Carely — User Manual

For every Carely subscriber. Keep this where your office can reach it.

This manual describes how Carely actually behaves, not how it is meant to. Where Carely does something surprising, it says so. Those are the parts worth reading before they cost you a morning.

If something is wrong right now, go to Troubleshooting — it is organized by symptom, the way you would actually search. If you are new, start with Understanding Carely and then your role's Quick Start.

Quick answers

TIPEvery screen has Ask Carely and Report a problem in the bottom corner. They sit as two small round buttons so they stay out of the way of your work — the speech bubble is Ask Carely, the life-ring is Report a problem. On a computer, point at either one and it widens to show its name. On a phone or tablet, just tap it. There is also a full practice agency at carelyapp.care/demo with invented data — do anything you are unsure about there first. Nothing you do in the demo touches your account. The demo opens with a short guided tour across the top of the screen. To look around on your own instead, select Explore on my own on the tour card.

1. Understanding Carely

Carely is software for running a home-care agency. It holds your clients and caregivers, builds the schedule, verifies that visits actually happened, turns those visits into timesheets, and turns approved timesheets into invoices and payroll figures.

You do not need to understand all of it. You need to understand the one rule below, and then find your own role.

The one rule that explains most of Carely

Work flows in one direction, and each step unlocks the next:

visit scheduled → caregiver clocks in and out → timesheet created
      → timesheet APPROVED → billable and payable
            → invoice raised → payment collected

If something downstream is missing, the cause is nearly always upstream. A missing invoice is usually an unapproved timesheet. A missing timesheet is usually a visit nobody clocked out of. A caregiver who cannot see their visit usually has no login yet.

Keep this chain in mind and most problems diagnose themselves.

Screens keep themselves current

You do not reload Carely to see what has changed. Every signed-in tab keeps a quiet connection open to the office, and when something in your agency changes — a caregiver clocks out, a family sends a message, a leave request comes in — the screen that shows it refreshes on its own within a few seconds. Two people looking at the same agency see the same thing. If that connection cannot be made (a network that drops idle connections, an older browser), Carely falls back to refreshing whenever you look at the tab again and every minute and a half while you do, so the worst case is a short wait, never a stale screen. Nothing sensitive travels over that connection: it carries only the name of what changed, and the screen fetches the data the same way it always did.

The five apps

Carely is five apps sharing one set of records. Which ones appear on your home screen depends entirely on your role.

There is also an Enterprise Portal for organizations that own several agencies. See Multi-location.

The app you are in is marked in the left-hand list: its row is filled, carries a bar down its left edge, and its name is in bold. The breadcrumb at the top of the page says the same thing. Inside an app, the tab you are on is marked the same way: filled in the app's color, in bold, with the bar down its left edge. On a phone the tab strip scrolls sideways, and the tab you are on is always scrolled into view.

Words Carely uses

Read these once. The rest of the manual assumes them.

There are five roles, and only five

IMPORTANTJob titles are not roles. Your scheduler, your biller and your compliance manager are all Administrator, and an administrator starts able to do everything. You can then take specific areas away from an individual — payroll, billing, clinical and the rest — so a scheduler need not be able to open payroll. → Limiting what an administrator can do It is a deny list, not a grant list: everyone keeps full access until you remove something. So restricting people is a decision you have to make deliberately, and it is still worth keeping a few admin accounts, removing departing staff the same day, and reading the activity log.

Two more things people assume and get wrong:

• A nurse is not a role. A nurse is a Caregiver record with a clinical profile and a license. The clinical module is switched on for the whole agency, not per person.
• Only Caregiver and Family Member can be chosen at sign-up. To create an administrator, the person signs up as one of those and you change their role afterward. See Adding a user.

Role-access matrix

What each role can reach. "—" means the screen does not exist for them at all.

What Carely deliberately does not do

Knowing these saves hours of searching.

NOTEThis list is honest rather than flattering on purpose. Every item on it is a real gap, and knowing about it now is cheaper than discovering it during a billing run.

2. Getting started

Where Carely runs

Carely is a website, and that is the whole product. Go to carelyapp.care in any modern browser — Chrome, Edge, Safari or Firefox — on a Windows PC, a Mac, a Chromebook, an iPhone, an iPad or an Android phone. There is nothing to install, nothing to update, and everyone sees the same information at the same time.

On Windows, the browser is the way to use Carely. There is no separate Windows program to download. Every feature described in this manual — scheduling, clocking in, billing, clinical records — works exactly the same in Edge or Chrome as it does anywhere else.

On a Mac, there is an optional desktop app at carelyapp.care/download if you prefer Carely in its own window rather than a browser tab. It is the same Carely, and it updates itself. Using it is a preference, never a requirement.

On phones and tablets, open carelyapp.care in the browser and add it to your home screen. Caregivers clocking in this way are prompted for location exactly as they would be in an installed app. Native App Store and Google Play versions are on the way.

Creating your agency

Purpose. Sets up your agency on Carely and gives you an organization code — the code every caregiver and family member uses to join you.

Who can do this. Whoever is starting the agency. You do not need an invitation.

Before you begin. Have ready: your agency's legal name, the state you operate in, an email address you control, and roughly how many clients you serve (it sets your price). Sign-up is two short steps and takes about two minutes.

Steps

1. Go to carelyapp.care, type your email address into the box beside Start free if you like, and select Start free — it opens the new-agency sign-up directly, with that address already filled in. (Create an account on the sign-in screen opens the form for joining an existing agency; choose Start your free trial at the bottom of it to start a new one.)

Step 1 of 2 — email and password

1. Enter an Email address and a Password. Use an email you control: your username and any password reset are sent there. Select the eye to check what you typed.
2. Select Continue.

Step 2 of 2 — your agency, care type and the agreement

1. Enter your Agency name — this appears on invoices and on documents families see — and Your full name.
2. Answer What kind of care do you provide? Nothing is chosen for you, and the form will not go on without an answer.
3. Companion care only for companion care, homemaker and personal care. Most agencies choose this.
4. Both companion care and licensed nursing or medical care then asks which license you hold — Nurse registry, Home health agency, or Nurse registry and home health. This is what offers you the clinical module; the tools themselves turn on after you verify the license inside Carely. See Clinical workflows.
5. Read the Business Associate Agreement (HIPAA) — select Read it to open the document. This is a real contract about protected health information. Your name is filled in under Signing as; change it if somebody else is the authorized signer. Enter Your title, then check the confirmation.
6. Check the box agreeing to the Terms of Service and Privacy Policy.
7. Note your username. Carely makes it from your email address and shows it above the button; select Change if you want a different one. It is also on your welcome card and in your welcome email.
8. If any of these apply, open Optional: client count and organization:
9. How many clients does your agency serve? This chooses your starting subscription tier and shows its price. Left blank, you start on the smallest tier. Give your real number — Carely re-counts from live data and adjusts later.
10. Part of a larger organization? — Sub-unit of another agency? Enter the Parent agency's organization code. This asks the parent to take you on as a branch: their administrator approves it before you can start, and then the parent pays for you and no payment step appears. Multi-location enterprise — enter the enterprise code (ENT-…) to join it.
11. Select Sign and start my trial. Your 30-day free trial starts immediately — no card, nothing to pay. See Your free trial.

What happens next. Your agency is created and you are signed in as its Administrator. Your welcome card shows your organization code and the username you sign in with.

How to verify it worked. Open the avatar menu (top right) → Account settings. You should see your agency name and an Organization code with a Copy button.

IMPORTANTThe BAA is signed once and stored permanently. It cannot be edited or deleted afterward — only viewed, exported and printed, from Account settings. That permanence is the point: it is your evidence that the agreement exists.

Common mistakes

• Signing up with a personal email nobody else can reach. Use an address the agency controls, or you will lose the recovery route.
• Choosing a nursing option "just in case." It adds clinical screens throughout. If you do only companion care, choose No — it can be turned on later in Account settings.
• Skipping the client count. It sets your price and your capacity band.

Your free trial, and what you pay after it

The trial is 30 days, and no card is needed to start. You create your agency and you are in — the whole product, every screen, from the first minute. Nothing is held back and nothing is locked while the trial runs.

You are reminded before it ends. Carely emails you a "Your free trial ends…" notice seven days before, naming the plan, the amount and how to subscribe. Add a card any time during the trial and it simply continues into a paid subscription.

If the 30 days pass without a subscription, the agency's access closes until you subscribe. Your data is not deleted — it is waiting for you when you come back. The screen you see then says so plainly — "Your free trial ended on …" — with the price charged today and the renewal terms above the button; there is no second trial. You confirm the automatic renewal, pay on Stripe's secure page, and are back in your agency within a few seconds of the card clearing. Everything is exactly as you left it.

If a renewal payment fails, you keep access while Carely tells you to update the card; a successful retry puts things right. You can cancel online at any time from Account settings → Subscription and keep access until the end of the period you have paid for.

If you sign up and get distracted, Carely sends you a short reminder after a couple of days, and once more about a week later — then stops. It only goes to agencies that have not added anyone yet; once you have a single client or caregiver on file, Carely takes that as you being under way and leaves you alone.

One trial per agency. Canceling and signing up again does not start a second one.

What it costs afterwards depends only on how many clients your agency actively serves:

Annual is ten months for twelve — paying yearly is two months free.

The tier follows your real client count. You give a number at signup, but Carely counts your active clients from live data and moves you to the correct tier, so a growing agency is never stuck on a plan that no longer fits and an under-declared number only delays the correction.

Sub-units of a parent agency pay nothing separately — the parent's subscription covers them, and no payment step appears at signup.

If your agency is a sub-unit

Sub-units are additional locations under one owner. You sign up with the parent's organization code, and:

• The parent approves you first. Until their administrator approves the request, signing in shows a screen naming the parent agency and nothing else opens. Choose Check again once they have approved. If they decline, the screen says so and nothing is linked or billed.
• The parent pays. No payment step, no separate subscription.
• The parent's administrators can switch into your location and operate it with full administrator access. This is by design; make sure your staff know it.
• Your client count is added to the parent's for tier purposes.

The Carely First-Day Setup Checklist

Do these in order — later steps depend on earlier ones.

1. Confirm your operating state. Avatar menu → Account settings → Operating state. It drives tax estimates in Revenue Center.
2. Set your agency name, reply-to address and brand color. Same screen. These appear on every email and invoice you send.
3. Copy your organization code and put it somewhere you can paste it into messages — you will send it to every caregiver and family.
4. Bring your data across if you have an export from another system: Operations Hub → Data Migration. Do not type clients in by hand if you have a file. See Migrating your data.
5. Add your office staff — Adding a user.
6. Add caregivers — Adding a caregiver.
7. Add clients, with rates — Creating a client.

The Carely First-Week Checklist

1. Upload every caregiver's credentials with expiry dates. An expired required credential stops that person from being scheduled — including stopping them from claiming open shifts — unless you have turned that blocking off in Certifications. Adding a credential
2. Confirm each client's EVV location. The anchor is set automatically from the first clock-in but stays unconfirmed, so distance is recorded and the geofence is not enforced until you confirm it. Operations Hub → Clients & Matching → the client's EVV location row. Confirming a client's location
3. Check every client has a charge rate. A client with no charge rate invoices zero and you will not notice for a month. Rates
4. Have caregivers fill in their availability — both weeks, not one day. The availability trap
5. Build your first week's schedule and confirm one visit reaches the caregiver's phone. Creating a visit
6. Do one full dry run: schedule a visit, have someone clock in and out, approve the timesheet, generate the invoice. Do it in the demo at carelyapp.care/demo if you would rather not touch live data. That whole dry run works in the demo. Three things are deliberately read-only there, because nothing the demo records is kept: recording a payer authorization, correcting an EVV exception, and resolving a leave conflict or setting the dates on an old leave request. Each of those screens says so rather than offering a form that cannot save.
7. Link families to their clients — approval alone shows them an empty portal. Family portal access
8. Take a backup. Avatar menu → Account settings → Backups.

3. Accounts, users, roles and access

Limiting what an administrator can do

By default every administrator can do everything — which is right for a small agency where one person does the lot, and wrong once you have a scheduler, a bookkeeper and someone handling compliance.

You can now take specific areas away from an individual administrator. The areas are:

Each can be set to view-only or removed entirely.

Where you do it. Operations Hub → Users, then Permissions beside the administrator you want to restrict. Everything is checked to begin with, because everything is allowed to begin with. Uncheck View to remove an area entirely, or uncheck only Change to leave someone able to look at it without altering it — a bookkeeper who should see the schedule but never move a visit. Save, and it applies immediately; the person does not need to sign out and back in.

The button only appears for administrators. Other roles are already confined to their own work, and protected accounts (super-admin, developer) cannot be restricted.

EVV is deliberately separate from Schedules. Building a schedule and swearing to hours nobody witnessed are different acts. An owner can leave a scheduler running the calendar while only a supervisor can close a visit whose evidence is missing. → Closing a visit when the evidence is missing

Nothing changed for anyone when this arrived. Every existing administrator kept exactly the access they had; permissions are only ever taken away, deliberately, one person at a time. Clearing the list gives someone their full access back.

It is enforced properly. Removing an area does not simply hide a tab — the server refuses the request, and tells the person which permission they are missing rather than just failing. Someone who knows the address of a page they no longer have access to still cannot reach it, and neither can anything acting on their behalf.

Every administrator keeps a few things whatever is taken away: signing in, seeing what they are allowed to do, the plan banner, and the signed agreements (the BAA and terms) on record.

You cannot lock yourself out. Carely will not let you remove your own ability to manage staff, and it will not let you remove it from the last administrator who has it — otherwise nobody could ever give it back.

How people get a Carely login

You do not create accounts for people. They create their own, using your agency's organization code, and then wait for you to approve them.

you send them the org code → they sign up → they appear as Pending
      → you approve them and choose their role → they can sign in

There is no email invitation to send or expire. If someone is not showing up in your pending list, they have not signed up yet or they used the wrong code.

The code is what ties someone to your agency, so it is required. Signing up without one, or with a code that does not match, is refused — nobody lands in your agency by accident or by guessing. A correct code still only creates a pending account: they cannot see anything until you approve them and choose their role.

Finding your organization code

1. Select your avatar (top right) → Account settings.
2. Find Organization code (share with your team).
3. Select Copy, and paste it into your message to them.

Adding a user (they sign up, you approve)

Purpose. Gives a person access to your agency — office staff, caregiver or relative.

Who can do this. Administrator.

Before you begin. Have your organization code, and know which role they need.

Steps

1. Send the person your organization code and the address carelyapp.care.
2. They select Create an account, choose I am a…, and enter the code along with their name, username, email and password.
3. In Operations Hub, open Users (in the Settings group of the left navigation).
4. Their request appears under Pending approvals with an Awaiting approval badge and the role they asked for.
5. Confirm who they are. Set Approve as to the correct role.
6. Select Approve.

What happens next. A confirmation appears: "[name] approved as [role] — they can now sign in." They move into the main user list and can sign in immediately. The notification about them disappears from the bell, because the decision it was asking for has been made.

NOTEYou do not have to go looking. When somebody signs up, a notification appears in your bell: "[name] is waiting for your approval", with the role they asked for and where to go. Only administrators see it. Previously a pending account was visible on the Users screen and nowhere else, so a relative who had signed up correctly could sit there unnoticed while every other screen behaved as though they did not exist.

How to verify it worked. They appear in the user list with the right role, and the Pending approvals section no longer lists them.

IMPORTANTAt sign-up a person can only choose Caregiver or Family Member. To create an office administrator, let them sign up as either one, then change their role to Administrator in the user list. This is the answer to "how do I add another staff member?"

Common mistakes

• Approving without checking who it is. Approving a Family Member is the moment you decide someone may see health information about a client. If you do not recognize the request, decline it. Leaving it pending is how the wrong person eventually gets approved by someone in a hurry.
• Approving a relative and stopping there. Approval grants a login; it does not connect them to a client. See Family portal access.
• Approving a caregiver and expecting them on the schedule. A login and a caregiver record are different things — see Adding a caregiver.

Agreeing to the terms

Everyone who uses Carely agrees to the Terms of Service and confirms they have read the Privacy Policy — admins, caregivers and family members alike. It is one checkbox on the sign-up form, with both documents one click away, plus a plain-English summary of what Carely is and is not responsible for.

Two things people ask about:

"I was asked again when I signed in." That happens for two reasons. Either the account was created for you — brought across from another system during a migration, rather than signed up for — so you had never been asked; or the terms have genuinely changed since you last agreed. You are asked once, not every time.

"Why can't I skip it?" No session is issued until you agree, so there is nothing to skip into. If you would rather not agree, simply do not — you are returned to the sign-in screen and nothing is created.

Each person can see exactly what they agreed to, and when. Admins can see it for their whole agency — see Compliance → Signed agreements.

Changing someone's role

1. Operations Hub → Users.
2. Find the person and change the Role dropdown.
3. The change applies immediately and a confirmation appears.

You can assign Administrator, Caregiver and Family Member. Super Admin and Developer are Carely staff roles, shown as protected, and cannot be assigned or altered by an agency.

WARNINGMaking someone an Administrator gives them everything to begin with: payroll, billing, every client record, and the ability to change other people's roles. You can then take specific areas away from them, but nothing is removed until you do it. Be deliberate about who gets the role, and narrow it afterwards. → Limiting what an administrator can do

Removing access

Do it the same day. An administrator holds every permission you have not taken away, so a forgotten account is live access sitting unattended — and narrowing someone's permissions is not a substitute for removing an account that should not exist.

Deleting your own account

Anyone can delete their own Carely account: avatar menu → Account settings → Delete my account. Enter your password, then confirm.

• Your sign-in ends on every device right away, and it cannot be used again.
• Your email address is removed from the account, and your phone stops getting Carely notifications.
• The records of care already given or received stay with the agency, which must keep them by law: visits, timesheets, pay records, invoices and EVV records keep the person's name.
• The agency's administrators get a notice in the bell. A caregiver who still works with the agency needs a new invitation to sign in again.

An agency's only administrator cannot delete their own account this way, because no one would be left to run the agency. Make someone else an administrator first, or delete the whole agency under Danger zone, which has a 30-day grace period and a full export.

Password and username recovery

Forgot password

1. On the sign-in screen select Forgot password?.
2. Enter the username or email on the account.
3. Carely emails a reset link. The link expires in 30 minutes. If the same email is on more than one Carely account (an admin login and a caregiver login, say), the email lists a link for each account, labeled with its username and agency — use the one for the username you sign in with. The confirmation names the username the new password applies to, and the sign-in screen fills it in. Forgot username works the same way: if the email is on several accounts, the email lists every username with its agency.

Which address the link goes to. The email on the account itself. A login that has none — a caregiver account created from the roster, a family login invited by name — is reached through the address on file for that person: the caregiver's profile, or the client's contacts. "Forgot password" finds the account by that address too.

If someone is stuck: the administrator's failsafe. Operations → Users → Send reset link on any account in your agency emails a fresh reset link to the address on file for that account (the Email column shows where, masked, and where it came from). The link goes to the person, never to you, so it can only ever let them set their own password; each send is in the audit log. An account with no address anywhere is refused with what to add.

Caregivers: when a username has been forgotten too. Enter the email on the caregiver profile under Forgot username.

1. Open the link (or paste the code into the app), then set a new password twice.

Forgot username

1. Select Forgot your username? Recover it.
2. Enter the email address on the account. Carely emails the username.

NOTEBoth screens show the same confirmation whether or not an account exists. That is deliberate: it stops someone using the form to discover who has an account. It is not a sign the request failed.

If no email arrives

• Check the address on the account is right and reachable. An account with no email address cannot be recovered by the user.
• Check spam.
• An administrator can trigger a reset for a locked-out user. If the only administrator is locked out, contact Carely support — see Getting help.
• Told the account is paused or locked after wrong passwords? See Locked accounts — a reset unlocks it.

Locked accounts (too many wrong passwords)

Carely counts wrong sign-ins per account, from any device or network:

• After 5 wrong passwords in a row, sign-in for that account pauses for 15 minutes.
• After 10 in a row, the account is locked until the password is reset. Carely emails the account's address to say so.
• Wrong two-factor codes count the same way.
• A correct sign-in starts the count over; so does a day with no wrong attempts.

While an account is paused or locked, even the right password is refused, so a lock never tells anyone whether a guess was correct. The sign-in screen says which it is and offers Reset your password to unlock. Resetting the password unlocks the account at once.

Administrators: Operations → Users shows a locked account in red with how many failed attempts locked it. Unlock lifts the lock without changing the password — the way back in for someone who has no email address to reset from. Every lock and unlock is in the audit log.

Confirming your email address

When a new agency is created, Carely emails the owner a link to confirm their address — the address password resets and security notices go to. Open the link once (it expires after 24 hours); until then a reminder bar at the top of Carely says by when, with Resend link.

• The owner has 7 days to confirm. After that, Carely shows only the confirm screen until the address is confirmed — the agency's data is untouched. From that screen you can send a new link or, if the address was mistyped, change it (your password is required).
• Caregivers and family members are never asked to confirm, and nothing changes for them.
• Accounts created before 17 October 2026 are not asked either.

Passwords

Passwords must be 8 to 72 characters with at least one letter and one number. Carely stores only a one-way bcrypt hash of your password — nobody at your agency or at Carely can read it.

Carely refuses a password that appears in a known breach list. It is not a judgment about how clever the password is — that exact string is on a list criminals use. Choose another, and never reuse a password from another site.

On first sign-in with a password an administrator set, Carely requires you to change it before you can do anything else.

Two-factor authentication (2FA)

Purpose. Adds a six-digit code from an authenticator app to your sign-in, so a stolen password alone is not enough.

Who can use it. Every role, on their own account.

Steps

1. From your Carely home screen, find Two-factor authentication.
2. Add the displayed key to your authenticator app.
3. Enter the current six-digit code to confirm, and 2FA is enabled.

From then on, signing in asks for the code after your password.

WARNINGThere are no backup codes, and your own agency administrator cannot reset 2FA. Only Carely support can clear a lost authenticator. Before you enable it: use an authenticator app that backs itself up to your account, and never enable it on your agency's only administrator login.

To turn it off, use Turn off 2FA in the same place.

Signing out other devices

If a phone or laptop is lost: avatar menu → Account settings → Session security → Sign out all other devices. Every other session for your account ends; you stay signed in where you are.

Carely also locks your session automatically after a period of inactivity — the exact number of minutes is shown in your avatar menu.

Reviewing who changed what

Avatar menu → Security activity. It shows recent actions with who performed them and when — records created, changed and deleted, approvals, exports and overrides.

Permissions decide what somebody can reach; this is how you see what they actually did. Review it periodically, and use it first if you suspect misuse of access.

Why a menu item is missing

Almost always your role, not a fault. Check the role-access matrix. The three that surprise people most:

• Administrators have no Family Connect. You are not attached to one family. Read and post to the same conversations from Chat.
• Caregivers have no Revenue Center. They have My Pay instead.
• Clinical screens are invisible unless the clinical module is switched on for the whole agency.

4. Client management

A client is the person receiving care. Clients never sign in — the person who signs in is their relative, on the Family Member role.

Everything lives in Operations Hub → Clients & Matching (People group).

Creating a client

Purpose. Creates the record everything else hangs off: scheduling, invoicing, the care plan, the family portal and your subscription count.

Who can do this. Administrator.

Before you begin. Know the client's care level and what you will charge. If a relative should have portal access, approve their account first (Adding a user) so they appear in the picker.

Steps

1. Operations Hub → Clients & Matching.
2. Select ＋ Add client.
3. Complete the top section:
4. Full name — required.
5. Age — required, numeric.
6. Care need — required. A short description, e.g. Diabetes management.
7. Referral source — required. Where this client came from: a hospital, a doctor, your website. Suggestions appear as you type. This feeds your referral reporting, and it is the one field agencies skip and later regret.
8. Care level — Low, Medium or High. This prefills the two rates.
9. Assign caregiver — the client's primary caregiver. Can be changed later, and starts on — Unassigned —. A new client is deliberately not assigned to anyone: matching is a decision you make, not one Carely makes for you.
10. Client charge rate ($/hr) — what you bill.
11. Caregiver pay rate ($/hr) — what you pay for this client.
12. Complete Billing:
13. Billing contact name and Billing contact email — who receives the invoice.
14. Invoice billing frequency — Weekly, Biweekly or Monthly.
15. Payment method — ACH, Credit/Debit Card or Check.
16. Only if you bill Medicaid or an insurer for this client, open Insurance & Medicaid identity and complete it. See Insurance and Medicaid identity below. Private-pay clients skip this entirely.
17. Under Family portal access, choose the family member who may see this client, or leave it on — No portal access —.
18. Answer any Client intake questions your agency has configured. (Responsible party and Power of Attorney are not asked here — they belong in Client Contacts, below.)
19. Select Save client.

IMPORTANT— Unassigned — means exactly that. If the box shows a caregiver's name, that caregiver is assigned; if it shows — Unassigned —, nobody is. This matters because three things follow from it: the caregiver sees this client in their own app, the family sees the caregiver's name in Family Connect, and visits can be scheduled against them. If a family tells you they cannot see who is caring for their relative, check this box first.

What happens next. The client appears in the list and becomes selectable when you create a visit. They count toward your active-client total from that day.

How to verify it worked. The client appears in Clients & Matching with status Active, and shows a charge rate and pay rate on their card.

WARNINGA client with no charge rate invoices zero. Their hours are real, the money is not, and nobody notices until the billing run. Check the rate on every client, especially any brought in by migration — rates are the field most often missing from another system's export.

Rates prefill from the care level. Low is $28 charge / $18 pay, Medium $30 / $20, High $35 / $22. Adjust for the individual client — the defaults are a starting point, not a policy.

Common mistakes

• Leaving the referral source blank, then finding your Growth Center cannot tell you where clients come from.
• Setting the caregiver's usual rate rather than the rate for this client. Pay follows the client's care needs, so a caregiver's effective hourly rate varies by who they saw. That is normal, not an error.
• Expecting the family to see anything after approval alone. See Family portal access.

The care team: more than one caregiver on a client

Caregivers on a client change constantly — cover, holidays, somebody leaving. Assign caregiver names the primary. Underneath it, Care team is everyone else who covers that client: check a name to add them, check it again to remove.

Being on the care team is not a label. Anyone on it:

• can open that client's record
• receives that client's care alerts
• is preferred by the scheduler when one of their visits needs filling

That first point is the one to weigh before you add a name. The care team is an access decision, so treat it the way you would treat handing somebody a key.

The primary caregiver is already covered and is deliberately left out of the list, so removing somebody from the team can never quietly unassign the client.

When somebody leaves — deactivating a caregiver takes them off every client's care team automatically, and hands the primary slot to a remaining team member where there is one. You do not have to walk the client list yourself.

Letting Carely choose the caregiver

ALIASES: auto match, auto-match, ai matching, match caregiver automatically, who should care for this client, change the match, override the match

On the Clients & caregiver matching tab, every card has a Match dropdown and an ✨ Auto-match button (Re-match once somebody is assigned). Select it and Carely chooses from your whole roster and says why on the card. Auto-match unassigned, at the top of the tab, does it for every active client who has nobody yet. When Carely's AI made the choice, the card shows it as a suggestion and nothing is saved until you select Use (see step 3).

How it chooses, in order:

1. Who may take this client's visits at all — the same rules that decide whether a shift can be published: credentials, licensure, approved leave, double-booking and weekly hour caps. Nobody the rules refuse is ever suggested.
2. Who can do the work. Carely reads what the client needs — the stated care need, the intake answers (personal care, memory, behaviors, mobility, language, pets, diet), the plan of care — against what each caregiver has on file: skills, experience, certifications, the answers to your own application questions, the clinical profile. The fit is a percentage with each need named as covered or missing. A client who speaks Spanish is matched to someone who does; a gap on a need that matters counts against a caregiver however available they are.
3. Carely's AI, when your agency has a model configured, ranks the top candidates the rules allowed — weighing a dementia specialty against months of continuity, a shared language against a lighter week — and writes the one-sentence reason you see on the card. It sees needs and facts only, never a name. If it is not configured or does not answer, the rules' order stands and the card says so.

When the AI chose, you decide. The card reads "Carely AI suggests …" with its reason, and the client keeps whoever they had until you select Use (name). Carely checks the rules again at that moment, so someone whose credential lapsed in the meantime is refused rather than assigned. Dismiss leaves things as they were. After you use a suggestion, the caregiver sees on that client's care plan that Carely AI helped suggest them and that your office approved it. Without the AI, Auto-match assigns the rules' choice in one step, as before.

You can always change it. The dropdown on the card is the override; pick anyone and that is who the client has (and the AI note goes away). Auto-match, each AI suggestion and each one you used are recorded in the audit log with what they were based on.

Asking Carely to suggest a team

Select ✨ Suggest a care team and Carely ranks your roster for that client using the same rules that decide whether a shift can actually be published — credentials, licensure, approved leave, double-booking and weekly hour caps. A name it offers is a name that would survive being scheduled, not one that merely looks plausible.

Each suggestion says why — including how well the person's skills fit the client's needs, with the needs named — and how many of that client's visits the person could take. The screen also says what the ranking is based on: the client's booked visits, the visits they usually have, or — for a client with no history yet — a typical weekday.

It suggests only. Nothing is added until you select it and save. That is deliberate: being on a care team grants access to somebody's health record, and that is not a decision software should take on its own.

Recording who pays

Where. Operations Hub → Clients & Matching → open a client → Insurance & Medicaid identity → Insurance company.

Choose from the list rather than typing. The list covers Medicare, Medicaid and the major national insurers, Medicare Advantage and managed-Medicaid plans, and the long-term-care insurers. Picking from it means every client on one payer reports as one payer — typing it freehand each time turns "Aetna", "AETNA" and "Aetna Better Health" into three payers in your reports and one in real life.

If your payer is not listed, choose Other — not listed and type it once. Carely stores your spelling and reuses it on every invoice for that client, so it stays consistent even though it is not on the list. Tell us and we will add it.

Private pay. Choose Private pay (no insurer) so it is recorded deliberately, rather than left blank and indistinguishable from "we haven't asked yet".

It flows through to billing. When you record insurance on an invoice, the payer is already filled in from the client. You can change it there for a one-off, but you will not be retyping it every month.

Insurance and Medicaid identity

Purpose. Records who this client is to a payer, which is rarely the same as who they are to you. You call her Peggy; her Medicaid record says Margaret. A claim has to use the payer's version.

Who can do this. Administrator only. Caregivers and family members never see any of it, even for a client they otherwise have full access to.

When you need it. Only when you bill Medicaid or an insurer for that client. If every client pays privately, leave the whole section closed — nothing elsewhere depends on it.

What to enter

IMPORTANTThe member ID is encrypted and only administrators can read it. It is never included in what a caregiver or a family member sees.

Common mistakes

• Entering the name the family uses instead of the name on the card. This is the single most common reason a clean-looking claim comes back rejected.
• Assuming Age is enough. It is a whole number that changes every year; a payer matches on the date.
• Guessing at a member ID. A wrong number is worse than a blank one — ask for the card.

Getting it at intake is worth a minute. Asking a family for a birth date and a member number during the first conversation is normal. Telephoning them for it eight months later, because a claim was rejected, is not.

Giving a family member access

Purpose. Connects an approved Family Member account to one specific client. Without this, they sign in to an empty portal.

Who can do this. Administrator.

Before you begin. The relative must already have signed up and been approved with the Family Member role.

Steps

1. Operations Hub → Clients & Matching, open the client.
2. Find Family portal access → Family member who can see this client.
3. Choose their account from the list. It shows their name and email.
4. Save.

What happens next. When they sign in to Family Connect, they see this client: visit reports, the care plan, the care-team conversation, incidents and their invoices.

How to verify it worked. The client's record shows the linked family member. Ask the relative to sign in and confirm they can see their loved one.

Troubleshooting. If nobody appears in the list, the message under it tells you which of two situations you are in, and they need different things:

The picker lists approved family accounts only, so somebody who has signed up but not yet been approved will not appear here — which is why the message distinguishes the two. A pending account also raises a notification in your bell.

NOTEThis is the most common family support call by a distance. Approval and linkage are two separate steps, and only the second one shows them anything.

The reference number on the client card

ALIASES: client number, patient number, client id, patient id, reference number, number on the card, what is client hash

Every client card shows a short reference number beside the care need — Client #538188, for example.

What it is for. It stands in for the client's name on documents that do not need to carry it. A caregiver's paystub lists the visits they are being paid for by this number rather than by the people they cared for, so a pay document can be handled, filed or queried without putting a client's name in front of anyone who does not need it.

Where the wording changes. Agencies with the Clinical module switched on see Patient #; companion-care and other non-medical agencies see Client #. It is the same number either way — only the word follows the kind of care you provide.

NOTEThe number is derived from the client record and stays with it. You do not set it, and you cannot edit it.

Client status and the lifecycle

Each client carries a status you set from their card:

Discharged, Deceased and Archived clients do not count toward your subscription tier. Everything else does — including On Hold and Hospitalized. If a client will be away for a long time, that is worth knowing.

Discharging a client

Setting a client's status to Discharged opens a confirmation that tells you exactly what will happen: they are removed from active care, their upcoming visits are canceled, and they drop off schedules and family access. Their history — invoices, past visits — is kept, and you can readmit them at any time.

How to verify it worked. Their card shows Discharged, and their future visits no longer appear on the schedule.

Deleting a client permanently

A permanent delete asks for a reason, archives the record to Operations Hub → Archived Records for HIPAA retention, and then removes them from the working app.

Removed from the live database

The client record and everything that only describes them going forward: their contacts, their open and upcoming visits, their payer authorizations and the units counted against them, and their rate history. All of it is in the archive first.

Deliberately kept

Completed visits, timesheets, invoices, EVV records, incidents and visit reports are not deleted, and they carry the client's name. That is not an oversight — they are the evidence that care was delivered and paid for, and destroying them would break your payroll and billing records and your retention obligations.

So a permanent delete is not an erasure of every mention of the person. If you have had a request to erase someone's data, read this section before answering it, and take the question to your compliance officer or attorney — the right answer depends on your state and your payer contracts, not on what this software can technically remove.

IMPORTANT"Delete" in Carely means "archive out of sight," not "destroy." This is deliberate: health records have retention requirements, and a truly destructive delete would put you in breach of them. Use Discharge for care that has ended and delete only for genuine mistakes such as duplicates.

The care plan

The plan of care records what this person receives: the services (which become the caregiver's visit checklist), goals, equipment and instructions. Families read it in Family Connect, and caregivers read it against their visits.

It is written and signed when you add the client. The Add client form ends with the plan of care and a Signatures block for the two people who sign it:

Under the signatures, before you press Save client, Carely tells you how the plan will be saved: Signed and approved, waiting for the responsible party, or not signed yet. Saving without signatures is allowed — a first phone call is rarely the moment — but the plan is then shown to everyone as not signed until it is.

The signatures are shown on the plan. At the top of the plan — for you, for the caregiver and for the family — each signer's signature appears with their name, how they signed (in person or online) and the date. Both are on the printed copy as well.

One status, shown to everyone. The plan is signed and approved only when both have signed it. Until then it reads Waiting for the responsible party to sign, Waiting for the agency to sign, or Not signed yet. The same line appears:

• on the client's card, beside Plan of care, so an unsigned plan is visible without opening anything;
• at the top of the plan for the caregiver, who can see whether the plan they are working from has been approved;
• at the top of the plan for the family, who see Signed and approved with who signed and when — or Waiting for your signature, with a place to sign.

Changing a plan takes the signatures off it. A signature belongs to the plan as it was worded when it was signed. Edit the plan (Edit plan on the client's card) and Carely warns you, before you save, that the signatures on the earlier version will be removed; after saving, the plan is unsigned until both sign the new wording, and the family is told it was updated and asked to sign again. The earlier signatures are kept in the client's history, not thrown away. Editing anything else about the client — an address, a rate, the care team — leaves the signatures alone.

An empty plan cannot be signed. A signature has to be on something; fill in the plan first.

TIPIf the responsible party cannot use the portal, sign with them at the next home visit: open Edit plan, hand them the screen, and have them type their name.

Emergency contact

The emergency contact appears on the client card and on caregivers' visit screens, in red, so the person in the house can find it fast. Families maintain it themselves in Family Connect; administrators see it but do not type it in. If it is wrong or missing, ask the family to update it — or update it with them on the phone.

The contact directory

Operations Hub → Client Contacts.

Everyone connected to a client's care beyond the emergency contact above — family, power of attorney, physician, case manager — with their role and how to reach them.

The useful part is what it shows you by absence: the screen flags clients with nobody on file. A client with no reachable contact is a problem you want to find on a quiet Tuesday, not at 2am when something has gone wrong.

Responsible party lives here, not on the client form. The intake form used to ask for the responsible party's relationship, phone and whether they hold Power of Attorney. It no longer does — this screen already records all three, for as many people as a client actually has, and keeping a second thinner copy on the client form only meant one of them went stale. Add the responsible party here and mark their type as Power of Attorney.

The screen follows the roster. Discharge a client, readmit one, rename one or delete one and this list keeps up on its own:

• A discharged or deceased client's contacts are kept — you may still need to reach the family, and the record is retained deliberately — but they are labeled with the status and Carely stops asking you to add an emergency contact for them. Only active clients are counted in the gaps above, and only active clients are offered when adding a contact.
• Renaming a client brings everything that is looked up by their name with them: contacts, shifts, timesheets, visit reports, recurring schedules, payer authorizations and their recorded usage, and open alerts. Their EVV history keeps the name each visit was verified under — that is the record of what happened, and of what was sent to the state, so it is deliberately not rewritten. The same holds for a caregiver rename, and for a locked pay run: it says who was paid at the time. For a patient in the clinical module, their whole chart moves with them — medications, the eMAR history, physician orders and clinical notes. What does not move is who gave a dose, wrote a note or signed an order: that names the clinician who performed the act at the time, and rewriting it later would falsify a clinical record. A caregiver's pay adjustments do follow, because those are applied to future runs. Renaming a caregiver also renames the account they sign in with, so their own schedule, clock-in, pay details and nursing license keep working — the change reaches them on their next action, without signing out and back in.
• Permanently deleting a client removes their contacts too, and stores them inside the archived record — so a deletion made in error can be undone, contacts included, from Operations Hub → Archived Records.

Client documents and intake questions

Your agency's intake questions are configurable: Operations Hub → Forms → Client intake questions. They belong to your agency alone — customizing them changes the form for your staff and nobody else's. Anything you add there appears on the client form for every new client, so you can capture what your state or your payers require without waiting for Carely to add a field.

Migrating from another system

Purpose. Moves your agency into Carely from your previous software without typing any of it in. What comes across:

Past EVV verification arrives as a record, never as something to re-send. Imported visits land in a separate history table, deliberately outside the column every transmission path keys off. Those visits were already verified and billed by your old system; re-sending them would be duplicate submission to your state.

User logins are a special case. An imported account arrives with a password nobody knows — not even whoever ran the import — so the person must set their own from the sign-in screen before they can get in. If the email on file is wrong, they never can; Carely lists anyone in that state under User management rather than leaving you to discover it when they call.

Who can do this. Administrator.

Before you begin. Export your data from your current system. Whatever it gives you is probably fine: a .zip export package, one or more .csv files, an Excel .xlsx, .json, or a .pdf. You do not need to rename columns, reorder anything, or know which file is which.

If all you have is a PDF

Some systems will only hand you a printed report. Upload it anyway — Carely reads the table out of the page and treats it like any other file, including the dry run and the ability to undo.

The row tells you what Carely did with that particular file. A plain printed table says “Read from a PDF — check the columns before importing”, and it is worth doing: a CSV states what its columns are, a PDF does not, so Carely works them out from where the text sits on the page. Read the dry run before approving.

While a PDF is being read, the row shows a spinner and "Reading <file> — Carely is working through the pages." A long packet goes to Carely's AI and can take up to a minute; the page has not frozen, and the file picker returns when it is done.

A fillable PDF — one whose boxes you can type into — is read from its fields directly, by the names the template gave them. The client is whatever the form's own client-name box says; a copy of the name elsewhere on the form (a Plan of Care header, say) is never used, and if it differs the row says so.

Where each answer goes is decided against your agency's own forms — the questions on your client intake form and plan of care (Settings → Forms), including ones you added or reworded — not a fixed list. A box the template calls txtPatientDOB or Member Date of Birth is still the date of birth; a checked Diet_Low_Sodium is the dietary answer; Preferred_Hospital lands on your "Preferred hospital" question if you have one. An answer to a question you have taken off your form is kept as a plan-of-care line rather than filed where nobody would see it. Whatever the names cannot place is offered to Carely's AI against that same list of places — it can only choose a place that exists — and the row says how many it placed. Anything still unplaced arrives as a care-plan line, never dropped.

Dates are read the way people write them: 4/30/1942, 04-30-42, April 30, 1942, and a slip like 04/301942 (a slash short) all become the same date, and a two-digit birth year is always in the past. A date that cannot be read is not written to a date field; the dry run says so. The packet's physician becomes a Contact (type Physician, with phone and fax) as well as the four physician lines on the plan of care. A case manager, guardian or power of attorney named on the form becomes a contact too; the emergency contact and responsible party, when they are the same person, are one contact.

A filled-in caregiver application is recognized by its fields (applicant, not client) and becomes a caregiver record: contact details, license, emergency contact, days and shifts available, skills, and the answers to your agency's own application questions, which are kept on the caregiver's profile.

A client created from a PDF carries a line on its card saying how its name was decided — from a Client Name field, from First Name and Last Name fields, from the page's own labels, or the reader's own choice when the page has no name field Carely recognizes — and the words on the page it came from. If a migrated client has the wrong name, that line says why.

A filled-in form — an intake packet, say — is not a table at all. It is one client with the field names down the page and the answers beside them. Carely reads those as fields rather than rows, and the row then says how many it placed and which ones it could not.

If Carely could not read the page as a form, the row says why, rather than leaving you to guess: too few label-and-answer pairs to be a form, no AI model configured on the server, the model unreachable, or nothing that passed Carely's own checks. In each case the file is still there and you can choose its data type yourself.

Three things it cannot do, and it will say so rather than import something wrong:

• A scan or a photo of a page has no text in it to read. Export a CSV if your old system offers one — it is more accurate anyway.
• A password-protected PDF has to have the protection removed first.
• A PDF built from cut-down fonts that do not say what their characters mean would come out scrambled, so it is refused instead of imported as nonsense.

Move my whole agency (the usual way)

1. Operations Hub → Data Migration (Settings group).
2. Under Move my whole agency, choose your export files — all of them at once, or a single .zip containing them.
3. Carely reads each file and tells you what it thinks each one is and how sure it is. Anything it cannot identify confidently is listed for you rather than guessed at.
4. Select Dry run. Nothing is written yet. You will see, per data set, how many records are ready, how many need attention, and how many are already in Carely. For invoices you also see the total amount that would be imported — check it against your own books.
5. If it looks right, select Approve & import.
6. If anything is wrong afterwards, select Undo this migration.

What happens next. Carely imports in the right order on its own — people before the visits that reference them — so a visit finds its client and caregiver even though they were only created moments earlier. You do not have to import anything first.

How to verify it worked. Open Clients & Matching, Caregivers, and the Schedule. Your records should be there and connected: a migrated visit should name a real client and a real caregiver, and opening that client should show their care plan.

What Carely can bring across

What it does not bring across, at present: the scanned document files attached to credentials, Medicaid claims, and referral records. Those stay in your old system, and you should keep access to it for your record-retention period regardless.

IMPORTANTMigrated timesheets are recorded as Migrated, never as EVV-verified. Carely did not witness that work, and its records will not imply otherwise.

One data type at a time

If you only want to bring across one thing — say, a corrected list of clients — the per-data-type flow is still there below the whole-agency panel. It is also where you import a file Carely could not identify: you tell it what the file is, and the rest is the same.

TIPPractice in the demo at carelyapp.care/demo first. It runs the same validation and dry run against invented data, so you learn the shape of your own file's problems before touching your real agency.

Common mistakes

• Migrating clients without rates and then invoicing zero. Check rates after any import — a charge rate is the one thing an export from another system most often leaves out.
• Approving without reading the dry run. The counts are there so you can catch a file with the wrong month or the wrong branch in it before it lands, not after.
• Assuming a rejected row was lost. Records that need attention are held, not discarded — fix the file and import it again. Carely recognizes what it has already brought in and will not create it twice.

5. Caregiver management

Everything lives in Operations Hub → Caregivers (People group).

IMPORTANTA caregiver record and a caregiver login are two different things, and new agencies trip over this constantly. - The record is what you create here. It carries their details, skills, availability and credentials, and it is what you assign to visits. - The login is what they create themselves with your organization code, and you approve in Users. They need both. A caregiver with a record but no login cannot see their schedule. A caregiver with a login but no record cannot be scheduled. For the two to connect, the name on the record and the name on the account must match.

Employment type: W-2 or 1099

On each caregiver's card you choose whether they are a W-2 employee or a 1099 contractor. This is not a label — it changes what your agency owes.

A caregiver costs more than their hourly rate when they are a W-2 employee — usually around 10-12% more once employer taxes are counted. Carely shows you that real cost in Revenue Center → Expenses, so you can price work off what it actually costs you.

New caregivers default to W-2, which is the correct answer for most home-care work.

Be careful with 1099. If you set the schedule, provide the training and supervise the work, the person is usually an employee in the eyes of the Department of Labor and your state — whatever the paperwork says. Getting this wrong is expensive: the back taxes, interest and penalties fall on the agency, not the worker. Carely will tell you when a classification is removing workers from your tax base, but it cannot make this decision for you. If you are not certain, ask your accountant.

Adding a caregiver

Purpose. Creates the staff record used for scheduling, matching, compliance and pay.

Who can do this. Administrator.

Before you begin. Have their application in front of you. Have them sign up with your organization code as well, so you can approve their login.

Steps

1. Operations Hub → Caregivers.
2. Select ＋ Add caregiver. (If your agency has the clinical module, there is also ＋ Add nurse — see Clinical.)
3. Work through the grouped form:
4. Personal information — Full name, Address, City, State, ZIP code, Phone number, Email address. City, state and ZIP are three separate boxes (they used to be one), so the same home is always written the same way. State is a list to pick from, by name. The state and the ZIP must agree: a mismatch is refused with the state the ZIP actually belongs to, so you can see which box to correct. Name, phone and email are required.
5. Emergency contact — who to call about them.
6. State EVV worker details — date of birth, gender, hire date, the state license or registry number (the HHA or CNA registry number, not the driver's license) and the EVV worker ID if the state issued one. Your state's EVV aggregator keys a caregiver on these; HHAeXchange refuses a caregiver without the first four. In a state that enrolls caregivers with Medicaid themselves (Arkansas, for example), also enter the caregiver's own Medicaid provider ID; CareBridge requires it on every Arkansas visit.
7. Employment eligibility — right-to-work and license confirmations.
8. Availability — the days they can work.
9. Experience, Skills & certifications, Employment history, Background, References — used by matching and by your compliance file.
10. Direct deposit (for pay) — optional, a note of the last four digits for your own payroll. Paying a caregiver through Carely needs them to connect a payout account on Stripe's page instead (My Pay → Direct deposit → Connect payout account; you can start it for them from their profile) — see Taking payments in Carely.
11. Save.

What happens next. They appear in the caregiver list and become selectable when you create a visit.

How to verify it worked. They appear in Caregivers, and their name appears in the Caregiver (optional) dropdown on Schedule → New shift.

Common mistakes

• Spelling the name differently from their login. Carely connects the record and the account by name. Keep them identical.
• Creating the record and forgetting to approve their account, so they never see a shift.
• Leaving credentials for later — an incomplete credential blocks scheduling. Do it now: Adding a credential.

The availability trap

Caregivers set their own availability: Schedule → My availability. They mark each of the next 14 days as Day, Evening, Overnight, Any or Off.

WARNINGUntil a caregiver saves availability for the first time, Carely treats them as available. The moment they save it once, every day they did not mark counts as unavailable. So a caregiver who saved a single Monday has just told you they cannot work the other thirteen days — and they will quietly get offered less work. Have them fill in the whole two weeks, and keep it current.

This is worth saying out loud in training. It is the most common cause of "why am I not getting hours?"

Assigning a caregiver to a client

Each client has a primary caregiver, set on the client record (Assign caregiver). Clients and caregivers also appear together in Clients & Matching, where Carely suggests matches based on skills, availability and the client's needs.

Matching is a suggestion. You decide.

Two caregivers with the same name

Carely will not let two caregivers in your agency share a name. Adding or renaming someone to a name already on your roster is refused, because a name is how their shifts, timesheets, credentials and pay are matched to them. Two people called John Smith need to be distinguishable — "John Smith" and "John R. Smith", for example.

A caregiver in another agency can have the same name as one of yours. Agencies never see each other's records, and Carely keeps the two entirely separate.

Renaming is safe. If someone marries, or you fix a typo, Carely moves their schedule, timesheets, credentials, availability, training, pay adjustments and their login across to the new name in one step. Records that are a point-in-time statement of fact — a completed EVV visit, a paystub already issued — deliberately keep the name they were made under, because that is what they said at the time.

Deactivating a caregiver

Purpose. Takes someone out of scheduling when they leave or go on extended leave, without destroying their records.

Steps

1. Operations Hub → Caregivers.
2. On their card, choose the deactivate action.
3. Read the confirmation and select Deactivate.

What happens next. Carely tells you exactly what it will do: they are removed from active matching and scheduling, taken off their clients' care teams, and their upcoming shifts are freed up. Their timesheets and certifications are kept, and you can reactivate them at any time.

How to verify it worked. They no longer appear in the caregiver dropdown when creating a visit, and their former shifts now show as open.

TIPFreed-up shifts become open shifts immediately, so the gaps are visible the moment someone leaves. Check the Schedule right after deactivating and refill them. Your caregivers will see the gaps in the next daily open-shift digest, but that is a day away — for a visit in the next 48 hours, fill it yourself. See Open shifts and Open shifts: a daily digest.

Do this the same day someone leaves, and remove their login in Users too.

Permanently deleting a caregiver

Deactivating is almost always what you want. Permanent deletion is for a record created in error, or a request to erase someone entirely. It asks for a reason, and the reason is recorded.

Everything of theirs goes with them. Their bank/payout details, certifications, training records, availability and any pay adjustments are removed along with the caregiver — not left behind. That matters because Carely looks these up by name: anything left over would be inherited by a future hire with the same name, and a payout could reach the wrong person's bank account.

What is deliberately kept:

• Locked pay runs — the record of what was actually paid, and to whom, at the time.
• EVV visit history — the compliance record of visits that were verified, and where applicable transmitted to your state.

Neither is looked up by name to decide anything new, so nothing is inherited from them.

It is reversible. The whole record — bank details included — is stored in Operations Hub → Archived Records first, so a deletion made in error can be restored with everything intact.

What caregivers see

A caregiver's Carely is much smaller than yours:

Family Connect reads differently for a caregiver. It is the same screen the family opens, so it names their clients rather than "loved ones", and it leaves out the two things that belong to the family: Billing, which is what the family owes for their care, and Quality, whose survey is the family's own rating of the agency. A caregiver can read a client's emergency contact — it is who they call — but the family maintains it, and the plan of care is signed by the family, not by the caregiver. Their private line to the office is under Messages → Caregiver ↔ office.

They never see other caregivers' visits, other caregivers' pay, agency finances, the family's billing, another caregiver's conversation with the office, or clients who are not theirs.

Leave and time off

Caregivers request leave in Operations Hub → Request Leave, and can see their remaining balance per category first.

Administrators approve in Operations Hub → Leave, where each request shows the person's remaining balance so you can decide with the facts in front of you. Annual and Sick accrue monthly; Maternity, Paternity and Emergency are fixed total allowances, and you can set these per caregiver.

Paid leave — a second, separate answer

A caregiver asks for the time off and, if they want it, asks for it to be paid. Those are two questions and Carely keeps them apart, because the commonest real answer is "take the week, but it comes out unpaid" — and a single Approved/Denied cannot say that.

Requesting it. On the request form there is Ask for this to be paid leave. It is off unless they check it. The form tells them the office answers it separately, so an approval later is not read as a promise about money.

Answering it. In Operations Hub → Leave, a request that asked for pay carries its own row: "[Name] asked for this to be paid leave", with Approve pay and Deny pay beside it. Refusing asks you for a reason, which the caregiver sees.

The two decisions never move each other:

Refusing the pay never takes the leave away, and taking the leave away never changes what you said about the pay. You can answer them in either order.

What the caregiver sees. Operations Hub → Request Leave → My requests shows both answers on the same row: the time-off status, and beside it Paid leave approved, Unpaid — pay was not approved (with your reason), or Paid leave requested — awaiting a decision.

How long a day of leave is

Leave is asked for in days far more often than in odd hours, so the request form counts days and multiplies. A full day is 8 hours unless you change it.

Where. Operations Hub → Leave → Leave allowance & accrual → A full day of leave is __ hours. Set 12 if you run twelve-hour shifts, 7.5 if that is your day. Anything from 0.25 to 24 hours, in quarter-hour steps.

A caregiver requesting time off then gets 1 day / 2 days / 3 days / 5 days buttons, and the form tells them what it worked out: "A full day is 8 hours — that is 3 days." They can still type an exact figure, and part-day leave records the hours its own times describe — a morning off is not rounded up to a whole day.

NOTEChanging this does not alter leave already requested. It changes what the next request starts from and counts in.

Where paid leave comes out of — your rules, per category

One pot, or two? That is the whole question. Each category is a pot of hours, and time off comes out of it; all this setting decides is what happens when that time off is paid.

Agencies do not agree on the answer, and both are normal. You choose, for each of the five categories independently: Annual, Sick, Maternity, Paternity and Emergency.

Where. Operations Hub → Leave → Leave allowance & accrual. Under each category heading is Paid leave comes out of:

Two pots is how you cap what you pay for without capping how much time off somebody can take. For example: 80 paid annual hours. A caregiver takes 60 paid — 20 paid hours left. They take 40 more unpaid, which comes out of the general pot and costs you nothing. They still got the time off.

Choosing "a separate paid one" adds a second box per caregiver — the dashed one — where you set their paid allowance. It follows the same shape as the general one: hours per month for Annual and Sick, a fixed total for Maternity, Paternity and Emergency.

Their balance chip then shows both, for example "Annual 152 hrs · 76 paid" — 152 hours of time off left, of which 76 can be paid.

Which pot a request draws depends on the pay decision, and moves if you change your mind:

• Pay not yet answered, or refused → the general allowance.
• Pay approved → the paid allowance, and the hours move out of the general one.
• Change the pay answer later and the hours move back. They are never in both.

Changing the setting re-files leave you have already approved. Switch a category to a separate paid allowance and the paid absences already approved move into it; switch back and everything consolidates into the one pot. Carely tells you how many caregivers it rebalanced.

NOTELeaving every category on this allowance keeps Carely working exactly as it did. Nothing moves until you change something.

How approved paid leave gets paid

Once you approve the pay, the hours go onto the pay run for the period the leave falls in. You do not add anything by hand.

On the Pay run the caregiver's line reads, for example, "20.00h worked · 8.00h paid leave · $240.00", and the line says which rate was used. Both exports carry Paid leave hours and Paid leave as their own columns, so a bookkeeper can still reconcile the worked hours against the visits behind them.

Paid leave is paid at their average hourly rate — what they actually earned divided by what they actually worked, which is the same weighted rate overtime is paid on. Leave is not attached to a client, so there is no single visit rate to use. If they worked nothing in the period, Carely uses their most recent paid work.

Paid leave does not count toward overtime. Overtime is owed on hours worked. Someone who works 36 hours and takes 8 hours of approved paid leave is paid 44 hours at their normal rate — not 40 plus 4 at time-and-a-half. The pay run says so on the line.

Leave across two pay periods is split. A week off that straddles a period boundary is paid in two parts, each on the run that covers those days, and the line tells you when that has happened.

It is paid once. Locking a run claims the leave it paid, the same way it claims timesheets, so no later run can pay it again.

NOTEIf a caregiver has no paid work on file at all, Carely cannot work out a rate and will not guess one. The line says so instead of paying zero quietly — pay it by hand, or record a visit at their rate and rebuild the draft.

What approving does to the balance

Approving spends the hours. The request's hours come off that caregiver's remaining balance for that category, as part of approving it — not as a separate step you could miss.

• Approving the same request twice spends nothing extra.
• Taking an approval back returns the hours. Deny or cancel a request you had approved and the balance goes straight back up.
• Correcting an approved request moves the difference. Change it from 32 hours to 40 and 8 more come off; change the category and the hours move with it.
• A category with no allowance set shows N/A, and approving against it spends nothing — there is nothing to spend. Set the allowance first if you want it tracked.
• By default, paid or unpaid makes no difference to the balance. The allowance is how much time off you grant, not how much of it you pay for, so every approved absence comes off it. You can change this per category — see below.

Approving the pay is refused if it goes past the allowance. That is the decision that puts money on a pay run, so a mistyped figure is a day's pay to claw back afterwards. Carely tells you what is available and by how much the request is over, and offers Approve the pay anyway — which is recorded against your name. Their time off is not affected either way.

If no allowance is set for that category, there is nothing to exceed and nothing is refused.

Approving the time off is never refused — only warned about. The warning names the allowance it is about — where a category has a separate paid one, approving the pay shows what that pot has left and how far past it you are going — and it stays on screen until there is no decision left to make.

NOTEOnly approved leave spends anything. A pending or denied request never touches the balance, so what you see beside a request is what the person has left if you say no.

Leave dates, and why they matter

A request asks for a first day off and a last day off (leave the second blank for a single day), and optionally only part of the day with a from/until time. Those dates are what let Carely check the request against the schedule — a request without them is recorded and approved as normal, but cannot be compared to anybody's visits.

Requests made before Carely recorded leave dates show "Dates not recorded" and are skipped by every check below. Carely does not try to read the old free-text wording: deciding that "Jun 20–24" meant this year rather than last would mark somebody unavailable on days they never asked for, and the mistake would look exactly like a correct answer.

To fix one: Operations Hub → Leave, find the request, and select Record the dates beside the warning. Give the first and last day off (the same date for a single day), and check only part of the day if it is not a whole one. Saving makes the request checkable from then on — and immediately lists any visits already booked inside those dates, in the same resolver approving uses. Recording dates moves no visit by itself; you decide what happens to each one. The original free-text wording is kept beside the dates, not overwritten.

Part-day leave only blocks the hours it covers. An afternoon off does not clear the morning's visit. In a range with times — Monday 13:00 to Wednesday 12:00 — the times apply only at the two ends; Tuesday is off entirely.

Approving leave when visits are already booked

When you approve leave, Carely lists every booked visit that falls inside it — and changes none of them.

For each visit you choose:

Carely will never empty somebody's schedule on its own. A system that silently released twenty visits because a holiday was approved would be worse than one that ignored leave, because the office would not know it had happened. Every one of these is a decision about somebody's care, so every one is made by a person.

You can also close the list and deal with them later — the visits stay exactly as they are.

Scheduling someone who is on leave

Carely warns and lets you proceed. Sometimes the caregiver has agreed to come in, and an agency that cannot record that in Carely will record it nowhere.

The warning appears when you book a visit, move one onto a leave day, or assign someone — and across a whole recurring series, so two weeks booked over a holiday are caught rather than just its first day. Every override is recorded with who did it, when, which caregiver, which leave, which visit and the reason given, readable in the leave override log.

A caregiver may claim an open shift during their own approved leave. That is their call — their employer's software should not stop someone picking up a day of their own leave — and it is recorded so the office can see it and the balance conversation is not a surprise.

Cancelled or denied leave stops blocking immediately. The caregiver is back on the schedule with no warnings.

6. Nurse and clinical workflows

NOTEThis whole section applies only if your agency has the clinical module switched on. Companion-care and non-medical agencies will not see any of these screens, and nothing here affects them. If your left navigation has no Clinical tab, skip to Scheduling.

What the clinical module adds

Ordinary home care does not need medication records or physician orders. Skilled nursing does. The clinical module adds that layer without changing anything about how companion care works:

• Clinical profiles on clients, with medications and clinical history
• eMAR — the electronic medication administration record
• Physician orders
• Structured clinical documentation with a sign-off step
• License-aware scheduling, so only qualified clinicians take skilled visits
• A clinical view for families
• Clinical integrations (EHR, drug terminology)

Turning the clinical module on

Who can do this. Administrator.

1. Avatar menu → Account settings.
2. Find Clinical / Skilled-Nursing module.
3. Select Activate Clinical module… and complete the form:
4. Care model — nurse registry, home health agency, or both.
5. States — pick each state the agency is licensed in from the list. Each one appears as a chip; the ✕ on a chip removes it. Pick as many as apply.
6. License type and number — the agency's own home health or nurse registry license.
7. Director of Nursing — the registered nurse legally responsible for clinical care. Every state that licenses home health requires one, which is why it cannot be left blank, and a name alone is not a credential: the DON goes on the record with a nursing license number. Pick them from your nurse roster and the license on their clinical credentials becomes the license of record (if they have no license number on file yet, add it under Caregivers first). Someone not yet on the roster can be named — type their name, license type and license number. The number is kept on the record for a surveyor and shown on screen only as "license on file". Clinical Supervisor and Medical Director are optional; a small agency's Director of Nursing often fills both roles.
8. Clinical sign-off — a licensed clinician's name, license type and number, and the acknowledgements. This is stored as a permanent record.

How to verify it worked. The panel shows Clinical module active ✓, and a Clinical tab appears in Operations Hub.

IMPORTANTActivation is agency-wide, not per person. It adds clinical screens everywhere they apply. If you provide only companion care, leave it off — the extra screens are noise you would have to explain to every new hire.

Adding a nurse

Nurses are caregivers with a license.

Add nurse appears once the agency has said it provides clinical care — before the module is activated, so the Director of Nursing can be on the roster when activation asks for them.

1. Operations Hub → Caregivers → ＋ Add nurse.
2. Complete the ordinary caregiver fields, plus Clinical credentials:
3. License type — RN, LPN and so on.
4. License number — stored encrypted.
5. Licensed state(s) — pick each state the license covers from the list; each appears as a chip you can remove.
6. License expiration.
7. NPI number.
8. Compact / multistate license?
9. Save.

How to verify it worked. They appear in the caregiver list, and they become selectable in the Clinician dropdown when you create a clinical visit.

Clinical visit types

When the module is on, Schedule → New shift gains a Visit type field:

Choosing a clinical visit type changes the Caregiver field to Clinician, and Carely checks the person's license against the requirement.

If they are not qualified, you see: "Not credentialed for [visit type]" with the reason. An administrator may select Assign anyway (supervised — this override is logged).

WARNINGA clinician cannot override this for themselves. Self-claiming a skilled visit they are not credentialed for is refused outright: "You aren't credentialed for this visit." Only an administrator can assign an exception, and the override is recorded in the activity log with your name on it.

Who may do clinical work

Clinical authority in Carely follows the license, not the job title.

Administering a medication, writing a clinical note and signing a physician order require a current nursing license on record. Being an agency administrator is not enough — your agency's license is not a practitioner license, and the person who runs the office is not automatically the person who may push a dose.

An administrator is not locked out of the clinical picture — they see all of it. What they do not have is the ability to perform it.

If you own the agency and you are a nurse, you keep doing clinical work. Record your own license in Account settings → My clinical license. From then on you practice under that license: it expires like any other nurse's, and every dose you give is signed with your name like any other nurse's. Until it is recorded, Carely treats you as an administrator.

An expired license grants nothing. On the day a license lapses, that person's clinical actions stop — for the owner exactly as for anyone else. Everything stays readable; only the ability to act ends. Renew it in the same place and the access returns immediately.

The Nurse Station (nurses)

Licensed nurses get their own app — Nurse Station in the sidebar. It holds the patients' medication record, clinical documentation and physician orders: everything a nurse does under their license, in one place. It appears only for people who hold clinical scope, so an aide, a scheduler or an unlicensed administrator never sees it.

It deliberately does not show the agency-wide rollup or the integration settings — those are the administrator's, not the bedside's.

Clinical oversight (administrators)

Operations Hub → Clinical. It contains:

Asking a nurse about their work. Next to any dose in the administration history is an Ask button. It sends your question to the nurse whose name is on that record — that nurse only, not the whole team — and it reaches them in their alerts. It does not change the record. Use it for the things oversight is actually for: an unsigned note, a missed dose, an order still waiting on a signature.

Adding a medication

Nurse Station → Medication record → Add medication. Pick the patient (every active client on your roster is offered), type the medication, and Carely codes it to RxNorm as you type — the standard drug vocabulary from the U.S. National Library of Medicine, read live from their public service, or from your own drug-terminology vendor if you connected one. The RxCUI appears under the name when it matches; a name with no match can still be saved and coded later.

Dose, units, route and frequency have their usual choices; mark PRN, High-risk or Controlled substance where they apply — the medication intelligence flags and the controlled-substance audit read those marks. Once added, the medication is on the eMAR with Administer, Hold and Discontinue beside it, and, when an EHR is connected and live, the patient's coded medication list is sent to it.

Stopping and holding a medication

Setting a medication to discontinued stops it. Nothing can be recorded against it afterwards — not given, and not missed or refused either, because there is no longer a dose to account for. If the drug is restarted, set it back to active first, which leaves a trail of who resumed it and when.

Held is different: the order is still live, the drug is temporarily withheld. Carely refuses to record it as given, but a nurse can and should still document the hold on the MAR.

Medication administration (eMAR)

The MAR records every dose: what was given, when, and by whom. Medication intelligence watches the same data and flags doses that are due or overdue, possible interactions, and duplicate entries.

IMPORTANTThese flags are prompts for a clinician, not clinical decisions. Carely does not decide whether a medication should be given. A qualified person does, and records what they did.

Clinical documentation

Structured forms rather than free text, so the same information is captured every time and can be signed. Notes are signed by the clinician who wrote them.

Clinical integrations

Operations Hub → Clinical → Clinical integrations. These use your credentials with your vendors:

Connecting your EHR. Click Connect, pick the vendor, and the form says where each credential comes from. What you enter is what your EHR's app registration gave you:

• Epic — register a Backend Systems app at fhir.epic.com. Epic does not use a secret: you generate an RSA key pair, upload the public key to the app, and paste the private key (the whole PEM block) into Carely with the Client ID. Carely signs a short-lived token request with it each time. The non-production Client ID works against Epic's sandbox until your organization's Epic team approves the production one.
• Oracle Health (Cerner) and athenahealth — a System app gives you a Client ID and Client Secret; enter those.
• Every vendor needs your organization's FHIR R4 base URL, which is on the same registration page. Carely discovers the EHR's token server from it, so there is nothing else to find.

The connection is tested the moment you save: Carely asks the EHR's authorization server for a token with your credentials and reads the FHIR capability statement. A refusal comes back in the EHR's own words, on the form, with what to change. Send test then sends a clearly-marked test bundle so you can see it land in the EHR before any patient's data does.

Each connection has a sandbox and a production setting; nothing is sent automatically until the connection is in Production and has verified. See EVV transmission — the same safeguard applies, for the same reason. A sync that fails raises an alert in the bell and is retried.

What the clinical module does not do

• It does not make clinical decisions or recommend treatment.
• It does not replace your clinical judgment, your policies, or your state's rules.
• Payer authorizations are handled outside the clinical module, in Operations Hub → Authorizations — approved units, service codes, dates and scheduling warnings all live there. → Authorizations
• Lab ordering and results are not available.

7. Scheduling

What the schedule shows

The calendar holds a working window — roughly the last four months and the next eight — because a schedule is what you are staffing and correcting, not an archive. If a single period holds more visits than Carely can show at once, it tells you the count and asks you to narrow the range, rather than quietly dropping the rest off the end.

The Schedule app is every visit on the books.

The counts along the top are your morning check: how many visits in total, and how many still need a caregiver.

Creating a visit

Purpose. Puts one block of care for one client on the schedule. This is the record that later produces the EVV punch, the timesheet, the pay and the invoice.

Who can do this. Administrator.

Before you begin. The client must exist. If you are assigning someone, the caregiver must exist and their required credentials must be current.

Steps

1. Open Schedule → New shift.
2. Client — choose from the list.
3. Visit type — clinical agencies only. Leave on Companion / non-clinical for ordinary care.
4. Caregiver (optional) — choose a person, or leave it on — Open shift — to post it for caregivers to claim.
5. Start date and End date. For a single visit, leave them the same day — which is how they start. For a run of visits, set the end date to the last day the client needs cover and Carely fills the weeks between.
6. Start and End times, set with three small dropdowns — hour, minutes, and AM or PM. There is no 24-hour clock to work out.
7. Repeat weekly — One-time, or a number of weeks. This and the end date are two views of the same run: change either one and the other follows, so they can never disagree.
8. Read any warning that appears (see below).
9. Select Create shift — or Create [n] visits for a run.

What happens next. The visit appears on the schedule. If you assigned someone, it shows on their My shifts. If you left it open, it appears in Open shifts for anyone eligible to claim.

How to verify it worked. The visit appears in All shifts on the right date with the right person, and the open-shift count at the top changed as you expected.

TIPSchedule the time the visit really takes. The schedule feeds payroll and invoicing, so an optimistic schedule quietly becomes an inaccurate invoice.

Overnight visits

A waking night or live-in visit runs from one day into the next. Book it as one visit, on the night it starts.

Set the start to the evening time and the end to the morning time — for example 9:00 PM to 9:30 AM. Because the end reads earlier on the clock than the start, Carely takes it as running into the next day, and says so under the form: "Overnight visit — 12.5 hours, finishing 9:30 AM the next day." Check that line before you create the visit; it is the number that will reach payroll and the invoice.

Everything downstream treats it as the single night it is:

• Hours count the whole span — 9:00 PM to 9:30 AM is 12.5 hours, not minus 11.5.
• The visit is filed on the night it began, so the hours land in that day's period.
• Double-booking is caught across midnight. A caregiver working 9:00 PM to 9:30 AM cannot also be booked later that evening, or the following morning. A day shift starting at exactly 9:30 AM is a handover, not a clash, and is allowed.
• Nobody is chased for running late while the night is still in progress.

The one thing Carely will not accept is a start and an end at the same time — that is not a visit. For a full 24-hour period, book two visits.

NOTEA visit longer than 24 hours cannot be entered. Live-in care is booked as one visit per night.

Warnings on the create form

Carely stops you before you create a problem. Each message tells you what to do.

NOTECarely cannot know about a second job at another agency. Conflict detection covers your schedule only.

Recurring visits

A run of visits is set by its two ends: a start date and an end date. Carely creates the visit on the same weekday each week between them, and tells you the range before you commit: "Creates 6 weekly visits, [first date] through [last date]."

Repeat weekly is the same fact said the other way, for when you think in weeks rather than dates. Change the end date and the number of weeks follows; change the weeks and the end date follows.

If the end date does not fall on a visit day — you set it to a Friday when the visits are on Wednesdays — Carely stops at the last visit that fits and says so: "The end date is [date], and the week after [last visit] would run past it." An end date before the start date is refused, with the reason on screen.

WARNINGIt creates separate visits, not a linked series. Change one and only that one changes. There is no "change all future occurrences," and no way to end a series early in one action. Start with two to four weeks. Extending is one form; unpicking twelve weeks is twelve edits.

Repeat covers one weekday. Monday, Wednesday and Friday is three separate runs.

Open shifts are a feature

A visit with no caregiver is an open shift. Caregivers see it in their Open shifts tab and can claim it themselves, with no approval step. It is the fastest way to fill a gap.

A caregiver can only claim a shift if:

• their own required credentials are current — they cannot override this;
• they are credentialed for the visit type, if it is a clinical visit; and
• it does not double-book them.

If any of those fail, they see a message explaining which one.

WARNINGReleasing a shift sends nothing immediately. There is no text, no push and no phone call at the moment you post it. Caregivers see it next time they open Carely, or in the daily open-shift digest — one email a day, listing everything open in the next 14 days, sent between 7am and 8pm and skipped entirely on a day with nothing open. That is useful for next week and useless for this afternoon: for a visit in the next few hours, post it and pick up the phone. → Open shifts: a daily digest

Changing an assignment

Open the visit and change the caregiver. The same conflict and credential checks apply as when creating. If the new person is double-booked, Carely names the clash and refuses.

Editing a visit

Open the visit and edit the Date, Start and End.

IMPORTANTOnce a visit is In Progress, administrators have no controls on it at all: no edit, no reassign, no delete. Only the assigned caregiver can clock out and end it. If they cannot — a dead phone, a caregiver who has left — the office closes it through Operations Hub → EVV Exceptions, which records a correction rather than editing the visit. See A visit is stuck In Progress.

When a visit is not going to happen

You have three options: reassign it, release it as an open shift, or delete it.

WARNINGThere is no cancel button. A visit that is not happening is deleted, which removes the evidence it was ever scheduled — including why. Before you delete, write a note on the client record. It takes fifteen seconds and it is the only trace you will have when someone asks in three months. The one place Carely will mark a visit Cancelled rather than delete it is when you approve leave and choose Cancel visit for a conflicting booking — there the visit is kept, so the record of it survives. → Approving leave when visits are already booked

Visit statuses

AI auto-schedule

Schedule → AI auto-schedule proposes assignments for the open shifts already on your calendar, and shows why it chose each person.

What it will not do. A caregiver is never proposed if they are already booked at that time, have not marked the day available, are missing or have an expired required credential, lack the license a clinical visit type requires, are on approved leave, or would pass their own weekly hour limit. Going into overtime is flagged rather than refused — and against your agency's own threshold, not a fixed 40 hours, so an agency that pays overtime after 35 sees the warning at 35. In a state with a daily rule, a long day is flagged too. Where you can override some of those by hand on a single visit, the auto-scheduler cannot: one override in front of you is a decision, fifty across a week is not.

Why it picked somebody. Each proposal carries its reasons — already works with this client, on the care team, marked available, has room in their week, has the skills this client's needs call for. Continuity counts for most, deliberately: a client who has had the same caregiver for months should not be reshuffled for a marginal gain elsewhere. A caregiver with nothing on file for some of a client's needs is flagged ("has nothing on file for…") rather than refused — a thin record is not proof of a thin skill set — so you see it before publishing.

Carely's AI. When your agency has a model configured, the whole run goes to it in one request: for each visit, the client's needs and the top few caregivers the rules allowed, as facts without names. It chooses among them — a dementia specialty over a lighter week, a shared language over a marginal continuity edge — and its one-sentence reason appears under the visit as Carely AI. It can only ever choose someone the rules already cleared and who is free at that time. If it is not configured or does not answer, the rules' choices stand and the screen says so. Either way, every name is a dropdown you can change before Publish. Where nobody could be found, it says which constraint stopped each candidate rather than just reporting a gap.

The caregiver is told. Once you publish, a visit the AI helped choose shows the caregiver "Carely AI helped suggest you for this visit. Your office reviewed and approved it." in My shifts. If you later move the visit to someone else by hand, the note goes away.

If it fills nothing, look at credentials first. With credential blocking switched on — which is the default — a caregiver whose required documents are not uploaded yet is not eligible for anything. On a brand-new agency that can mean every caregiver, so the first run comes back entirely unfilled and looks broken. It is not: open a proposal and it names the constraint that stopped each person. Upload the credentials, or turn blocking off in Certifications, and run it again. → Adding a credential

IMPORTANTIt proposes. Nothing is assigned until you publish it, and publishing re-checks every rule at that moment. Check the proposal against what you know about the people involved — Carely does not know that two of them do not get along, or that one asked for lighter weeks.

If somebody edits a visit while you are reviewing, the proposal for that visit is refused rather than applied, and Carely tells you which one and why. The person who made the change wins; regenerate to pick the schedule back up.

Standing arrangements (recurring schedules)

Operations Hub → Recurring Schedules.

The rule behind the visits — "Ada, Mondays and Wednesdays, 9–1, with Cara" — rather than the individual visits it creates. Change the arrangement here instead of editing every remaining visit by hand.

Ending one tells you exactly what it will remove and what it will keep before you confirm.

Visits that have already been worked are never removed. Anything completed, or with a clock-in against it, is delivered care and stays on the record — ending an arrangement only clears upcoming, unworked visits.

Open slots — arrangements with no caregiver assigned — are flagged, so a standing commitment nobody is covering does not sit unnoticed.

Filtering and finding visits

All shifts carries a week grid and search. To see one caregiver's week, filter by their name. To see one client's schedule, open the client.

8. Electronic Visit Verification (EVV)

One visit, one timesheet

A visit can be clocked in once and clocked out once. Tapping Clock in again on a visit that is already running, or Clock out again on one that is finished, is refused with a message rather than accepted — so a double tap, or a retry after a patchy signal, cannot turn one visit into two.

That matters in both directions. A second clock-out used to create a second payable timesheet, so the family was billed twice and the caregiver paid twice for the same hours. A second clock-in used to reset the start time, so a full day's work could record as zero hours.

If a visit was genuinely recorded wrong, an administrator corrects the timesheet — the EVV record of what the device actually reported is deliberately not editable, because it is the evidence the visit happened.

Visits can only be booked for people on your roster

A visit needs a client and a caregiver who both exist and are both active. A name that matches nobody is refused when you schedule it, rather than becoming a visit that is worked, paid, and exported to your payroll provider for someone who does not exist.

Discharged clients and deactivated caregivers cannot be scheduled. Readmit the client, or reactivate the caregiver, if care is resuming.

What EVV is, and why it exists

Electronic Visit Verification is proof that a visit actually happened: the right person, at the right place, at the right time. Many payers — Medicaid in particular — require it before they will pay. Even for private-pay work it settles the question of what hours were really worked, so payday is not a debate.

What Carely captures

A visit is EVV-verified only when both punches are verified. Clocking in correctly and clocking out from the parking lot gives an unverified visit.

Choosing how your agency does EVV

Administrators choose the mode in the Schedule app:

Integrated mode works in any state and does not require Carely itself to be certified, because your vendor remains the system of record.

The caregiver's workflow

1. Open Schedule → My shifts.
2. On the visit, select Clock in. The phone asks for location permission the first time — allow it.
3. Provide care.
4. Select Clock out & complete visit.
5. Complete the visit report: Client mood, Signature (client/caregiver) and a Visit note (shared with family).
6. If the client is billed mileage, answer Driving during this visit: No driving, or I drove — add trips and pick each place From → To as you type it. Carely measures the miles and pays them back to you. → Mileage

IMPORTANTIf location does not work, clock in anyway. Carely never blocks a visit because of a phone. The visit is recorded and flagged for review. An unverified visit takes a minute to resolve; a visit with no record at all is a hole in your billing and your audit trail. Say this to every caregiver during training. It is the single most useful sentence in this chapter.

There is no time window. Carely does not refuse a clock-in for being early or late. The time it records is the real time, and any variance shows up on the timesheet for you to see.

Clocking in works on any device that knows where it is

ALIASES: no location, no gps captured, clock in on a laptop, clock in on a tablet, computer clock in, gps not working, location not captured, clock in without gps, browser location

A phone, a tablet or a laptop can all record EVV — anything that can report a location and runs Carely in a web browser. A phone is the most accurate, because it has a satellite receiver. A laptop usually locates itself from the Wi-Fi around it, which is less precise but is normally well within the range EVV asks for.

Carely asks twice, in two different ways. Devices without a satellite receiver often cannot answer a request for a precise fix at all, so if the first attempt finds nothing Carely immediately asks for the network-based location instead. That second attempt is what makes a laptop or a desk tablet work.

A laptop's fix counts. A network (Wi-Fi) fix is coarser than a phone's — tens of meters, sometimes a few hundred. Carely keeps which kind of fix each punch was: a fix within 200 m is GPS; one between 200 m and 1 km is a network fix, and it verifies the visit when it places the device inside the client's location, recorded as a network fix so an auditor sees what kind of evidence it is. Only a fix coarser than a kilometer is recorded as too imprecise to place the visit. A network fix never sets a client's location anchor — only a GPS-grade fix does, because every later visit is measured against it.

The installed desktop app asks the computer's own location service — no key, no setup. The first clock-in on a Mac shows the system prompt "Carely Location Helper" would like to use your location: click Allow (Carely waits up to 45 seconds for the answer). On Windows there is no per-app prompt; the switch is Settings → Privacy & security → Location → Location services and Let desktop apps access your location.

If somebody said no. The operating system asks only once. From then on, every clock-in without a location shows Carely's own dialog saying exactly what to turn on, with an Open Location Settings button that opens the right pane (macOS: Location Services → turn on Carely Location Helper; Windows: the Location page). In a browser, the address-bar lock or tune icon → Location → Allow, then reload. The dialog appears on every attempt until location works; "Clock in anyway" still records the visit, flagged for review.

When a location will not come. No Wi-Fi (a computer finds itself by the networks around it), no internet, or Location Services off — the dialog names the check to make.

If your device still gives nothing, Carely says why before it records the punch — the location permission is off, the device cannot report one at all, or it simply took too long — and offers to cancel so you can use a different device.

A visit with no location is not sent to the state. If your agency has connected a state EVV aggregator, that visit is held rather than transmitted, and Operations Hub → Medicaid shows it as Held with the reason. Location is the whole point of Electronic Visit Verification, so Carely will not submit a record without one and let you believe it counted.

To release it, raise and close an exception for the visit under Closing a visit when the evidence is missing. Closing it sends the visit on, with the exception reason attached — an administratively closed visit still is not a device-verified one, and the state is told which it is.

Carely tells you when it holds one. A held visit puts an item in your notification bell — and sends the same notice to your administrators' phones — saying which date is waiting and what to do about it. You do not have to remember to open the Medicaid screen.

The reminder stays up until the visit is released, and then takes itself down. It is not a message you dismiss and lose; it clears when the visit actually goes. Nothing asks you twice for something you have already done.

The location anchor

The first clock-in at a client automatically establishes that client's location anchor — but stores it unconfirmed.

While the anchor is unconfirmed, Carely records the distance and enforces nothing. Visits show the note "client location not confirmed yet — not enforced" next to the EVV badge.

That behavior is deliberate. If the first clock-in happened in the parking lot or at the end of the street, an automatically-trusted anchor would enshrine the wrong location and fail every honest visit afterward.

IMPORTANTUntil an anchor is confirmed you have complete GPS and time records but no geofence enforcement: out-of-range visits are recorded and shown, not rejected. Do not read a "verified" badge as "inside the geofence" while the anchor is unconfirmed.

Confirming a client's location

Purpose. Turns the geofence on for that client. Until you do this, distance is recorded but nothing is enforced.

Who can do this. Administrator.

Two ways, and the second is usually better.

From the client record — Operations Hub → Clients & Matching. Each client shows an EVV location row:

• not set — nobody has clocked in yet. Set it now lets you enter coordinates.
• Not confirmed · not enforced — set automatically from a clock-in. Confirm this location if the coordinates are right, or Enter the correct one if they are not.
• Confirmed · geofence enforced — done. Change if the client moves.

From a visit you trust — Schedule → All shifts → Verify location → Use as client's location on either punch. This is usually the better route: it takes the coordinates from a real punch you know happened at the house, rather than asking you to find coordinates from somewhere else. It is also the moment you normally notice the anchor is wrong.

What happens next. From then on, a punch more than 250 m from that point is recorded as outside the client's location and the visit is not EVV-verified.

NOTEThe 250 m radius is the same for every client and cannot be changed. It is generous on purpose: it absorbs ordinary GPS wobble and large properties without waving through somebody a block away. If you have a client where it is genuinely wrong — a big rural property, or a large apartment complex with the anchor at the wrong entrance — tell us, because that is the case that would justify making it adjustable.

How to verify it worked. The client's EVV location row reads Confirmed · geofence enforced, and the coordinates are shown beneath it.

Working through your whole client list. There is no single screen listing which clients are still unconfirmed — you check them on their cards in Clients & Matching, where the status shows on every client. Two approaches that work:

• New agency: confirm each client as their first visit happens. The first clock-in sets the anchor; open that visit, check the distance looks right, and confirm from the punch.
• Existing agency: work down the client list once and confirm the ones that already have an anchor. Budget a few minutes; it is a one-time job per client.

Until a client is confirmed, their visits still record everything — you simply have no enforcement for that client.

WARNINGConfirm only coordinates you actually believe are the client's home. Confirming a wrong anchor is worse than leaving it unconfirmed: every honest visit afterward fails the geofence, and the failures look like the caregiver's fault. If the distances on a visit look wrong for one client, suspect the anchor first.

Every confirmation is recorded in the activity log with your name on it.

Reading an EVV result

Each visit shows EVV verified or Not EVV verified, and when it failed, why.

IMPORTANTOut of range is not evidence of dishonesty. The most common cause is an anchor set from a first clock-in that happened somewhere slightly wrong. Getting this the wrong way round costs you good staff. Check the anchor first, then talk to the person.

Verifying exactly where a visit happened

Purpose. Shows you precisely where a caregiver clocked in and out, so you can confirm a visit for payroll, a payer, or an audit.

Who can do this. Administrator only. Caregivers see their own punches but not this panel; families never see caregiver locations at all.

Steps

1. Open Schedule → All shifts and switch to the List view.
2. Find the completed visit.
3. Select Verify location.

What you see, for each punch

Below that is a plan view: the client at the center, the geofence as a dashed circle, and a pin for each punch with a line back to the client. A shaded ring shows how precise that GPS fix was. It is drawn inside Carely — no map service is contacted, so nobody outside your agency learns where your clients live.

Open in Google Maps is there when you need a real map, but it is a deliberate click: following it sends that location to Google. Carely never does so on its own.

How to verify it worked. You see two pins and two distances. A clean visit has both pins inside the dashed circle.

TIPThe pattern to look for is a clock-in inside the circle and a clock-out well outside it. That is somebody leaving early and closing the visit from the road, and it is invisible on the schedule — the visit still shows as Completed with the right hours.

IMPORTANTBefore you act on a distance, check whether the client's location is confirmed. The panel says so when it is not. An unconfirmed anchor was set from a first clock-in that may itself have been in the wrong place, which makes every later distance wrong in the same direction. See The location anchor.

The EVV decision guide: caregiver cannot clock in

Work down this list. Most cases are resolved by step 3.

1. Does the visit exist? Look in Schedule → All shifts for that client and date. If there is no visit, create one — a caregiver cannot clock into a visit that was never scheduled. → Creating a visit
2. Is it assigned to them? An open shift has no Clock in button for anybody. Assign it, or have them claim it from Open shifts.
3. Are they looking in the right place? Clock in appears in Schedule → My shifts, not on the Operations Hub dashboard. This is the most common answer of all.
4. Do they have a login, and is it approved? A caregiver record alone shows them nothing. Check Operations Hub → Users. → Adding a user
5. Does the name on their login match the name on their caregiver record? Carely connects the two by name. A middle initial on one and not the other separates them.
6. Has the visit already been started? A visit already In Progress has a Clock out button, not Clock in.
7. Is the visit already Completed? Then it was clocked out. Check whether somebody already did it.
8. Is their device online? Clocking in needs a connection. Carely does not queue punches offline.
9. Has location permission been denied? This does not block clocking in — they can and should proceed. The visit will simply record no GPS captured.
10. Still stuck? Contact Carely support with: the client name, the caregiver name, the visit date and time, what they see on screen, and the exact message. → What to send support

The EVV decision guide: caregiver cannot clock out

1. Is the visit In Progress? Clock out only appears once clocked in.
2. Is the person clocking out the assigned caregiver? Nobody else can — not another caregiver, and not an administrator.
3. Are they on Schedule → My shifts?
4. Are they online?
5. If they are unable to reach the app at all — lost or broken phone — the visit stays In Progress until they can. There is still no way for anyone to clock out as them. What the office can do instead is raise an EVV exception, which closes the visit with a correction that is labeled as entered by a person rather than measured by a device. → Closing a visit when the evidence is missing

Closing a visit when the evidence is missing

Operations Hub → EVV Exceptions.

Phones die. Caregivers forget. There is a fall and nobody thinks about the app. The visit still happened, the caregiver still has to be paid, and the payer still has to be billed.

This screen is how the office deals with that without anybody editing a clock time.

This list does not chase you. A visit reaching this screen has already been nudged once — an hour after it was due to end, Carely emails the caregiver and raises it in the office's bell. What it will not do is remind you again about the ones nobody acted on. So still open this screen as part of the weekly run, before you bill. → When a visit is still running

What it shows you

• Visits needing attention — visits still running more than 16 hours after clock-in (which is a clock-out that never happened, not a long shift), completed visits carrying a punch the device could not verify, and completed visits that recorded no time at all. It should be rare for one to get this far, because of the nudge an hour after the visit was due to end.

> No time recorded means the clock-in and clock-out landed within a few minutes of each > other on a visit booked for real hours — a double-tap, a phone in a pocket, a punch that > sent twice. Hours are rounded to the quarter, so the visit ends up worth zero. Both > punches can even be GPS-verified: verification says the caregiver was at the address, not > how long they stayed. Carely holds these off the billing worksheet until you correct > them, because billing a nought closes the visit off for good and the corrected hours > could never be charged afterwards. A genuinely short visit — a quarter of an hour because > the client was in hospital — is real time and is left alone.

• Open corrections — ones you have raised and not yet closed.
• Corrections already applied — the permanent record, and where you reverse one.

How to correct a visit

1. Find the visit and choose Deal with this.
2. Say what happened — a dead phone, no signal, an emergency — and describe it in your own words. This is the record an auditor reads. It is kept permanently and cannot be edited afterwards.
3. Enter the corrected times and apply.

What Carely will and will not do with those times

This is the point of the whole screen. "Verified" in Carely means a device was there. A system that will mark an office keystroke as verified is worth nothing to you in an audit. Every correction says on its face that a person asserted it.

The caregiver still gets paid

Closing a visit this way produces a normal Pending timesheet for the corrected hours, which goes through your usual approval and pay run. It is marked EVV correction rather than EVV, so payroll can see at a glance that these hours were asserted by the office rather than measured by a device. The client's authorized units are drawn down as they would be for any other visit.

If the visit already had a timesheet, that timesheet moves to the corrected hours instead. A timesheet already inside a locked or paid pay run is not rewritten — Carely records a payroll adjustment against it, so money that has already gone out stays reconciled.

Corrections over 30 minutes need two people

A correction of more than 30 minutes does not take effect when you save it. It waits, visibly, until a second and different person confirms it. The same person approving twice is one opinion, and Carely refuses it by name. Under 30 minutes, one approver is enough — correcting four minutes and adding four hours are not the same act.

Reversing a correction

Open Corrections already applied and choose Reverse, with a reason. The hours go back to what they were. Nothing is deleted — the reversal is a further record shown alongside the correction.

Who can do this

Administrators, and it is its own permission: EVV — resolving exceptions, separate from managing the schedule. An owner can leave a scheduler able to build the schedule but not able to attest to hours nobody witnessed. → Limiting what an administrator can do

Caregivers cannot raise or resolve an exception on their own visits.

A correction longer than 24 hours is refused as a mistyped date, because a wrong one would flow straight into pay and into a Medicaid claim.

External EVV and state systems

If you transmit to a state aggregator, this lives in Operations Hub → Medicaid. You supply your own credentials for your own vendor account — Claim.MD, Sandata or HHAeXchange.

Carely transmits a real visit only when the connection is set to production and has passed a live test. In sandbox, or before that test passes, nothing real is sent.

IMPORTANTThis is a safeguard, not a fault. A real visit landing in a sandbox is a visit your state never received, and you would not find out until the money did not arrive. Send a test, read your vendor's response, fix whatever it objects to, switch to production, then test again.

A completed visit queues for transmission automatically once a connection is live.

When a transmission is rejected, the message comes from your vendor, not from Carely — take it to them. Common causes: a member ID that does not match the state's record, a service code outside the authorization, or a visit outside the authorized dates.

Carely does track authorizations, and warns you about these before the visit rather than after the denial — see Authorizations: what a payer has approved. It warns; it never blocks a booking.

Looking up past visit verification

Operations Hub → EVV History.

Every verified visit Carely holds, including any brought across from a previous system, with the clock-in and clock-out times, where they happened, and how the location was captured.

This is the screen for the question "can you prove that visit happened?" — a payer audit, a family query, or a dispute months later. Filter by client, caregiver or date range.

Imported history is shown as imported. Carely does not re-transmit historical visits to a state aggregator — they were already submitted by whichever system recorded them, and sending them twice would create duplicates in the state's records.

Importing verified visits from a vendor

In integrated mode you can import your vendor's already-verified visits. Paste the export as CSV in the format shown on screen:

clientName,caregiverName,date,start,end,evvId

Each imported visit becomes a completed, externally-verified visit with a timesheet, and is reconciled against the matching scheduled visit. Imported visits show "via [vendor]" next to their EVV badge.

9. Visit documentation

The visit report

When a caregiver clocks out, Carely asks for the visit report:

What happens next. The report is filed against the visit and shared with the family automatically. It appears in Schedule → Visit reports for administrators, My reports for the caregiver, and Care Reports in Family Connect for the family.

Who can write one. A caregiver on the client's care team, a caregiver who has a visit with that client, and administrators. Anyone else is refused, so a report a family reads was always written by someone who was there or who is responsible for that client.

TIPTrain caregivers on the visit note specifically. It is the single piece of writing families actually read, and a good one prevents a phone call. "Mrs Davis ate a full lunch and we walked to the mailbox and back" is worth more to a worried son than any dashboard.

Incidents

Falls, injuries and other events are recorded in Operations Hub → Incidents. Caregivers can report and view; administrators see everything; families can see incidents for their own relative in Family Connect.

Record an incident as soon as it happens, while the detail is fresh.

Clocking in with no signal

What happens. If a caregiver clocks in or out where there is no cell or Wi-Fi signal, the punch is saved on their phone and sent automatically as soon as they have service again. They see: "No signal — this clock-in is saved and will send itself when you are back online." They do not have to do anything, and they can close the app.

The visit keeps the time it really happened. A punch captured at 9:02 a.m. in a basement and delivered at 11:40 a.m. is recorded as a 9:02 a.m. visit, not an 11:40 one. That is the time of service, and it is what a payer is owed.

The record says where the time came from. A punch the server saw as it happened is marked as such. A punch captured on a device with no connection is marked as device-captured, together with how long it waited before it arrived. Both are real visits; they are different kinds of evidence, and an auditor is entitled to know which is which.

GPS still counts. The location is captured at the door, at the moment of the punch, exactly as it would be online. Losing signal does not lose the location evidence.

You can see a punch that has not been sent. While anything is waiting, a band appears across the top of the screen — "1 clock punch saved on this device" — and it stays there on every screen until the punch has gone. It tells you whether you are offline, reminds you that the visit time is already recorded, and offers Try now if you want to force a retry rather than wait. When the queue empties, the band disappears on its own.

The visit report is saved too. The note, the client's mood and the care-plan checklist written at clock-out are held with the punches and sent with them. While one is waiting the band counts it — "2 clock punches and 1 visit report saved on this device" — and the family sees the report once it arrives, exactly once.

If the sign-in has ended, the punch waits for it. A sign-in lasts eight hours. If it ends before the phone finds service — a clock-out in the evening, service the next morning — nothing is lost: the band reads "1 clock punch saved on this device. Sign in to send it." and the punch goes as soon as the caregiver signs in again, still stamped with the time it really happened. A saved punch is only ever sent by the person who made it; on a shared device, someone else signing in is told that a record is waiting for its owner, and nothing more.

Opening the app with no signal still shows today's visits. The caregiver stays signed in, and the app keeps their own visits for the days around today on the phone, so a visit can be started in a driveway with no bars. A notice above the list says so — "No connection. These are your visits as they were at 8:15 AM" — because a schedule can change after the phone last heard it. What is kept is deliberately little: the client's name, the date and times, the visit's status and type, and the care-plan checklist. No address, phone number or care notes, and nobody else's visits. It is removed whenever the caregiver signs out or is signed out.

Two limits are worth teaching. The copy exists only if the caregiver has opened Carely with a connection, signed in, within the last day; and a sign-in lasts eight hours, after which they must sign in again, which needs a connection. In practice: open the app once at home or in the car before going in.

Signed out in the middle of a visit? They can still clock out. Carely signs a person out after fifteen minutes without touching the app, and with no signal there is no signing back in. The sign-in screen therefore offers one thing to a caregiver with a visit in progress on that phone: "Nia, your visit is still in progress — started 9:02 AM. You can clock out without signing in." Pressing Clock out records the time and location, takes an optional mood and visit note, and saves them on the phone. Nothing about the client is shown, and nothing is sent until that caregiver signs in, when it goes by itself. The visit is recorded as ending when they pressed the button, and the record notes that it was made from the sign-in screen. The care-plan checklist is not available there; the note is.

A visit still waiting to send says so. Until the punch arrives, the visit shows as started with "saved on this device, not sent yet" beside the time, and Waiting to send — EVV is checked when it arrives in place of a verified or unverified badge. There is no second Clock in to press by mistake.

If Carely refuses a saved punch, the caregiver is told. This is rare — it happens when the visit was deleted or already completed by the office while the phone was out of reach. A red line appears in the band: "A clock-in saved at 9:02 AM was not accepted", with the reason and Tell your office so the visit can be added for you. It stays until they press Dismiss. If a caregiver mentions it, record the visit yourself with the reason — see Closing a visit when the evidence is missing.

A punch that cannot be believed is not silently accepted. If a device's clock claims a punch happened in the future, or more than a week ago, Carely records the visit but stamps it with the server's time and flags it. A punch arriving a month late is not a connectivity problem — it is a correction, and corrections belong to an administrator with a reason attached.

Sending twice is not possible. Each punch carries its own identifier, so if a phone retries one it already sent — which is what usually happens on a dying connection — Carely recognizes it as the same punch. This matters most on clock-out, which creates the timesheet: a repeated clock-out can never produce a second payable entry.

What can be changed afterward

This rigidity is the point: EVV records that can be edited afterward are not evidence of anything.

10. Timesheets

A timesheet is one row of hours for one caregiver, one client and one date.

Timesheets arrive in two ways:

1. Automatically, when a caregiver clocks out. The hours come from the real clock-in and clock-out times, rounded to the nearest quarter hour — not from the scheduled times. Source shows EVV.
2. Manually, when a caregiver uses ＋ Log time on their own timesheets. Source shows Manual. Start and end are set in AM/PM — hour, minutes, AM or PM — and a shift worked through the night is entered as it happened: 9:00 PM to 6:00 AM logs nine hours. The running total under the form shows what will be submitted.

Find them in Operations Hub → Timesheets (Billing & Pay). Caregivers see My Timesheets.

Timesheet statuses

Approving timesheets

Purpose. Approval is the gate between "work happened" and "we can bill and pay for it." Nothing downstream moves without it.

Who can do this. Administrator.

Steps

1. Operations Hub → Timesheets. It opens on Needs approval whenever anything is waiting. → How the timesheet list is organized
2. Work through one caregiver at a time. Each row shows the client, date, hours, rate, pay and source; the card view on a phone also shows the variance between scheduled and actual hours.
3. For a client billed mileage, select the row's Mileage button to see the trips the caregiver logged, From → To with the miles. Trips with typed miles are marked. You can edit them, correct the total, or approve from that window. → Mileage
4. Select Approve — or Deny and give a reason.

When a caregiver's visits all look right, Approve in their heading approves them together; Carely first says how many, how many hours and how much pay. The visits you approve or deny stay where they are, marked Approved or Denied, so the list does not jump while you work; they leave the queue the next time you open it.

How the timesheet list is organized

Along the top, the filters Needs approval, Denied, Approved, Paid and All each show how many timesheets they hold. The search box narrows every one of them to a caregiver's or a client's name.

• Needs approval is grouped by caregiver, because that is how timesheets are checked. The caregiver whose oldest visit has waited longest comes first, and each person's visits run oldest first. The heading totals their visits, hours and pay, and flags anything from an earlier pay period — approve those first, before payroll closes — and any visit with no time recorded.
• Denied, Approved, Paid and All are grouped by pay period, newest first, using your agency's pay schedule. This pay period and last pay period are open; older periods are folded — select Show to open one.

A caregiver's My Timesheets is grouped by pay period the same way, with each period's hours and pay. If you send one of their timesheets back, the top of their list says so first, with a button to see it, read your reason and resubmit.

Approving the whole queue at once

When you have gone through the whole list, Approve all at the top of the screen approves everything still pending in one action. The button shows how many that is. With a search typed, it reads Approve all shown and approves only the timesheets that match.

Carely asks you to confirm first, and the question tells you exactly what you are about to do: how many timesheets, across how many caregivers, how many hours and how much gross pay. Approved hours go straight into payroll and invoicing, so this is a money decision — the confirmation is there on purpose.

It approves the list you were looking at, and only that. If a caregiver clocks out while you are reviewing, their timesheet is not swept in — you never saw it, so it stays pending for next time. Carely tells you afterwards if anything was left alone.

Timesheets you have already approved or denied are not touched. A denied one is never quietly flipped to approved by a bulk action; if you want to reverse a denial, use Approve anyway on that row.

The whole batch is recorded in the activity log as a single decision, with its size — so an audit shows one person approving forty timesheets at 9:04, rather than forty separate events that look like careful individual review.

What happens next. Approved hours flow into Payroll and become available to invoice. Denied rows are excluded from both, and the caregiver is notified with your reason.

How to verify it worked. The row shows the Approved badge, and the client appears in the invoicing worksheet.

IMPORTANTUnapproved timesheets are invisible to billing, with no warning. This is the most common cause of "the invoice is missing hours." Make the weekly order explicit: approve → bill → pay.

Correcting a timesheet

WARNINGNobody can type over the hours on a timesheet — not the caregiver, not you. There is no edit field anywhere in Carely. This is deliberate: hours come from server-stamped EVV punches, and hours that can be typed over are not evidence. Hours can change, but only through an EVV exception: a correction recorded beside the original evidence, with a reason, an approver, and a second approver over 30 minutes. → Closing a visit when the evidence is missing

So a timesheet with wrong hours — usually a forgotten clock-out — is corrected like this:

1. Operations Hub → Timesheets. Find the wrong row.
2. Select Deny, and write a reason in Reason for denial. The caregiver sees this note. Be specific: "25 hours logged — clock-out was the next morning. Please log the real 4 hours."
3. Select Deny timesheet. The row becomes Denied and is excluded from pay and billing.
4. Ask the caregiver to add the correct entry: Operations Hub → My Timesheets → ＋ Log time.
5. Approve the corrected entry.

How to verify it worked. The wrong row shows Denied with your reason; a new row shows the right hours and Approved. Payroll totals now match reality.

NOTEThe caregiver's ↺ Resubmit for review button sends the same hours back for another look. It does not change them. Use it when the denial was a misunderstanding, not when the number is wrong.

You can also select Approve anyway on a denied timesheet, if you decide the original was right after all.

Exporting timesheets

Export CSV on the Timesheets screen produces: Caregiver, Client, Date, Scheduled Hours, Verified Hours, Rate, Gross Pay, Source, Status.

The scheduled-versus-verified pair is the useful part. Sort by the difference and your problem visits rise to the top.

Troubleshooting hours

11. Billing and money

Money lives in two places. Revenue Center is the main one; Operations Hub has the same invoices, payments and payroll under Billing & Pay for convenience.

Rates: charge and pay

Every client carries two rates:

• Client charge rate — what you bill.
• Caregiver pay rate — what you pay for that client.

By the hour or per visit. Each client is billed by the hour (the default) or per visit. By the hour, a visit bills hours × the charge rate and pays hours × the pay rate. Per visit, each completed visit bills the visit rate and pays the caregiver's visit rate, however long it ran. Choose it under Billed on the client's Care rates card (Operations → Clients & Matching → Edit), or on the client form.

Switching a client between the two is a new rate on both sides — $30 an hour is not $30 a visit — so Carely asks for the charge and the pay in the new unit, and the switch takes an In effect from date like any rate change. Visits before that date keep the terms they were delivered under, on the invoice and on the caregiver's pay; if visits have already been worked on or after the date, Carely asks whether the switch should cover them (see Changing a rate below). The rate history shows each period with what it was per.

Per-visit pay and overtime: a caregiver paid per visit who works more than 40 hours in a week (or over a daily limit, where your state has one) is still owed overtime. Carely works it out the way federal rules require for flat per-visit pay — the week's visit pay divided by the hours worked gives the regular rate, and each overtime hour adds half of it.

The gap between them is your margin. Both prefill from the client's care level (Low $28/$18, Medium $30/$20, High $35/$22) and should be adjusted per client.

IMPORTANTRates follow the client, not the caregiver. A caregiver's effective hourly rate therefore varies with who they saw that week. This is by design — pay tracks the difficulty of the care — but it surprises new office staff, and it surprises caregivers even more. Explain it during onboarding.

Who can see which rate. The charge rate and the pay rate are both the agency's business, and Carely shows each party only their own side of it:

So a caregiver never reads what you bill for their hours, and a family never reads what you pay the caregiver. Your margin stays inside the office.

Changing a rate: it applies from a date

Rates are effective-dated. Changing one does not overwrite the old figure — it opens a new period and keeps the old one, so a visit is always priced at the rate that applied on the day it happened.

This reaches the invoice, not just the reports. An un-invoiced visit is billed at the charge rate that was in force on the day it was worked. If you raise a client's rate on 1 July, work delivered in June still bills at June's rate when you invoice it in July, and the figure on the invoice matches the one the Growth Center reports.

Work already done is your decision, and Carely asks. Rates run by the day, so a rate starting today covers all of today — including a visit that finished this morning. When a change would reach work that has already been done, Carely stops before saving anything and tells you what it would move — how many visits, and the difference in money. You choose:

Visits that are already invoiced or paid are never re-priced by this; their money has already gone out. On the pay side, Carely raises a payroll adjustment for approval instead, as it always has.

When you edit a rate you are asked In effect from, which defaults to today.

A pay run that spans a change pays both rates. Raise a caregiver's rate from the 1st and a run covering the last week of March and the first week of April pays March hours at the March rate and April hours at the April rate, in one paycheck. You do not have to time a rate change around your pay calendar.

Reports agree with the records. Profit & Tax, the Growth Center and Copilot's estimates all price a past hour at the rate that applied then. Before this, they valued every past hour at today's rate, so raising a rate quietly rewrote what previous months appeared to have cost — while the money already paid stayed where it was.

To see the history, open the client and choose Rate history. Every rate, the days it covered, and who set it.

Rate history is read-only. It is written by changing a rate and cannot be edited or deleted from any screen. A record of what you used to charge that a screen can rewrite is not a record.

Rates already on file when this arrived are shown as applying from the beginning, and say so. Carely has no record of when they were originally set and does not pretend otherwise — so nothing you had been billing or paying changed when this feature landed.

Backdating into payroll you have already paid

If the date you choose falls inside a pay run that is locked or paid, Carely stops and tells you which runs, which caregivers, and what the difference comes to.

It will not reprice a paycheck that has gone out. If you confirm, the paid run is left exactly as it was and Carely raises a payroll adjustment for each affected caregiver, for someone to approve through the normal adjustments process. Your books stay reconciled and the correction is still recorded.

Approving a timesheet after its pay run has closed

Approving late is normal — a supervisor gets to it after the run was locked. That work is carried into the next run rather than lost, and the run says so: the line shows how many hours came from an earlier period and which dates they cover, so a larger paycheck is explained rather than mysterious.

Those hours are paid at the rate that applied on the day they were worked, not the rate today — so work carried forward from three weeks ago is not accidentally repriced by a raise that happened in between. → Changing a rate: it applies from a date

This only applies to work whose own period has already been locked or paid. Approved work from a period that was never run stays where it is, because that period can still be run.

Raising the same invoice twice

An identical invoice — same client, same amount, same date — posted again within a few minutes is refused, and Carely points you at the one that already exists. That is a double click or a retry, not a decision.

If you genuinely need a second invoice for the same client and amount on the same day, confirm it and Carely will raise it.

What Carely accepts as a rate

A rate must be a plain number. 25, 25.50 and 0 are all fine — zero is allowed, because unpaid and volunteer arrangements exist.

Anything ambiguous is refused with a message rather than guessed at: a currency symbol ($25.00), a thousands separator (1,234.56), a stray second decimal point (12.34.56), or a word. Carely will not read 12.34.56 as 12.34, because a rate it guessed wrong becomes somebody's pay.

Rates finer than a cent are refused too — rounding 0.001 down to 0.00 would quietly turn paid care into free care.

The Pay Rates screen

Operations Hub → Pay Rates (under Billing & Pay).

Every rate your agency uses, on one screen:

Rates are set per client because they follow care needs — a client requiring a higher-skilled caregiver carries a different rate from a companion-care client.

You can edit the same rates from the client's own record in Clients & Matching; both places write to the same figure. Changes flow straight through to invoices and payroll — there is no separate step to apply them, and nothing is recalculated retroactively for invoices you have already sent.

State forms

What this is. The forms your agency has to file and sign against a client's record. Two kinds live here: forms you upload yourself, and forms Carely has already built from a state's published documents.

Where. Operations Hub → State Forms (under Care & Compliance). Choose a client at the top right.

Uploading your agency's forms

Carely cannot hold an authoritative copy of every form every state, county and payer asks for, and waiting for one to be added is not a workflow. So Your agency's forms, at the top of the screen, is yours to fill.

Who can do this. Administrator. Caregivers and family members cannot upload or remove forms, and do not see this.

Steps

1. Form name — what you call it. If you pick the file first, Carely offers the file name.
2. Jurisdiction (optional) — a state, county or payer, so a long list stays sortable.
3. File — a PDF or an image, up to 12 MB.
4. Select Upload form.

What happens next. The form is listed with its size and who added it, and stored encrypted on your agency's record. Select Open to read or print it, Remove to delete it. Removing a form does not touch anything already filed against a client.

Uploading the same file twice is refused rather than duplicated — Carely compares the file itself, not its name.

This screen works before you have set a state. Your own forms are yours regardless of what Carely knows about your state.

Forms Carely has already built

Where Carely holds a state's published documents, the forms are pre-built: the screen shows which are ready to file, which are still pending, and everything already filed for that client. If Carely has nothing pre-built for your state, the screen says so plainly — upload what you file and carry on.

Virginia (DMAS). Three forms are built from the documents DMAS publishes:

DMAS-98R (Home and Community Based Services Request) is not built — DMAS publishes it as a Word document and Carely does not yet hold an authoritative copy.

Carely records the answers; the state's own document is what you file. The DMAS-99 says in print "Do not alter or revise this form in any manner", so Carely does not produce a substitute for it. What Carely does is keep the answers on the client's record, track the dates these forms depend on, and let you complete them once rather than from memory.

Check the revision before you rely on one. Each built form links to the published PDF it was transcribed from and shows which revision that was. States reissue forms; nothing here claims to have been checked against the current one, and the screen says so.

A form Carely does not have yet cannot be opened. If Carely does not hold the authoritative field list for a form, it says so and refuses to open it. That is deliberate: a regulatory form with guessed fields looks official, gets signed, gets filed, and is wrong — and nobody checks it a second time. Send us the current published form and it gets built from the real document.

Signing. A signature records who signed, in what role, and when — and is bound to a fingerprint of the exact answers as they stood at that moment. That is what lets the record prove what was signed, not merely that something was.

A signed form cannot be edited. Carely refuses the change rather than quietly accepting it. If a signed form is wrong, void it with a reason and file a corrected one. The voided form is kept, along with the reason and who voided it, because "we amended it" and "it never happened" are different claims and an auditor is entitled to tell them apart.

Checking a filed form. Carely can re-check any signed form against its own fingerprint and tell you whether it still says exactly what was signed.

Privacy. A filled-in form is health information. The answers are encrypted at rest, and never appear in the activity log — only the fact that a form was signed.

Authorizations: what a payer has approved

Operations Hub → Authorizations.

An authorization is a payer's approval to deliver a service: how many units, for which service code, between which dates. Care delivered outside one is usually not paid.

Recording one. Choose Record an authorization and enter what the payer approved. The client is picked from your roster rather than typed — an authorization recorded against a misspelled name can never match a visit, and would look perfectly correct on this screen forever. You can also bring existing authorizations across in Data Migration.

Modifiers. When the payer approved the service code with modifiers (for example T1021 TE), enter them in Modifiers, separated by spaces, up to four. Visits under that authorization are sent to the state's EVV aggregator with the same modifiers, so it can match them to the approval.

Units count down on their own as visits are delivered — at clock-out, on an imported EVV visit, and when an EVV exception is corrected. A visit consumes its units exactly once however many times it is retried, and deleting a visit gives them back.

The screen separates two things deliberately:

Booked but not yet delivered. Visits on the calendar are shown as scheduled units, and are not deducted from the balance. Only delivered care consumes units — otherwise a canceled shift that failed to release would quietly destroy units you still have. If more is booked than remains, Carely says so.

Delivered work on an invoice. Carely shows how much delivered work falls inside a period some invoice covers, and how much of that sits on an invoice that has been paid. It does not claim to match a visit to a bill — nothing links them — and where your invoices carry no service period it says it cannot tell rather than reporting nothing as billed.

So the full picture reads: approved → booked → delivered → on an invoice → paid.

Export puts all of it in a spreadsheet, with whatever filters you have applied.

Claims are not gated on an authorization. Carely warns you at scheduling time, which is when it can still change what happens; it does not stand between a built claim and your clearinghouse.

Warnings when you schedule

When a visit is booked, Carely checks it against the client's authorization and tells you if:

• there is no authorization for a client billed to a payer
• the date falls outside the approved dates
• it would take the client over the approved units, or the authorization is used up
• the service code does not match what was approved
• the authorization ends soon after this visit

It warns. It never blocks. The visit is created either way. An agency covering an emergency at 6pm has to be able to send somebody, and a hard stop there does not prevent unbillable care — it moves the scheduling outside Carely, where nothing is tracked at all.

You are asked why you are booking it anyway, and that reason is recorded permanently against the visit. Say something useful: it is what explains the decision months later.

Private-pay clients see none of this. A client who is not billed to a payer — no Medicaid ID and no authorization on file — produces no warnings at all, ever.

Removing an authorization

An authorization that visits have already been counted against cannot be deleted — those usage records are the evidence for a balance you have been scheduling against. Set its status to Cancelled instead, which keeps the record. One nothing has consumed can be removed.

Where invoicing lives

ALIASES: where are invoices, cant find invoices, invoices empty, no invoices yet, operations hub invoices, invoice screen, raise an invoice

Carely shows invoicing in two places, and they do different jobs.

If you are looking at Operations Hub → Invoices and there is nothing to do there, that is expected — select Open invoicing at the top right and Carely takes you straight to Revenue Center → Invoices.

A new agency sees four zeroes here, and that is correct. Approved visits do not become billable until the client's billing period closes, so there is nothing to bill on the day you start. Revenue Center → Invoices names what is waiting and the date it becomes ready.

The billing worksheet

Revenue Center → Invoices shows every client with approved, unbilled work, split into two groups:

• Ready to bill now — their billing period has closed, plus your send buffer.
• Upcoming — still inside their period.

What each row is made of. Each approved visit is priced by the server at the client's terms on the day it was worked — hours × the hourly rate, or the visit rate for a client billed per visit — plus its mileage when the client is billed mileage. A row says how many hours or visits it covers and its mileage, and the invoice generated from it lists every visit and every mileage charge as its own line, on the family's copy and on the printed invoice.

Correcting a visit after it has been invoiced bills the difference. If a visit was billed at four hours and is later corrected to five, the next invoice carries one hour — never five, and never nothing. Carely records how many hours of each visit have actually been billed, so putting a visit right months later still reaches the client, and the hours already charged are never charged twice. The worksheet row says includes 1 correction when a figure is a difference rather than a whole visit.

Corrected the other way, the family is owed a credit. A visit billed at five hours and corrected to three leaves $60 owed back at $30/hr. That credit nets against the same client's next invoice automatically. If there is nothing to net it against, Carely holds it and names it rather than raising an invoice for a negative amount — refund it, or leave it and it comes off their next bill.

A visit that recorded no time is held back, not billed. If a clock-in and clock-out land seconds apart — a double-tap, a phone in a pocket — the visit closes with zero hours. Carely keeps it off the worksheet and says so, because billing it would charge nothing and close the visit off for good: once a visit is invoiced it never returns to the worksheet, so correcting those hours afterwards would pay the caregiver and never charge the client. Fix the hours in Operations Hub → EVV Exceptions and the visit appears here to be billed properly. A genuinely short visit — fifteen minutes because the client was in hospital — is real time and bills as normal.

A visit is only billed once the period it falls in has closed. Work done today is not swept onto this month's invoice early; it waits for its own period. The invoice then states the period the work is actually in, and if an old visit was never billed, the range stretches back to cover it rather than filing it under a month it predates.

Send buffer (days after cycle close) is yours to set. It stops a client from being invoiced the instant their period ends, before late timesheets are approved.

TIPTrust the ready/upcoming split. It exists so nobody receives a half-month invoice, which is the fastest way to generate a billing dispute.

Mileage

Mileage can be billed to a client's family on the same invoice as the care. It is off until an administrator switches it on for a client: Operations → Clients & Matching → the client's Mileage card → Edit → Bill mileage to this client, and the date it starts. Switching it off works the same way, from a date; visits before the date are not affected. History lists every change.

The caregiver logs the trips at clock-out. While mileage is on for a client, the caregiver's clock-out panel asks about Driving during this visit. They choose No driving, or I drove — add trips and list each drive From one place To another: the client's home to the pharmacy, the pharmacy to the clinic, back home. Each stop is chosen as it is typed. The client's home is always offered first, and anything else comes up as suggestions written in full, such as CVS/pharmacy, 1201 Brickell Ave, Miami, FL 33131. Add another trip starts from where the last one ended, and Back to the client's home adds the drive home in one tap. The miles and what they pay show as the list is built.

How the miles are worked out. Each trip is the straight-line distance between its two places, measured by Carely from their positions on the map. The caregiver never types a distance for a place that was found. Two different places are never 0 miles apart, even in the same ZIP code.

When an address cannot be found — or address lookup is not available at that moment — the caregiver can use the address as typed and enter that trip's miles. The office sees those trips marked Miles typed by the caregiver — not measured when it approves.

Fixing trips afterward. A caregiver who could not log trips at clock-out, or got one wrong, opens My Timesheets → the visit's Mileage button → Add trips or Edit trips. That works while the timesheet is Pending or Denied. Once the office approves it, only the office can change the trips. Once the visit has been paid in a locked pay run, nobody can: the caregiver was paid those miles, and changing them would make the pay and the family's bill disagree.

Limits. One trip can be at most 200 miles, and a visit's trips at most 400 miles in all; anything more is refused with the reason, because it is almost always the wrong place picked. A visit with miles typed by the caregiver, or more than 100 miles of trips, is never approved automatically. It waits for a person to look at the trips.

The office reviews the trips when it approves. Operations Hub → Timesheets has a Mileage column for clients billed mileage. Each row says 5.0 mi · 2 trips, No driving, or Not logged. Select it to see every trip From → To with its miles, the total and what it pays. From there an administrator can Edit trips, Correct the total miles, or Approve timesheet. A correction replaces the trips' total for billing and pay, and Use the trips puts the trips' total back. The same correction is also under Invoices → the client's row → Mileage by visit.

The rate is the IRS standard mileage rate for business use, for each visit's date. The IRS sometimes changes it mid-year (it did on 1 July 2026, from 72.5¢ to 76¢), so two visits a day apart can carry different rates, correctly. Carely reads the IRS rate table at irs.gov every day and uses each new rate from the date the IRS sets; the Mileage panel shows the rate in use, the IRS release it came from, and when Carely last checked. Check now reads it immediately. If irs.gov cannot be read, the rate already on file is used and the panel says the check failed. A rate already on file is never changed by the check: if the IRS page ever shows a different figure for a date Carely already has, nothing changes and the panel says so.

On the invoice the family sees a mileage line for each visit (Mileage, 5.0 mi @ 76¢/mi (IRS standard rate)) with the number of trips. Where the caregiver drove stays with the agency.

The caregiver is paid the same miles back. For every visit whose mileage is billed, the caregiver is reimbursed the same miles at the same IRS rate. The reimbursement is not taxed and is not wages: it is kept out of gross pay, so it adds no withholding and does not count toward overtime. It appears on the pay run as Mileage (not taxed), on each caregiver's line, in My Pay, on the printed paystub, and as its own column in the payroll CSV, where Total to pay is gross plus mileage. Direct payouts include it.

Where addresses are looked up. Suggestions and map positions come from Amazon Location Service, inside the same Amazon Web Services account Carely runs in and under the same AWS Business Associate Addendum. Only the text being typed and the client's area are sent — never a name or anything about the visit. Addresses are not written to the audit trail.

Generating and sending invoices

Purpose. Turns approved, unbilled visits into invoices for the client's billing contact.

Who can do this. Administrator.

Before you begin. Approve the timesheets first. Confirm every client has a charge rate and a billing contact email.

Steps — the careful way (recommended)

1. Revenue Center → Invoices.
2. Check the Ready to bill now total and client count.
3. Select Generate [n] ready (review before send).
4. Open each generated invoice and check the hours and the total.
5. Select Send invoice on each.

Steps — the fast way (once you trust your setup)

1. Revenue Center → Invoices.
2. Select Bill & send [n] due now. This generates and emails in one action.

What happens next. Invoices are created, and sending emails them to each client's billing contact from your agency's own name, reply-to address and brand color, with only a small "Powered by Carely" at the bottom.

How to verify it worked. The clients drop out of Ready to bill now, the invoices appear in the invoice list, and the billed visits no longer appear as unbilled.

If nothing is in the send window you see: "Nothing is in its send window yet." That is not an error — no client's period has closed yet.

How the invoice list is organized

Revenue Center → Invoices works the same way as Timesheets. Along the top, the filters Unpaid, Paid, Void and All each show how many invoices they hold, and the search box finds a client, a family or an invoice number.

• Unpaid is where the list opens while anything is owed. It is grouped by when each invoice is due: Overdue, Due today, Due in the next 7 days, Due later and No due date, with the most overdue invoice at the top. Each heading totals what is still owed after insurance. An invoice you Mark paid stays where it was, marked Paid, until you next open the list.
• Paid, Void and All are grouped by the month the invoice was issued, newest first. This month and last month are open; older months are folded — select Show to open one. Months rather than billing periods, because each client is billed on their own cycle.

Families see their bills in Family Connect → Billing the same way: To pay first, grouped by when each is due, then Paid and All by month. A bill they pay stays on screen, marked Paid.

IMPORTANTBilled work is tied to its invoice on the server, and Carely refuses to bill it twice. You can run the billing screen without fear of double-charging a client.

What a family actually receives when you send an invoice

The email is a knock on the door, not the invoice. When you select Send invoice or Send statement, the payer receives a short message saying an invoice is ready, with a link to sign in and read it. It carries no amount, no dates and no client name.

It comes from your agency's name and brand color, with a small "Powered by Carely" at the bottom. The mailbox does not take replies, so give families your own contact details inside the message they read in the portal.

Why. Naming a person and their care in an email would put health information in an inbox and in a mail provider's store. Carely keeps it in the portal instead, which is what lets your agency send through a mail provider without a business associate agreement.

The message you compose is what they read once signed in. The subject and message in the send dialog go to the payer's portal, not into the email body. Write it for them — it just isn't what travels over email.

How to verify it worked. The invoice moves out of Waiting to send and shows as sent. If the mail provider refused it, Carely tells you why on the spot and leaves the invoice unsent, so it stays in your list and you can try again after fixing the cause. An invoice is only ever marked sent when a provider accepted it.

If it says there is no billing contact email, the message names the client. Open that client's record, add a billing contact email, and send again.

Invoice numbers

What it is. Every invoice carries a plain number — 1001, 1002, 1003 — that appears on the emailed invoice, on the printed copy, in the family's portal and in the invoice list. It is the number to quote on the phone, write on a check stub, and match against a bank deposit.

Where the numbering starts. Each agency has its own series beginning at 1001, counting up from there. Your numbers are yours: another agency on Carely reaching invoice 1042 has no effect on yours, and two agencies having a 1042 each is normal and correct.

One number, one invoice. Carely will not let the same agency issue two invoices under one number, so a number always identifies exactly one invoice for your books.

Invoices raised before this existed were given numbers automatically, in the order they were issued, and keep those numbers permanently. Nothing you have already sent has changed.

Where to find it. Revenue Center → Invoices, on each row. Families see it at the top of their invoice in Family Connect.

Verifying a billing run before you send

Do these four checks and most billing disputes never happen:

1. Approve everything first. Operations Hub → Timesheets should have no Pending rows for the period.
2. Look for zero or near-zero amounts. A client with hours but no money has no charge rate. → Rates
3. Compare the total against last period. A large swing is usually a missed approval or a duplicated visit, not a real change.
4. Check the billing contact email on any new client. An invoice sent nowhere is worse than an invoice not sent, because you will think it went.

Why a visit is not on an invoice

Work down this list:

1. Its timesheet is not Approved. By far the most common cause.
2. The visit was never clocked out, so no timesheet exists at all.
3. It is already on another invoice.
4. The client's period has not closed yet — it is in Upcoming, not late.
5. The client has no charge rate, so it appears with a zero amount.

Payments

Revenue Center → Payments records what has been collected, grouped by the month each payment came in, newest first, with older months folded. Invoices update in real time as payments arrive, including family payments made through the portal.

Administrators can also collect from a family by ACH, send a payout to a caregiver, and send money back to a family — from Operations Hub → Payments.

Taking payments in Carely (Stripe)

Operations Hub → Payments offers two ways to work. Track only is a ledger: you pay caregivers and bill families through your own payroll and bank, and record it here for your books. In-app payments moves real money through Stripe, and three things have to be true before it does — the screen says which are not:

1. Your agency is connected. Your payout account (Stripe) → Connect payments takes you to Stripe's page to verify the agency's identity and bank; family payments land there and caregiver payouts come from it. The banner at the top reads Live only once Stripe reports the account can take charges and make payouts. Carely never sees your account numbers.

> NOTE — Stripe's page is headed Sign in to Express, which reads as though you > need a Stripe account already. You do not. It is one entry point for both: enter your > email address and, if Stripe does not recognize it, the same flow sets an account up — > phone, business details, then your bank. A caregiver connecting their own payout > account (point 3 below) meets the same screen and the same answer applies. (The one exception is the agency the business that runs Carely operates itself — its Stripe account is the platform's, so there is nothing to connect; Carely staff mark it as such: Platform Console → Agencies → open the agency → Stripe account for payments → tick Platform's own Stripe and confirm. The same card says whether any agency has connected its own account yet.)

1. The family has a bank on file. Collecting by ACH debits a bank account the family has saved on Stripe's secure form — either you, on the client's card → Payment on file → Add bank account, or the family themselves in Family Connect → Billing. Carely keeps the bank's name and last four digits, nothing more. The Collect from a family list says beside each family whether a bank is on file.
2. The caregiver has connected a payout account. Each caregiver connects their own from My Pay → Direct deposit → Connect payout account (an administrator can start it for them from their profile). The Pay a caregiver list says who has.

Carely refuses rather than pretends. If any of those is missing, the payment is refused with the reason on the card — nothing is written to the ledger as if it had happened. A Reimburse is a refund of a collection Carely made: pick the family from those you have collected from, up to what was collected and only once the ACH has settled (a few business days). Money you took outside Carely is paid back outside Carely.

What the ledger shows. Every collection, payout and reimbursement with Stripe's status: Processing while an ACH debit or transfer is in flight, Completed when Stripe confirms it, Failed with the bank's reason if it bounced or was reversed. Carely checks with Stripe every fifteen minutes, and the moment it happens when Stripe's webhook for connected accounts is set up (Platform Console → Diagnostics says whether it is).

Paid by check or cash? Mark it paid. Every unpaid invoice in Revenue Center → Invoices carries a Mark paid button. Use it when the money arrived outside Carely — a check in the post, cash, or a bank transfer. It stops the reminders, counts the invoice toward collections, and is recorded against your name in the audit trail. Families who pay through the portal are marked automatically; you never need to do it twice.

NOTEIn track-only mode, and before the platform has Stripe switched on, entries are recorded as simulated and no money moves. Once in-app payments are live, a payment that cannot be made is refused, never recorded as simulated. No bank or card numbers are ever stored in Carely — Stripe handles them.

Collecting what an invoice is owed

ALIASES: collect payment, charge the family, debit the bank, autopay, auto pay, collect balance, take payment, charge an invoice

Mark paid and Collect are different things, and the difference is whether money moves.

Collecting needs a bank on file. The responsible party saves theirs once — either you, from the client's card → Payment on file, or the family themselves from Family Connect → Billing. Without one, collecting is refused and says so; the invoice stays unpaid rather than quietly reading as settled.

A family's Pay button collects. In Family Connect → Billing, Pay debits the bank the family saved, the same way Collect does — the bill reads Paid only once the debit is accepted, and a family is told if their bank is missing or the debit was declined. Only the office can Mark paid. If your agency does not take payments through Carely, Pay tells the family to pay you directly.

A declined debit is not a payment. If the bank or Stripe declines a debit, the invoice stays open and the office is told why — whether it was AutoPay, Collect or a family's Pay.

A part payment stays a part payment. Collect $80 against a $200 invoice and $120 is still owed — the invoice does not round up to paid.

AutoPay charges the enrolled balances on the day they fall due, every morning. A family turns it on for their own loved one from Family Connect, or the office turns it on from the client's card; either way a verified bank must already be on file, and the choice is kept on the client record so the office and the family see the same thing. Each morning Carely debits every enrolled balance whose due date has arrived, the same way Collect does, once per invoice per day. Run AutoPay on the Revenue Center does the same run now rather than waiting for the morning. Any balance it could not charge is named in the result and raised once in the bell — a family with no bank on file, or a debit the bank refused — and the invoice stays open; it is tried again the next morning once fixed. Turning AutoPay off takes those reminders down. The Revenue Center's billing policy has Enroll clients in AutoPay once their bank is saved: with it on, the moment a family saves a bank the client is enrolled, and the family can turn it off from Family Connect.

After an insurance payment, the rest is collectable the same way. When a payer settles part of a claim, what they did not cover stays on the invoice as the family's balance, and Collect takes it from the responsible party.

NOTEACH takes a few business days to settle. The invoice reads as paid once the debit is accepted; if it later fails, the payment on the ledger shows why.

Invoice aging

Revenue Center → Reports → Invoice Aging buckets what you are owed: Current, 31–60 days, 61–90 days, 90 days and over. Work the oldest bucket first — collectability falls off sharply after 90 days.

What comes off a paycheck

Carely works out withholding properly rather than applying a rough percentage.

For each pay period it annualises the pay, applies the worker's W-4, subtracts the standard deduction and any personal exemption, runs the result through that year's tax brackets, takes off any credits, and divides back down. That is the same method the IRS and the states publish — it is what a payroll bureau does.

It covers all fifty states and the District of Columbia: the forty-two that tax wages have their full 2026 brackets, standard deductions and exemptions on file, and the nine that do not tax wages withhold nothing.

Why this matters: the previous version applied each state's highest tax rate to every dollar earned. For a caregiver in California on about $37,000 a year that withheld 6.6% where the correct figure is nearer 1.3% — roughly five times too much, out of the pay of somebody who can least afford to lend money to the state until their refund.

The W-4

Every worker's Form W-4 can be recorded against them: filing status, the multiple-jobs checkbox, dependent credits, other income, extra deductions and any extra withholding they have asked for. Carely uses all of it.

Where no W-4 has been recorded, Carely withholds as single with no adjustments — which is exactly what the IRS requires an employer to do in that situation. It is not a guess.

Local city and county tax

Fifteen states let a city or county charge income tax on top of the state's. Local tax follows the caregiver, not the agency — two people on your roster owe different amounts if they live in different counties. So each caregiver has a tax locality on their record, and Carely offers the real list for your state rather than a free-text box.

Carely calculates local tax in full for:

If you have a Maryland caregiver whose county you have not recorded, Carely withholds at the maximum county rate. That is what Maryland's own payroll bureau does when it has no address, and it is the safer error: too much comes back as a refund, too little becomes a bill. For Indiana, Carely tells you the county is missing rather than withholding nothing.

Pennsylvania, Ohio and the rest: import your state's register

The remaining states are not a fixed list Carely can carry. Pennsylvania has around 2,500 municipal and school-district Earned Income Tax rates keyed by PSD code; Ohio has around 600 municipalities plus 200 school districts collected by RITA and CCA rather than published as one file. Both change every year, some mid-year.

So Carely imports them instead. Upload your state's official register and it takes effect immediately — no waiting for a Carely update. The states this covers are Pennsylvania, Ohio, Kentucky, Michigan, Missouri, Alabama, Delaware, Iowa, New Jersey, Oregon, West Virginia and Colorado.

Carely checks a register before it uses it. A rate above 10% is refused outright — no US local income tax comes close, so that is a decimal point in the wrong place or a bad export. Duplicate codes are refused, because a duplicate quietly picks a winner. An import with no effective date is refused, because these rates change annually. And if the file is far shorter than the state is known to have, Carely says so — a truncated download makes every missing place look like it charges nothing.

Upload the CSV or spreadsheet your tax collector sends — Carely reads the column names loosely, because every collector labels them differently, and it understands rates written either way ("1.5%", "1.5" and "0.015" all mean the same thing). There is a blank template if you would rather start from one.

You can validate a file first and see exactly what Carely thinks of it without importing anything.

Once imported, Carely applies each state's own rule. Pennsylvania withholds the higher of the caregiver's total resident rate or the non-resident rate where they work — so record both their home and work localities if they commute. Ohio stacks: a municipal tax where they work plus a school-district tax where they live.

Until a register is loaded, Carely tells you your state has local taxes it is not withholding, rather than withholding zero and leaving you to find out.

Puerto Rico and the U.S. Virgin Islands run their own tax systems. Carely marks them unsupported rather than treating them as zero.

Carely prepares payroll; it does not file it. Give these figures to your payroll provider or accountant, who is responsible for the filing.

Expenses

Revenue Center → Expenses. What your agency actually spends, in one place.

Carely builds the total from three parts, and it is worth knowing which is which because only one of them is yours to type in.

Pick a period at the top — this month, last month, the last 30 days, this quarter, or the year so far. Every figure on the screen covers exactly that window.

Recording a cost

Add a cost, choose a category, enter who you paid and how much, and set the date.

If it repeats, set Repeats to monthly, quarterly or yearly. You record it once, and it counts in every period it falls into. Change the amount later and the whole series changes with it — so when your rent goes up, you edit one line, not twelve. Set an end date if the cost stops.

Removing a repeating cost removes every occurrence of it.

Your state unemployment rate

Carely sets this for you when you tell it which state you operate in. Every state publishes a rate for employers who have no claims history yet, and a taxable wage base — the slice of each person's annual pay the tax applies to. Carely holds both for all fifty states, the District of Columbia, Puerto Rico and the U.S. Virgin Islands, and applies your state's figures the moment it knows where you are.

Once your state assigns you your own rate, enter it. After a couple of years a state stops using the starting rate and gives your agency a rate of its own, based on your own claims history. It arrives on a notice each year and it will not match the starting figure. Enter it here and Carely will use it — and will never overwrite it. That distinction is deliberate: a rate your state assigned you is yours, and nothing automatic is allowed to change it.

Seven states — Louisiana, Minnesota, Montana, New Mexico, Utah, Washington and Wyoming — do not publish a starting rate at all. They give each new employer the average for their industry. In those states Carely records the wage base, leaves the rate empty, and tells you it is missing rather than inventing one.

Carely watches for changes. States revise wage bases and rate schedules every 1 January. The State Tax Agent checks yours daily against what your state publishes, and when it has moved it appears in Operations Hub → Copilot as an approval waiting for you. The approval names both figures — what you are using and what the state now says — so you can judge it without going and looking.

It never changes a tax figure on its own. Not at any autonomy level, including full. A wrong rate does not announce itself; it quietly mis-costs every payroll until somebody notices, so a person decides. Approving it applies the change and records who approved it; rejecting it leaves everything alone, and you will be asked again on the next check.

One approver is enough — many agencies have a single administrator, and a rule you could not satisfy would leave you stuck on the wrong figure. And a rate you recorded from your own state notice is never overwritten, approved or not.

What the total does not include

At the bottom of the screen Carely lists what is missing from the number above it — an unrecorded SUTA rate, a client with no pay rate on file, a period with no overhead entered, and travel time between clients, which Carely does not record. A profit figure is only as good as what went into it, so Carely tells you what went into it.

Profit

Revenue billed is what you invoiced in the period. Operating profit is that minus everything above. The margin percentage next to it is the one number most agencies want and few can produce quickly.

These figures keep themselves up to date. Record a cost, raise an invoice or approve a timesheet and the next time you look at Profit & Tax the numbers already include it — there is nothing to refresh.

When the numbers are not right

Carely works from what it can see, and it cannot see everything: cash work that never went through an invoice, an owner's draw, a cost paid from a personal account, or a year you part-migrated from another system.

So on Profit & Tax you can enter your own revenue and expense figures for a tax year. Carely uses them for the tax calculation, and goes on showing what it worked out beside them — so you can see the gap, and notice if your correction stops matching reality later. Add a note saying why, and your accountant will thank you.

Your figures are saved against that year only and stay until you press Use Carely's calculation. Nothing is overwritten: reverting brings the calculated figures straight back.

These figures are for running your business, not for filing. Give your accountant the export, not the screen.

Payroll

Operations Hub → Payroll.

IMPORTANTCarely prepares payroll. It does not run it. No tax filing, no W-2s, no payments to anyone. You get verified hours, gross pay and payroll-ready exports; your payroll provider does the rest.

Your pay period

At the top of the Payroll screen, choose Weekly or Biweekly. That one choice gives Carely four dates it did not have before, and everything else here follows from them:

Carely does the chasing. Once a period is set, reminders go out on their own:

• To a caregiver — only if they are actually missing something. 48 hours and again 12 hours before submissions close, and only about their gaps: a visit they completed with no timesheet, or one they clocked into and never clocked out of. A caregiver with nothing outstanding hears nothing, which is what keeps the reminder worth reading. These go to that caregiver alone — they do not fill your notification bell.
• To you — one message per period, not one per timesheet. When submissions close you get a single message with the count and the deadline ("14 timesheets await approval, due Tuesday"), a second if any are still outstanding twelve hours before the deadline, and a separate one naming any caregivers whose visits were never recorded — because those hours cannot be approved or paid until somebody records them.

Reminders never repeat. Each is sent once per pay period, however often the check runs.

The pay run

Below the pay period is the pay run — the thing you actually pay from.

Payroll used to answer "what are we paying?" by adding up approved timesheets every time you asked, so the number moved as people approved things, and a payment could never say which hours it covered. A run fixes that: it is a snapshot, and it goes through three states.

Locking is what stops a week being paid twice. Once a run claims a timesheet, no other run can ever include it — not by mistake, not by rebuilding, not by someone repeating the export. If a run somehow tries, the whole lock is refused rather than half-applied.

A locked run is never edited. If something was wrong, correct it as an adjustment against the next run — that is what keeps the history trustworthy.

How the amounts are worked out

Every figure is counted in whole cents. Carely converts the pay rate to cents, multiplies by the hours, and rounds once — so 6.5 hours at $41.15 is $267.48, which is what the arithmetic gives on paper.

That matters because the obvious alternative — multiplying the dollars and rounding at the end — quietly loses a cent on some lines, and can leave an invoice whose line items do not add up to its own total. If you check a Carely figure with a calculator, it will match.

Overtime

Carely works out overtime, and the way it does so is worth understanding — this is the part payroll spreadsheets most often get wrong.

Overtime is owed on the person's week, not the client's. A caregiver with 25 hours for one family and 20 for another worked 45 hours and is owed overtime, even though neither family alone passed 40. Carely totals the caregiver.

The premium is paid on the weighted-average rate, not the rate of whichever visit ran late. If someone worked 30 hours at $20 and 15 at $30, their regular rate that week is $1,050 ÷ 45 = $23.33, and the overtime premium is based on that. This is what federal law requires (29 CFR 778.115), and it is a different number from 1.5× either family's rate.

The week is fixed, Monday to Sunday. A two-week pay period is split into its two weeks before the 40-hour line is applied — otherwise the threshold lands on 80 hours and almost nothing counts.

Your state's rules are applied on top. In California a 14-hour day is 8 hours straight, 4 at time-and-a-half and 2 at double time — even in a week under 40 hours. Daily and weekly overtime are never both charged for the same hour.

Getting the run to people

Once a run is locked, there are two ways it reaches caregivers. Which one you use is a decision about tax, not about software.

Export to your payroll provider — the default, and the right answer for most agencies. The run becomes a file your provider ingests (ADP, Gusto, Paychex, QuickBooks), with straight and overtime hours in separate columns so they can apply the right tax treatment. Carely keeps the pay period, the approvals, the overtime and the audit trail; your provider does withholding, W-2s and filings.

Pay caregivers directly from Carely — opt-in, and for 1099 contractors only.

Read this before turning it on. Most home-care caregivers are W-2 employees — the 2015 federal Home Care Rule pushed hard against classifying them as contractors. Paying a W-2 employee by direct transfer skips income-tax withholding, FICA and W-2 filing, and that exposure is yours, in every state you operate in. Carely cannot check a classification, so it asks you to confirm it by name, with a timestamp, and records that you did. If these people are employees, use the export.

Paying twice is not possible. Each caregiver's payout in a run carries a fixed key. Ask twice — a timeout, a double-click, a container restarting mid-run — and the second attempt is refused by Carely, and refused again by Stripe if it ever got that far. A run that is fully paid declines further payment outright.

A payout that fails is not stranded. If one transfer bounces, you can run the payment again: the person who failed is retried, and the people already paid are left alone. Payouts that are still in flight are reconciled to completed or failed — a transfer that was reversed is recorded as failed, never as paid, because money that came back is not money someone received.

Setting your own overtime rate

Most agencies pay the statutory time-and-a-half. If you pay more — a weekend premium, a hard-to-fill client — set it under Overtime on the Payroll screen.

You can change four things:

The law is a floor, not a rate. You can always pay more than the minimum, or start overtime sooner. You cannot pay less, and you cannot start it later. Carely refuses those outright rather than saving them — so you can never be in the position of believing you set 1.25× when the law requires 1.5×. If your state mandates double time, the same applies to it.

Under the fields Carely shows a worked example — what a real mixed-rate week comes to under your settings — so you can see what you have chosen rather than trusting a multiplier you cannot check. Back to standard clears everything to the statutory rule.

Two things Carely cannot know, shown on the line rather than left out: travel time between clients is compensable and Carely does not record it, so a week containing travel is understated; and live-in and personal-attendant exemptions are not applied, so where one genuinely applies the figure may be higher than what is owed. These are computed figures for you and your payroll provider to check — not a filing.

Kentucky's seventh day. In Kentucky, someone who works all seven days of the week (Monday to Sunday in Carely) is paid overtime on every hour of the seventh day, even when that is more than their hours over 40. It counts toward weekly overtime rather than being added to it. Carely applies it automatically for an agency whose state is Kentucky, and the pay run names the week. The rule does not apply to someone who was not allowed to work more than 40 hours that week; Carely cannot know that, so it says so on the line.

Approve clean timesheets automatically

Off unless you turn it on, at the bottom of the pay run.

Most timesheets are unremarkable: the caregiver clocked in and out at the client's address, and worked what was scheduled. Clicking approve on each one is usually why approvals run late — and late approvals are what make people paid late. Switch this on and those clear themselves, leaving you only the ones worth your attention.

A timesheet approves itself only if all of these hold:

• it came from a clock-in, not typed in by hand
• both punches verified at the client's address
• the time worked is within your tolerance of what was scheduled (15 minutes by default)
• the client has a pay rate on file
• the caregiver is on your roster

Anything else waits for you and tells you why — "entered by hand", "42 min over the scheduled time". You are approving by exception rather than by item.

Steps

1. Set Pay period from and to. Do this first, every time — only visits dated in this period are included in the totals and both exports.
2. Review the figures: Verified Pay (Approved), Approved Hours, Pending Approval and Avg Effective Rate.
3. If Pending Approval is not zero, go and approve those timesheets before you export.
4. Export:
5. Summary CSV — totals per caregiver.
6. Per-visit CSV — every approved visit, for Gusto, ADP or QuickBooks.

How to verify it worked. The exported row count matches the approved timesheet count for the period, and the totals match what the screen showed.

WARNINGWithholding shown on a Carely paystub is an estimate. The authoritative paystub is the one your payroll provider issues. Do not give a Carely estimate to an employee as their tax record.

Medicaid billing

Operations Hub → Medicaid is where you connect your own vendor accounts with your own credentials, file claims, and watch what comes back. See External EVV and state systems for the visit side.

Carely does not hold any certification on your behalf — your agency holds all of it. Carely is the transmission vehicle.

Two things happen on this screen.

• EVV visits transmit on their own. When a caregiver clocks out, that visit is queued for your state aggregator automatically. The screen shows the queue, what succeeded, what failed, and a Retry button for anything that did not go through.
• Claims are filed from the screen and answered on it. Connect a clearinghouse, tell Carely who your agency is to the payer, and file a claim for a client's approved visits. The payer's answer — accepted, rejected, denied or paid, and the remittance that paid it — is pulled back from the clearinghouse and shown against the claim.

Connecting your clearinghouse

Under Clearinghouse (claims), click Connect and pick your vendor. The form says where its credentials live, asks up front for anything your vendor issues per customer, and tests the connection the moment you save — a refusal comes back in the vendor's own words, with what to change. What each one needs:

The host and folders for each of these are filled in from the vendor's own published connection sheet, so nothing but the username and password is asked for; both can still be changed under Advanced Setup if your packet says otherwise.

Each claim file is addressed the way that clearinghouse's companion guide asks — the interchange sender and receiver codes Availity, Office Ally, Waystar, TriZetto and Claim.MD each publish — and carries its own control number, so no two files are ever mistaken for a duplicate. Claim.MD's guide puts your Claim.MD account number in the file; enter it under Advanced Setup (your tax ID stands in until you do). Waystar's puts your client account ID there and Availity's your customer ID; both are under Advanced Setup too.

Every answer the clearinghouse sends is read, whatever its format: acknowledgements (999, 997 and TA1), claim statuses (277CA) and remittances (835). A file the clearinghouse throws back as a whole marks each claim in it rejected, with the reason.

Two clearinghouses also send reports of their own, and Carely reads those claim by claim: Availity's IBR (its own edits, minutes after upload), EBR (the payer's front-end verdict) and DPR (answers a payer sends later), and Office Ally's File Summary and EDI Status reports. A rejection in any of them marks the claim rejected with the reason in the vendor's words — the error, and for Availity the loop and segment it points at — and the bell says so once. A clearinghouse's own acceptance never overwrites a later answer from the payer. A report Carely cannot read — a zipped or encrypted file, or a format neither vendor publishes — is named in the bell with where to open it, never skipped in silence. A check that could not finish says why on the Remittances card.

The test checks two separate things, because passing the first does not mean the second is true: that your credentials are accepted, and that your account is actually allowed to file claims. Optum in particular will issue a working Client ID whose health check answers normally while claim submission is turned off for it — most often on a sandbox account. Carely refuses that connection rather than showing it as connected, and tells you to ask the vendor to enable professional claim submission for that Client ID. A connection that cannot file is never marked ready for live claims, so you find out now instead of at your first claim.

Claim.MD is the clearinghouse Carely recommends: it takes claims, reports each claim's status, and hands back 835 remittances, all through one Account Key. The file-based clearinghouses close the same loop: every quarter hour, and on Check now, Carely reads their response folder — a 277CA sets each claim accepted or rejected with the payer's words, a 999 that throws a whole file back raises an alert naming the fault, and every 835 is downloaded and reconciled against your claims. Optum is the one connection whose answers arrive in its own portal.

1. Open an account at claim.md. The account owner signs up under Get Started (ID verification and a card are required; ask their sales team for a test account for certification before paying); your production account is the one you enroll with the payers.
2. Find your Account Key. In Claim.MD: Settings → Account Settings → API Key (Account Key) → Generate. Copy it — Carely never shows it again after you paste it.
3. On the Medicaid screen, under Clearinghouse (claims), click Connect, choose Claim.MD — recommended, and paste the key. The screen repeats where the key lives, under Where to get it.
4. Choose Sandbox for a test account key, Production for your live one. Check the acknowledgment and click Connect (encrypted). Carely tests the key at once: a wrong key is refused with Claim.MD's own words.
5. In Sandbox, use Send test claim to certify. When Claim.MD is satisfied, click Replace / go live, paste your production key, choose Production and connect again. The card reads LIVE once the production key has verified.

IMPORTANTNothing real is sent while the connection is in Sandbox or unverified. Real claims file only from a LIVE connection.

The billing provider

Before a claim can be filed, Carely needs to know who your agency is to the payer: the NPI and tax ID (EIN) on your Medicaid enrollment, and the address the payer has on file. Enter them once in the Billing provider card under File a claim; they go on every claim. The state starts as your licensed state from agency setup and can be changed.

The card also asks for the Medicaid payer ID: your clearinghouse's own code for the Medicaid plan you bill, from that clearinghouse's payer list (each one publishes it). The clearinghouse routes every claim by this code, so a claim cannot be filed without it.

Each claim also needs the client's home address, date of birth and Medicaid member ID from the client record, and a diagnosis code. Type the code the way it is written (Z74.1) — Carely sends it the way claims carry it (Z741). Anything missing is named when you prepare the claim, before anything is sent.

The state and city are picked from lists, not typed, so the address on the claim is spelled exactly one way — "Miami", not "miami" or "MIAMI" — and matches what the payer holds. The city list follows the state you pick; a town the list does not carry goes under Other city… and is capitalized for you. A ZIP that sits in a different state from the one picked is refused, naming both, so the wrong box can be corrected.

A mistyped NPI is refused when you save it — it has to be ten digits and pass the check digit — because a wrong NPI rejects every claim in the batch a day later. Look yours up at npiregistry.cms.hhs.gov if the screen refuses it.

Filing a claim

1. Under File a claim, choose the client and the service period, and click Fill in from approved visits.
2. Carely fills the claim in from what it already knows: the client's Medicaid member ID, legal name, date of birth and payer from their record (Clients → the client → Billing identity), your billing provider, and one line per approved timesheet in the period. Each line is HCPCS T1019 (personal care) in 15-minute units — four hours is 16 units — charged at the client's charge rate.
3. Review the lines. Change the code, add the modifiers your state expects (for example U1), or adjust units and charge per line; remove a line with ✕. Your state's fee schedule is the authority here, not Carely's default.
4. Add the diagnosis code and confirm the place of service (Home, 12, unless the client lives elsewhere), then click File claim.

If something is missing, the screen says so before you can file — a client with no member ID or no date of birth on record, no charge rate, or no approved visits in the period. Fix the record and fill in again.

A visit already on a claim that is in flight or paid is left out of the next one, so the same visit cannot be billed twice. A rejected, denied or failed claim releases its visits so they can be corrected and refiled.

What comes back

Carely checks the clearinghouse every fifteen minutes, and Check now on either the Claim submissions or Remittances (835) card asks immediately. Each claim shows its latest answer:

• Accepted — the clearinghouse and payer took it; payment is pending.
• Rejected — turned away at the door, with the reason in the clearinghouse's words (usually a member ID or NPI the payer does not recognize). Fix and Retry.
• Denied — accepted, then paid nothing, with the adjustment reason (for example CO-197, not covered). This is the payer's decision; take it up with them.
• Paid — the remittance arrived and matched, with the amount.

Remittances (835) lists every payer payment the clearinghouse handed back: payer, check or EFT number, date, amount, and how many claims it matched. A remittance that names a claim Carely does not hold raises an alert in the bell rather than being dropped. The 835 itself is kept, encrypted, for the record.

Connecting your state EVV aggregator

Under State EVV aggregator (visits), click Connect and pick the aggregator your state uses. Your state's onboarding packet (the "Alt-EVV" or third-party EVV welcome letter) carries everything the form asks for:

• Sandata — API username and password, your Sandata Account ID (Ohio calls it the Business Entity ID) and your state Medicaid provider ID, all from the packet. Under Advanced Setup pick the interface your state runs — most run Alt-EVV v1.1, Ohio runs its own — and enter the Payer ID and program from the packet, a default procedure code if some clients have no authorization, and the state's reason code for a manually entered call. Choose Sandbox to certify on Sandata's UAT host, then Production. Sandata needs each caregiver and each client on file before it accepts their visits, so Carely sends the caregiver record (keyed by SSN, or the state-issued worker ID if you chose that) and the client record (keyed by Medicaid member ID) first, then the visit with both clock calls and their GPS. A time the office entered goes as an adjusted time with the reason code. Sandata answers each batch a little later; Carely asks for the verdict on Check now and every 15 minutes, and shows accepted or rejected with Sandata's own words.
• HHAeXchange — open a ticket with the HHAeXchange Provider EDI Integrations portal titled "Test Environment API Credentials Request", naming Carely as your EVV vendor. The Client ID and Client Secret arrive by secure email with your state's EVV Data Aggregator Specification, which carries the Payer ID HHAeXchange assigned; the form also asks for your agency's nine-digit tax id. Choose Sandbox to certify in HHAeXchange's Implementation environment, then Production once they issue production credentials. Each visit goes to HHAeXchange's Aggregator API keyed on the client's Medicaid member ID and the caregiver's worker ID, with both clock calls, their GPS and the service code; the aggregator's verdict on each visit is read back and shown on the transmissions card. HHAeXchange will not accept a caregiver without a date of birth, gender, hire date and the state license or registry number, so fill those in on each caregiver (Caregivers → the caregiver → State EVV worker details). A visit for a caregiver missing any of them is held back with a message naming what to add.
• Netsmart (Tellus / Mobile Caregiver+) — the aggregator for Florida's Medicaid managed care plans (Aetna, Anthem/Simply, UnitedHealthcare), Georgia DCH, Montana DPHHS, Nebraska (DHHS, Molina, Nebraska Total Care, UnitedHealthcare of the Midlands) and Virginia's Anthem HealthKeepers Plus. There is nothing to paste: Carely is the certified Alternate EVV Vendor and holds the Netsmart account, so you pick your state and payer from the list and press Connect. Your agency's NPI and tax id come from Billing provider on this screen; Georgia, Montana and Nebraska also need your agency Medicaid provider id under Advanced Setup, and Florida, Georgia, Montana and Virginia need a diagnosis code on each client's clinical profile or a default under Advanced Setup. Until Carely's certification with Netsmart is complete for an environment, the test reports certification pending there and nothing is transmitted — your settings are kept. Each completed visit goes out as a Rendered Services file in Netsmart's own layout — member by Medicaid ID with legal name, birth date and service address, both punches GPS-verified with coordinates, UTC times, the time-zone code and the reason code for an office-entered punch. Netsmart answers in its response folder within the quarter hour; Carely reads it on Check now and every 15 minutes, so a visit shows sent until its verdict is in, and a rejection carries Netsmart's own error line. Before the first file: Netsmart's UAT confirms two settings Carely cannot know in advance, the delivery system code for a managed-care payer and the exact layout of the response file — both are settings, not code.
• CareBridge — the aggregator for Arkansas, Iowa, New Jersey, North Carolina, Tennessee and Wyoming. There is nothing to paste: Carely is the certified Third-Party EVV Vendor and holds the CareBridge SFTP account (credentialed by Carely's SSH key, from Carely's own server address). Pick your state and health plan and press Connect; your agency's NPI and tax id come from Billing provider, and every CareBridge state except North Carolina and Tennessee also needs your agency Medicaid provider id under Advanced Setup. Arkansas also needs each caregiver's own Medicaid provider ID (on the caregiver's profile) and, for home health visits (T1021, S9131), the attending provider's name under Advanced Setup; the attending provider's NPI is your agency NPI unless you enter another. The member attestation says complete only when someone other than the caregiver signed at clock-out. Each completed visit goes out in that state's own column layout — member by Medicaid ID, both punches with GPS and the service address, UTC timestamps, the service code from the client's authorization (or the default you set), the authorization number, the member attestation, and the late or missed reason codes CareBridge asks for when a check-in was late. CareBridge answers in its response folder; Carely reads it on Check now and every 15 minutes, so a visit shows sent until the verdict is in, and a rejection carries CareBridge's own pre-billing error code and words. Claims through CareBridge (the ClaimAction column) stay off until an agency chooses it under Advanced Setup.

Deleting a visit that was already sent cancels it at CareBridge too: Carely sends the same visit marked canceled, from its encrypted copy of what was sent, and the visit shows canceled in EVV transmissions once CareBridge accepts it. A visit deleted before it was sent shows withdrawn; nothing reached CareBridge. Void claim, beside a sent visit, asks CareBridge to void the claim it generated for that visit. It appears only when your agency bills through CareBridge, asks you to confirm, and cannot be undone; the claim shows Void sent until CareBridge's daily report says it is voided.

CareBridge also sends two daily reports to the same folder, and Carely reads both. The Appointment Errors Report lists every pre-billing error still open on any visit, including ones CareBridge found after accepting the file, such as eligibility or a refusal from the plan's own aggregator; a visit listed there turns rejected with each code, and the bell tells the office once. The Appointment Status Report gives CareBridge's status for the visit and, when CareBridge bills, the claim: queued, submitted, accepted by the plan, paid, denied and so on, with the payer claim number and the billed units and amount. These show under the visit in EVV transmissions. A denied, rejected or failed claim raises an alert with the plan's own reason. When the report says a visit has no errors left, its rejection clears by itself, and a visit CareBridge holds only for a pre-billing rejection that has since cleared is re-sent once, unchanged, as CareBridge's specification directs. A rejected visit has a Re-send button once the record is corrected. The reports are on by default in North Carolina; elsewhere Carely asks CareBridge to switch them on.

• AuthentiCare, Therap — these aggregators issue their third-party interface to an EVV vendor only under an integration agreement, and Carely has not completed certification with them yet. They are listed so you can save your credentials and the endpoint from your packet (encrypted); the test reports certification pending, and no visit is transmitted until Carely is certified. Contact Carely support to start it.

A visit the aggregator rejects — a member ID the state does not know, a caregiver with no identifier on file — comes back with the state's reason on the EVV transmissions card and in the bell, so it can be corrected and re-sent.

Before you send anything real, send one test. Each connection has a Send test visit / Send test claim button. It works only while the connection is set to Sandbox — on Production a test record would land in your live account as a real one, so the button says so instead. That is the way to find out that your credentials, your endpoint, and your payload are right before a real claim or a real visit is on the line.

Visits completed before the aggregator is live wait for it. While the connection is in Sandbox or has not passed its test, a completed visit shows waiting with the reason and the bell says so once. The moment the connection passes a test in Production, every waiting visit is sent. A connection that fails a test in Production is tested again every hour, so one bad moment at the vendor does not stop your visits until somebody notices.

12. Credentials and documents

A credential is a certificate or license with an expiry date. Carely tracks them, warns you before they lapse, and — critically — stops a caregiver from being scheduled when a required one is missing or expired, unless you have turned that blocking off.

Everything lives in Operations Hub → Certifications (Care & Compliance). Caregivers see My Certifications.

Adding a credential

Purpose. Puts a certificate on file with its expiry date, so the person can be scheduled and so you can prove it at an audit.

Who can do this. Administrator, for anyone. Caregivers can view their own.

Before you begin. Have the document as a PDF, PNG, JPG or SVG, 5 MB or smaller.

Steps

1. Operations Hub → Certifications.
2. Select Upload a certificate.
3. Choose the caregiver.
4. Choose the credential type. The list covers CPR / AED Certification, Certified Nursing Assistant (CNA), Home Health Aide (HHA) License, First Aid Certification, Medication Administration, Background Check, TB Test, COVID-19 Vaccination and Driver's License. For anything else, choose Other (specify)… and type the name.
5. Enter the Issuer, e.g. American Heart Association.
6. Enter the Expiry date.
7. Attach the file.
8. Save.

What happens next. The credential appears against that caregiver, and the counters at the top update: Total Credentials, Valid, Due Soon, Expired.

How to verify it worked. The credential shows under the caregiver with its issuer and "Expires [date]", and the person can now be assigned to visits.

WARNINGA credential with no expiry date counts as incomplete, and still blocks scheduling. The message says "(needs expiry date)". Uploading the document is only half the job — always enter the date.

Common mistakes

• Uploading a file larger than 5 MB, or a format that is not accepted. Photograph the certificate rather than scanning it at maximum quality.
• Recording the issue date instead of the expiry date.
• Uploading to the wrong person. Check the name at the top before you save.

How credentials block scheduling

When a caregiver is missing a required credential, Carely refuses the assignment:

"[Name] can't be scheduled — missing required credential(s): CPR / AED Certification (expired). Upload them in Operations Hub → Certifications."

The states Carely distinguishes are missing, expired and incomplete (needs expiry date).

NOTEBlocking can be switched off for the whole agency. Operations Hub → Certifications → Block scheduling when required credentials are missing. It is on for every new agency, and leaving it on is the right default. Turned off, a missing or expired credential is still flagged everywhere — it simply no longer stops the assignment. Everything else in this section describes the setting left on.

TIPThe reason a caregiver cannot override for themselves is that credential rules exist to protect clients, and the person under time pressure to cover a shift is the worst person to decide the rule can be skipped. If an exception is genuinely warranted, an administrator makes it and their name goes on it.

Renewing before expiry

Watch the Due Soon counter on the Certifications screen. Renewing before the date is a five-minute job; renewing after it means someone could not be scheduled that morning.

Make it a weekly habit — see Best practices.

Viewing and replacing documents

Each credential shows the document, with Download available. Some file types cannot be previewed in the browser; the screen says so and offers Download instead.

To replace an expired certificate, upload the new one with the new expiry date.

Client and agency documents

• Client documents live on the client record, alongside the care plan.
• Signed BAA — avatar menu → Account settings. It can be viewed, exported and printed, but never edited or deleted.
• Backups — avatar menu → Account settings → Backups. This is how you get a full copy of your data.

13. Compliance

Operations Hub → Compliance shows live compliance findings, computed from your real caregiver certifications and required-training completion. It is not a static checklist — the findings change as your data changes.

Findings include expired and missing credentials and training gaps.

Related screens:

• Training — required training and who has completed it. Caregivers see My Training.
• Quality — quality metrics derived from real survey responses and EVV clock-in data.
• Incidents — recorded events.

State rules

Carely holds a per-state rule set covering EVV requirements and models, HIPAA, and other state-specific obligations, and applies your operating state's rules. For example EVV is federally required for Medicaid personal care under the 21st Century Cures Act, but how you transmit differs: Texas uses a state-designated HHAeXchange aggregator, Florida runs it through managed-care plans, New York lets providers choose a system whose data must reach the state aggregator.

Resolving a compliance warning

1. Open Compliance and read the finding — it names the person and the problem.
2. For a credential: upload the current document with its expiry date.
3. For a training gap: have the person complete the training and record it.
4. Return to Compliance. The finding disappears once the underlying data is fixed.

WARNINGCompliance is a screen, not an export. There is no download. If you need a copy for an auditor or your own records, print the page.

Signed agreements — your permanent record

Avatar menu → Account settings → Signed agreements.

One place holding everything your agency and its people have signed:

For every record you get the person, the document, its version, the date, and a document fingerprint — a code proving which exact wording they agreed to, not merely that they clicked something.

Export everything gives you a file to keep or hand to a lawyer. Print / Save PDF gives you something to file.

These records are permanent. They cannot be edited or deleted by anyone — including Carely — and they are kept even if the account is closed, so they can still be produced if they are ever needed.

A departed caregiver's acceptance stays on file and is still named; the record simply notes the account has closed. That is deliberate: an employment dispute is exactly when someone asks what a person agreed to, and it is usually after they have gone.

What Carely does not guarantee

IMPORTANTCarely monitors the data you put into it. It does not certify your agency, it does not guarantee you are compliant with your state or with any payer, and it is not legal or regulatory advice. A clean Compliance screen means the records in Carely look complete — not that your agency would pass a survey. Keep your own relationship with your licensing body and your own counsel.

14. Communication

Chat

Every conversation is in one app: Chat. It is laid out like Microsoft Teams. Your chats are listed on the left, newest first. The conversation you open is on the right, and you type at the bottom. Chat is in the sidebar for administrators, caregivers and families, and on the bar at the bottom of a phone.

Each chat is named for the people in it, from your side:

Under every chat name is what it is about, and the lock line at the top of an open chat says exactly who can read it — "The family and the office. Caregivers can't see this chat." See Who you are messaging for the rules.

To write. Open the chat, type, and press Enter to send. Shift + Enter starts a new line. The camera button attaches a photo. Your message is in the chat the moment you send it, and the other side's replies appear within a few seconds — see Screens keep themselves current.

New chat (top of the list) starts a conversation you have not had yet: type a client, family or caregiver name and pick from the list. Families and caregivers already see all of their chats, including empty ones.

Search chats filters the list by name, client or what was last said.

Group chats, and adding people

New chat → To: — type a name and pick the people. One person makes a chat with just the two of you; two or more make a group. A group can have a name (for example Weekend cover) and can be about one client. Then Start.

You can put in a chat only people you can already reach:

In a group, the people button at the top shows who is in it. From there:

• Add people. Pick them, then choose whether they can see the earlier messages in this chat. Leave it off and they see only what is said from then on.
• Remove someone — the person who started the chat, or the office, can do this.
• Rename the chat.
• Leave this chat. You stop seeing it; someone in it can add you back.

Add people on a care conversation (an Office or Caregivers chat) starts a new group with the people already in that conversation, plus whoever you add. The conversation you came from stays as it is.

The office can read any group that has a family in it, the same way it reads the family's chat with the caregivers: supervising care is the agency's job. The office cannot write there unless it is added. A group of staff only is private to the people in it.

Editing a message

Point at your own message (or, on a phone, look for the pencil beside it) and choose Edit message. Change the words, then press Enter to save — or Esc to leave it as it was.

Everyone in the chat sees the new words, with Edited beside the time. You can edit only your own messages — not even the office can change what someone else wrote — and only the words: to change a photo, delete the message and send it again. A deleted message cannot be edited.

Carely keeps every earlier wording for your agency's records, encrypted, as it does a deleted message, and shows it to nobody in Chat.

Deleting a message

Point at a message (or, on a phone, look for the bin beside it) and choose Delete message. You can delete your own messages. The office can delete any message in the agency — for example one a family sent to the wrong chat.

Everyone in the chat then sees This message was deleted where it was. Carely keeps a copy for your agency's records, because care communications have to be kept, and shows it to nobody in Chat.

Every message is encrypted where it is stored — the words and the photos — including messages sent before this was added.

Reporting a message

Point at someone else's message (or, on a phone, look for the flag beside it) and choose Report message. Pick why, add a note for the office if you like, and press Report. The person who wrote it isn't notified.

Your agency's administrators get a notification in the bell, and an email that says only that something was reported. They review it in Chat → Reported messages, which shows the message, who wrote it, who reported it and why. From there they Remove message, which leaves This message was deleted for everyone, or Dismiss the report. The office can remove a reported message even from a staff chat it is not in, because the person who reported it chose to show it to them.

Blocking someone

In a chat with just one other person, press Block at the top. In a group, open the people in the chat and press Block beside the person. You can also tick Also block when you report a message. Once you block someone:

• you no longer see their messages anywhere in Chat, and you get no notifications for them;
• neither of you can message the other one to one, start a chat with the other, or add the other to a chat;
• they can still write in a group you are both in, but you won't see it.

The person you block isn't notified. Your agency's office is told in the bell, so a care relationship, such as a family and their loved one's caregiver, isn't cut off without anyone knowing. To undo it, press Unblock in the chat, or open Blocked people at the bottom of your chat list.

Words Carely won't send

Chat refuses a message containing slurs or abuse aimed at a person, and asks you to reword it. What you typed stays in the box so you can change it. The filter is deliberately narrow: care work means writing about bodies, so medical and personal-care words are never refused.

On a phone, the list and the conversation are separate screens. Tap a chat to open it, and the back arrow to return to the list.

Message family on a caregiver's client card (Operations Hub → My Families) and Message care team on a family's loved one (Family Connect) open the right chat in Chat directly.

IMPORTANTAdministrators are in Chat like everyone else. There is no separate admin inbox: the office reads and answers every conversation from the same Chat app.

Build checking Chat into someone's morning, exactly like the phone. A family that messages and hears nothing rings the office anyway, then tells the next family that Carely does not work. When a family or caregiver writes, the people on the other side get a notification on their phone that says only "You have a new message in Carely" — no name, no words.

WARNINGTell families plainly: chats are not monitored around the clock. In an emergency they must call, not message. Chat says this above the box a family types in, but say it during onboarding too.

Incident reports, and the count a family sees

An incident report written by a caregiver does not go straight to the family. It waits in Operations Hub → Incidents, marked Awaiting your review, where an administrator can read it, reword it if it needs to land more carefully, and then send it.

Until you send it, the family cannot see it. Once you have sent it, the wording is fixed — file a follow-up rather than editing something a family has already read.

The family is emailed that something is waiting for them, and nothing about what it says. To help them find it, their Incidents tab carries a count of the reports they have not opened yet — the same red pill Chat uses. Opening the tab clears it, because the list shows each report in full.

The count is per person: two relatives with access to the same client each have their own. You cannot clear it on their behalf, which is the point — it means what it says.

Announcements: reaching your staff or your families

A chat is about one client. Announcements are for everything that is not: the office is closed Monday, timesheets are due a day early, a policy has changed. Agencies usually send these by group text, which reaches personal phones, leaves no record, and tells you nothing about who actually read it.

Go to Chat → Announcements, pinned at the top of the chat list.

Write a subject and a message, then Send. Recipients are told a message is waiting; the message itself is read in Carely, never sent in the body of an email.

Knowing it was received

This is the part a group text cannot do. Every message you send is listed under What you have sent with a count:

11 of 14 read

Select Who has read it to see the names — who has opened it, and who has not. Chase the ones who have not, rather than assuming everybody saw it.

An announcement is marked read when the recipient opens Announcements in their own Chat. You cannot mark it read on someone's behalf, which is deliberate: a receipt is only worth having if it means that person actually opened it.

PRIVACY — Recipients never see who else received a message. A family reading an announcement cannot tell that it went to every other family, and cannot learn their names.

Where recipients read it

Caregivers and families read announcements in Chat → Announcements, pinned at the top of their chat list, with a count until they have opened it.

Announcements from your agency are separate from Carely's own service notices (maintenance windows and the like), which appear on your dashboard and are sent by Carely, not by you.

Notifications

The bell in the top bar carries alerts — care alerts, approvals waiting and other events. Assigned caregivers and administrators receive care alerts there.

Select an item to go to the screen it is about. The ✕ beside an item puts it away once you have seen it, and Clear all puts away everything at once. An item you have put away comes back only if what it says changes — "3 timesheets awaiting approval" becoming "4" is news again. Something that stays the same day after day — an incident nobody has resolved — is one item in the bell, not a new one each morning.

Email from Carely

Carely sends email; it is not a place you write email. There is no compose box for messaging a family — the automatic emails below go out on their own, and anything you want to say to a family is said in Chat, where it stays with the client's record and the whole team can see it.

Emails Carely sends — invoices, payment reminders, care updates — go out as your agency: your name in the header, your brand color, and only a small "Powered by Carely" at the bottom.

Nothing Carely sends accepts a reply. Every email is addressed from a mailbox that receives nothing, and says so in its own footer. This is deliberate. An email a family can answer is an email a family will answer with detail — "she fell this morning, her sugar was 3.2" — and that reply lands in a mail provider's store, outside Carely's encryption and outside its audit log. The emails carry no health information going out; closing replies means none can come back in either. A family who wants to answer signs in and answers in the care-team conversation, where the message is encrypted, kept with the client's record, and seen by the whole care team rather than one person's inbox.

Set these in avatar menu → Account settings:

• Agency name (shown on invoices & family documents)
• Your reply-to address (emails & invoices) — used to identify your agency and for your own records; Carely does not put it on outgoing mail as a live reply address
• Brand color — the header on your emails

Agencies that want mail sent from their own domain can verify one in the same place, when your platform has that option enabled.

What a caregiver is emailed about their schedule

A caregiver is emailed when a visit is added to their schedule, when one they hold is moved, when one is taken off them, and when one is deleted — calling a visit off is the change most likely to send somebody to a house for nothing, so it is the one they are most surely told about. Reassigning a visit emails both people — the person losing it first, since they are the one who would otherwise turn up.

These emails never name the client. They carry the day and time, and say to open Carely for the client, the address and the care plan. A schedule email lands in a personal mailbox over a channel you do not control, and a subject line is readable on a lock screen; keeping care details out of it is deliberate.

Two more things worth knowing:

• One email per change, not per visit. Booking a recurring pattern of twenty visits sends one email listing them, not twenty.
• Nothing is sent about the past. Backfilling last month's roster emails nobody — there is nothing left to turn up for.

A caregiver with no email address on file simply is not emailed; the visit books normally. Adding their address in Operations Hub → Caregivers starts it.

Open shifts: a daily digest

Once a day, every active caregiver with an email address is sent the visits that are currently open in the next two weeks — day and time, never the client's name. Opening Carely is how they see who and where, and how they claim.

It is deliberately a digest, not an alert per shift. Emailing the roster every time a visit opens would send a busy Monday's re-scheduling to forty people a dozen times over, and the real cost is not annoyance: complaints and ignored mail damage the reputation of the domain your password resets are sent from. So it is at most one email per caregiver per day.

Two consequences worth knowing:

• A day with nothing open sends nothing. An empty digest teaches people to ignore the next one that matters.
• It is honest that shifts are first-come. The digest lists what was open when it was sent; somebody may have claimed one by the time they look.

It goes out during the working day in your agency's own timezone — never at 3 a.m.

When a visit is still running

A visit still showing as in progress an hour after it was due to end now tells two people, on the hour:

• the caregiver, by email — "you are still clocked in", with the reassurance that if they are simply running late they can ignore it;
• the office, in the bell — naming the client, because an admin cannot act on a visit they cannot identify, and pointing at Operations Hub → EVV Exceptions if the caregiver cannot clock out themselves.

The hour of grace is deliberate: care runs late, and chasing every visit on the stroke of the hour would train everybody to ignore it.

This is what stops an EVV exception happening. A visit left running for sixteen hours becomes one, and the only way to close it then is for a person to attest to times nobody witnessed — a worse record than a punch, on the evidence Medicaid pays against. One email at hour one is the difference. It is said once per visit per day, not once an hour. → Closing a visit when the evidence is missing

What a family is emailed, and what it says

A family is emailed when something is waiting for them: a message from the care team, a care update, a visit report, an invoice, a payment reminder, an incident, or a change to the plan of care.

The email names the kind of thing and nothing else. No client name, no health detail, no amount, no date of birth — those are in the portal. So a notification on a lock screen, or in an inbox somebody else can read, discloses nothing about the person receiving care. Where a reference is unavoidable the client's ID is used, never their name.

Testing your email without emailing a family

ALIASES: test email, send a test, is email working, check email setup, try sending an email

Avatar menu → Account settings → "Send a test email to myself".

It sends one message to the address on your own account — nobody else, and there is no box to type a different recipient. If it arrives, everything else will: invoices, payment reminders, care updates and the notices caregivers get about their schedule all leave the same way.

If it does not send, Carely repeats what your email provider said, word for word. That sentence is usually the whole fix — most often that your sending domain has not been verified yet, which is a DNS record to add rather than anything wrong inside Carely.

Do this after any change to your email setup, rather than finding out from a family who never heard from you.

Checking an email actually arrived

ALIASES: email not working, email did not send, message not delivered, sent but not received, email failed, family did not get the email

Operations Hub → Sent emails logs every email Carely sent to a family, and the badge on each one is the truth about that email, not about your setup:

A "Failed" badge tells you what to fix. The provider's own words appear beneath the message — most often that your sending domain has not been verified yet, which is a DNS record to add rather than anything wrong inside Carely.

If every message says Failed with the same reason, it is your sending domain, not Carely. An email provider will not send from a domain until you have proved you own it, by adding the DNS records it gives you. Until then every message is refused the moment it is offered.

What Carely does not send

• No replies. Every Carely email is from an unmonitored mailbox — see above.
• No SMS to clients or families. Messaging is in-app; Carely sends email.
• No push notifications yet. Everything below reaches caregivers by email; phone push is built but dormant until the agency's push credentials are configured.
• No SMS to caregivers. Same reason.

Language

Carely's interface is available in English and Spanish. Each person sets their own: avatar menu → Language.

15. Reporting and dashboards

Where each report lives

There is no single Reports module. Reports live in the app that produced them.

The last two are behind your own name in the top-right corner. Nobody finds them by accident — show your team.

The Agency Growth Center

Opens with your Business Health Score and a daily briefing of what needs attention. Further down are forecasts, financial intelligence, benchmarking and the KPI scorecard.

NOTEIf you have just started, you will see "Not enough data yet". That is correct behavior, not a fault. The coach needs clients, visits and timesheets before it can say anything useful. It fills in over your first few weeks.

The written analysis, and the Regenerate button

Enterprise owners get the same thing across every agency under Enterprise Growth Center → Executive analysis, and a parent agency gets AI family analysis for its branches. Both carry a Regenerate button and a badge saying who wrote the words:

• AI ANALYSIS — written by the AI, and it will read differently each time.
• ENGINE ANALYSIS — written by Carely's own rules from your numbers. This is what you see unless an AI key is configured for the platform.

Your own Agency Growth Center says the same thing, above the executive summary: either "written by Carely AI" or "from Carely's analysis engine". If you asked for the written briefing and the AI was unavailable, it tells you so plainly rather than quietly showing you the other one:

The written briefing is unavailable right now, so this is Carely's own analysis of the same numbers. Every figure and recommendation below is unchanged.

That last sentence is the important one. Only the prose is affected. Your Business Health Score, the risks, the opportunities and every number on the screen are calculated the same way either way — the AI writes the summary, it does not do the analysis.

If you press Regenerate and the words do not change, nothing is broken. The engine is a straight reading of your metrics, so the same numbers produce the same sentences. The timestamp beside the button tells you it ran — it will say Updated just now — and the wording changes when the underlying figures do.

The Business Health Score

A single number built from 13 weighted measurements of your own data:

IMPORTANTThis is arithmetic on your own data, not an opinion and not a prediction. If a factor looks wrong, the underlying data is usually wrong — most often unapproved timesheets or missing referral sources. Fix the data and the score corrects itself. A low factor is a prompt to look, not a verdict. Referral Performance scores badly for agencies that simply do not record referral sources.

Forecasts

Revenue — next month, Capacity headroom and Hiring forecast, each shown with a confidence level.

WARNINGForecasts are estimates from your trend, not guarantees. Read the confidence alongside the number, and never commit money on a forecast alone.

Financial intelligence

Projected inflow (30d) (recurring plus collectible receivables), Projected payroll (30d) (trailing run-rate) and Net cash flow (30d). Revenue is also broken down by service line, by payer and by referral source.

Benchmarking — how you compare

Anonymous comparison against agencies of your size.

NOTEWhen your peer group is too small to be anonymous, Carely falls back to published reference points rather than showing you a comparison that could identify another agency. The screen tells you which one you are seeing.

Quality metrics

Operations Hub → Quality derives its figures from real survey responses and EVV clock-in data — not from static sample numbers.

16. Carely AI features

Carely uses the word "AI" for several different things. Knowing which is which tells you how much to trust each one.

What Carely Copilot is — and is not

IMPORTANTCopilot is not a chatbot. Each agent is ordinary code running checks somebody specified. They propose; anything that changes your data waits for your approval.

The eleven agents watch: credentials, compliance, scheduling, staffing, documentation review, billing integrity, payroll integrity, client risk, incident response, analytics and revenue optimization, plus a Medicaid rails agent for agencies that transmit.

Operations Hub → Copilot shows:

• What the agents found — the latest findings across every domain, refreshed by events such as an incident or an expiring credential.
• Actions needing a decision — what the agents prepared. Approving executes immediately.
• Agent activity ledger — every run: which agent, why, what it saw, which tools it used.

Autonomy: how much agents may do alone

One dial for the agency, with per-agent overrides:

There is also a break-glass control that halts every agent for your agency immediately, recorded as an auditable event.

TIPStart at level 2 and stay there until you have read a few weeks of the ledger and agree with what the agents propose. Raising the dial is a decision about your agency's risk tolerance, not a performance setting.

How to check an AI recommendation

1. Open the finding and read why — every finding states what it saw.
2. Go to the underlying screen and confirm the facts yourself.
3. If it is wrong, the data is usually wrong. Fix the data.
4. If the data is right and the recommendation is still wrong, use Report a problem.

Privacy: what reaches a language model

IMPORTANTOnly Carely's already-calculated numbers go to the language model that writes the briefing prose, and client and caregiver names are removed first. Never type client names or health details into Ask Carely. It is a help assistant for using the software, not a place for patient information.

What AI is not allowed to do

• No clinical decisions, ever.
• No moving money on its own.
• No changing your data without approval, except low-risk reminders at level 4.
• No deciding anyone's pay.

Who you are messaging

There are separate conversations, not one room. Every client has up to three:

Caregiver-to-office is one conversation per caregiver, not one per client. If two caregivers attend the same client, each has their own private line to the office and neither can read the other's. A caregiver raising a safety concern, or saying they are struggling, is writing to the office alone. When the office replies it picks the caregiver it is answering, so an answer meant for one person never appears in front of the whole team.

Pick the chat before you type. Each chat is one of these conversations, and the lock line at the top of an open chat says who can read what you are about to write.

Why it matters. A family with a concern about the person coming into their home needs a way to tell the agency without that person reading it. That is the family-to-office conversation, and it is enforced by Carely, not by remembering which screen you are on.

The office sees family-to-caregiver messages. Supervising care is the agency's job and a complaint is usually for them. It is said on screen rather than left as a surprise.

Where to find each one. All of them are in Chat. The office sees every care conversation; families and caregivers see only the ones they are in. Group chats follow the rules in Group chats, and adding people. The office's chats with caregivers are listed by caregiver, because that is who the office is writing to.

Messages sent before this change have been sorted into the right conversation. Carely placed each one by who wrote it: office messages into family-to-office, caregiver messages into family-to-caregiver, and a family's reply into whichever conversation it was answering. Older caregiver-to-office messages were matched to the caregiver who wrote them, and an office message with no caregiver before it was left as it was — visible to that client's team, as it already was — rather than guessed at. Nothing was made visible to anyone new — a family's old message never lands in caregiver-to-office, and the only change in who can read what is that a caregiver no longer sees what the office wrote to the family. Every message was placed, so Earlier messages is gone from every screen; nothing was deleted, and no one lost their own history.

Unread counts

Every chat carries its own count. A red bubble on a chat in the list says how many messages are waiting in it, and unread chats are in bold. The number on Chat in the sidebar (and on the phone's bottom bar) is exactly the total of the bubbles in the list, unread announcements included. If it says 3, three messages are waiting, and the list shows which chats they are in.

Opening a chat clears only that one. Reading Mary's family's messages does not silence Robert's, and for the office, reading one caregiver's chat does not silence another's. Nothing opens by itself: Chat starts on "Pick a chat", and a count clears only once you open the chat it belongs to.

NOTECounts are per person. Clearing yours does not clear your colleague's.

17. Multi-location and multi-agency

Carely supports two different shapes. Choose by asking whether you are one business or several.

Sub-units

A new location signs up normally but enters the Parent agency's organization code. No payment step appears — the parent's subscription covers them.

That code is also the one your caregivers and families sign up with, so entering it only asks to become your branch. Nothing is linked, billed or covered by your Business Associate Agreement until you approve. Each administrator gets a notice in the bell, and the request waits in Operations Hub → Sub-units → Branch requests, showing who asked and when. Approve links the location: it bills through you from that day ($19/month, up to 3 locations) and opens for its staff. Decline asks you to confirm, then links nothing; their sign-in tells them the request was declined.

Parents track their locations in Operations Hub → Sub-units, which appears once the first request arrives. It shows each location's metrics and health, and lets you switch into one.

IMPORTANTAnyone administering the parent agency can switch into a sub-unit and operate it with full administrator access — every client record, every financial screen. This is by design for a single business with several offices. Make sure local managers know it, and that it matches how you actually want your business run.

WARNINGYour subscription tier is set by the combined active-client count across the parent and every sub-unit. A sub-unit's growth can move the parent's tier. Check Subscription before adding a location.

Enterprise

An enterprise owns several agencies. Set it up from avatar menu → Account settings → Enterprise, or join an existing one at signup with its ENT-… code.

The Enterprise Portal adds a corporate dashboard across agencies, alerts, per-state compliance, and regional directors with access scoped to their agencies.

Switching between agencies re-issues your session for the agency you moved to, so you are only ever operating in one at a time.

Which do I have?

Look at your Operations Hub navigation. A Sub-units tab means you are a parent agency. An Enterprise Portal in your app list means you are in an enterprise. Neither means you are a single agency, which is what most subscribers are.

18. Settings

Almost everything agency-wide is in one place: avatar menu (top right) → Account settings. Some settings appear only for Administrators.

Other settings live where they are used:

Backups

Purpose. A full copy of your agency's data, on demand.

1. Avatar menu → Account settings → Backups.
2. Select the option to create a backup.
3. Download it, or leave it in the list.

Backups can be restored and deleted from the same screen. Carely also runs continuous encrypted off-site backups of its own — the on-demand backup is your copy, under your control.

Once your agency has any records — clients, caregivers, visits or timesheets — Carely also takes an automatic backup every 2 days. The Off-site backup line under the heading says when the last encrypted copy reached off-site storage. If the newest copy did not arrive, it says last copy failed instead. Your backup is still saved in Carely, and Carely's own health checks flag the failure for the Carely team.

TIPTake a backup before anything large: a migration, a bulk change, a rate revision across many clients. It costs a few seconds and it is the difference between an inconvenience and a very bad day.

Deleting your agency

The Danger zone in Account settings starts agency deletion. There is a 30-day grace period in which you can export your data or reactivate. After that the account is archived and purged.

WARNINGDo not start this to "see what happens." Export your data first, and be certain. Once the grace period ends, it is over.

Your agency's time zone

Carely works on your clock. Days, pay periods and "today" all begin at midnight where your agency is, not in UTC.

You do not normally have to set this — Carely uses the time zone of the state you are licensed in. Set it explicitly only if your agency keeps a different clock from its licensed state: a border town, or a multi-state operator run from another office.

Why it matters. A pay period runs from your Monday to your Sunday. Before this, periods ran on UTC — so a visit worked on a Sunday evening fell into the following pay period and was paid up to two weeks late. The gap was four hours on the east coast and ten in Hawaii. Arizona and Hawaii, which do not observe daylight saving, had the widest gap of all.

Daylight saving is handled for you. A week containing a clock change is still seven days, and pay day stays on the date it should.

To change it: avatar menu → Account settings → the agency section. Clearing it returns you to your state's zone.

Times that record when something actually happened — a clock-in, a clock-out, an audit entry — are stored in universal time on purpose. They are evidence of a moment, and are shown back to you in your own zone.

19. Subscription and account management

How you are priced

On active clients, counted daily and de-duplicated — not per user. Add your whole office at no extra cost.

Every plan includes every feature. Tiers buy capacity, not capability. There is no feature locked behind a higher plan.

What counts as a billable client

A billable client is one whose record or care generated qualifying activity in Carely during the last 30 days — a visit, an EVV clock-in, a visit note, or an invoice. It is measured from what Carely recorded, not from the status field on the client's card.

On Hold and Hospitalized clients count. They are still your clients; you are still managing them here.

Changing a status does not undo care already delivered. If you archive or discharge a client in the middle of a month, the visits they received earlier that month still count for the rest of the 30-day window. This is the one rule that surprises people, so it is worth stating plainly: archiving is not a way to lower this month's bill.

And a client who genuinely leaves stops counting on their own. No action is needed from you. Once they have had no visits, notes or invoices for 30 days, they drop out of the measurement — whether you archived them, discharged them, or did nothing at all. Deleting a client's record does not change this either way.

Clients are also de-duplicated by name, so a client entered twice is billed once.

How tier changes actually work

IMPORTANTCrossing your limit does not bill you immediately. There is a grace band above your plan's maximum, and you must stay above it for 14 consecutive days before an upgrade is scheduled. Coming back down takes a 60-day window. Those numbers are deliberately different. Growth should be recognized promptly and shrinkage should not be — a bad month should not bounce your plan up and down.

You are notified along the way: at 80% of your capacity, when a sustained streak begins, part-way through it, when a change is scheduled, and when it applies. Read those notifications — by the time the bill changes, you have had several warnings.

Upgrading or downgrading manually

Avatar menu → Account settings → the billing section. Your current plan, active-client count and any scheduled change are shown there.

If you have grown past 500 active clients, Carely does not raise your price automatically — it tells you that your agency has grown into Enterprise territory and asks you to contact the team.

Paying monthly or annually

Avatar menu → Account settings → the billing section. If annual billing would save you money, Carely shows Switch to annual — save $X/yr with your actual saving worked out from your current plan. If you are already on annual, the same place offers Switch to monthly.

Either way Carely asks you to confirm first, and the confirmation spells out what happens before you commit:

• Monthly → annual. You are credited for the unused part of the current month, then charged the annual rate.
• Annual → monthly. The unused part of your annual term is credited against your monthly charges, and you are billed every month from then on.

Nothing else about your account changes — same plan, same features, same data. You can switch back at any time, and switching is not the same as canceling.

If your agency has a custom arrangement

Some agencies are on a negotiated plan. If so the billing screen says "Your plan is set by a custom arrangement with Carely", and automatic tiering is paused.

If you paid and still can't get in

Carely learns that you have paid when Stripe tells it. If that message is lost or fails to process, your agency can be left waiting even though the payment went through — you would have a receipt and no access.

Carely checks for this itself. The Payment Recovery agent looks for agencies sitting in the holding state that Stripe says are paying, confirms it against Stripe directly rather than trusting any message, and restores access. It only ever opens an account, never closes one, and it changes nothing when Stripe cannot be reached or does not confirm the payment.

If it happens and is not resolved within a few minutes, contact support with your Stripe receipt — the account is waiting, not lost, and no data is affected.

Payment problems and the holding state

If payment has not completed or has failed, your agency enters a holding state: everyone can sign in but the apps do not open.

Your data is not deleted. It is waiting.

To fix it: complete payment, then sign out and back in so your session picks up the new status. This last step is what people miss — the session is stamped with your status when you sign in.

Sub-units and billing

If you are a parent agency, your tier is based on the combined count across you and every approved sub-unit, and each approved sub-unit adds $19/month to your bill. A request you have not approved is not counted or billed. A sub-unit pays nothing itself.

What you agree to before the first charge

Before any subscription starts, Carely shows you the renewal terms in full and asks you to check a box confirming you understand them. That box is not pre-checked and the Subscribe button will not work until you check it — deliberately, because it is the record of what you agreed to.

What is shown, and what is stored: the amount, how often it charges, when the first charge lands if you are on a trial, that your plan can change if your active-client count changes, and how to cancel. If the price or the billing cycle later changes, you are asked again — your earlier agreement does not carry over to a different price.

You can see everything you have agreed to at any time in Account settings.

Reminders before you are charged

Carely writes to you before money moves, not after:

Each notice arrives by email and in your notification bell. If you have already canceled, these stop — Carely will not tell you that something is about to renew when it is not.

Changing your card or bank account

Avatar menu → Account settings → Payment method.

Your card or bank account is changed inside Carely — you are not sent off to a separate payment website. The screen shows what is currently on file (for example Visa ····4242, expires 4/2029), and warns you if it is about to expire.

Carely never sees or stores your card number. The details are typed into a secure field supplied by our payment processor and go straight to them; Carely only ever receives a reference to the saved method. That is deliberate, and it is why the field looks slightly different from the rest of the form.

You do not need an account with the payment processor. There is nothing to sign up for and no second password — you are only ever in Carely.

Once saved, the new method is used for every future charge, including the subscription you already have. Your existing invoices stay on this screen.

Who your invoices are addressed to

Avatar menu → Account settings → Billing details.

What Carely prints on your invoices and where it sends them: the company name they are addressed to, the billing email, a phone number, and the postal address. Change any of it here — it applies from your next invoice, and does not alter invoices already issued.

Tax ID (EIN). If your accountant needs your EIN on the invoice, add it here. Tax IDs are added and removed rather than edited: a corrected number is a new entry, so add the right one and remove the old.

This is your agency's own billing contact — it is not connected to any client record, and nothing here is patient information.

Canceling your subscription

Avatar menu → Account settings → Cancel subscription.

You are asked to confirm, and the confirmation is deliberate rather than a quick "are you sure?": it lists what will happen, what will not happen, and the exact date your access runs to. You then check a box confirming you understand before the button will work. The prominent choice is Keep my subscription — canceling should be easy to do, but not easy to do by accident.

Canceling your subscription is not deleting your agency. They are different things and they live in different places.

When you cancel:

• You are not charged again.
• You keep full access until the end of the period you have already paid for.
• Nothing is deleted. Your clients, visits, documents and records all stay exactly where they are.
• You can resume at any time before the period ends, with one click in the same place.

If you want your data removed as well, that is a separate, deliberate step — see the Danger zone below.

Your own copy of the data, kept automatically

Carely already replicates your database continuously and keeps 90 days of history. That covers a failed server or a corrupted database. It does not cover losing the hosting account itself, because the replica lives in the same place as the thing it is protecting.

For that, Carely can keep a verified copy on a computer you control, on a daily schedule. Each copy is checked after it is downloaded — if it does not open cleanly as a database it is quarantined rather than kept, so the newest copy in the folder is always one that actually works. Fourteen are retained and older ones are removed automatically.

The copy is a complete database. Keep it on an encrypted disk, and keep any external drive you move it to encrypted as well.

Deleting the agency entirely

Agency deletion lives in Account settings → Danger zone. It is a 30-day grace period in which you can still export your data or reactivate, followed by archive and purge. It also stops billing.

Export before you start. Most people who think they want this actually want to cancel the subscription above, which keeps the data.

20. Security and privacy for users

You are handling health information about vulnerable people. Most of what protects it is habit, not software.

For everyone

A sign-in belongs to one browser tab

In a web browser, signing in signs in that tab. Reloading the tab keeps you signed in; closing it, or opening Carely in a new tab or window, asks for your password again. That is deliberate: on a shared computer, someone opening carelyapp.care in a fresh tab should not find themselves inside your agency. The installed desktop and phone apps have one window each, and there your sign-in persists as before.

If a device is lost or stolen

1. You, from any other device: avatar menu → Account settings → Session security → Sign out all other devices.
2. If you cannot get in at all, tell your administrator immediately so they can remove or change your access.
3. Change your password.
4. If the device had 2FA on it and you have no other authenticator, contact Carely support — your administrator cannot clear it.

For caregivers

• Do not photograph client documents on a personal phone. Upload through Carely.
• Do not discuss clients on social media, however anonymous it feels.
• Write visit notes as though the family will read them — they will.
• Do not let a client or relative use your login, even to "just look at something."

For administrators

• Keep the number of Administrator accounts small, and remove the areas each one does not need. An administrator starts with full access until you narrow it. → Limiting what an administrator can do
• Remove departing staff the same day, both the caregiver record and the login.
• Review Security activity periodically. Permissions decide what someone can do; the activity log is how you see what they did.
• Keep a second administrator. If your only admin loses their 2FA device, you are dependent on support.
• Do not post your organization code publicly. Anyone with it can request access. You approve each request, but you do not want a queue of strangers.
• Never share credentials for a Medicaid or EHR connection over email or chat. Enter them directly in Carely.

What Carely does on its side

Protected health information is encrypted at rest, access is scoped to your agency, actions are recorded in a tamper-evident log, backups run continuously and encrypted off-site, and a signed BAA is held for your agency.

NOTEThis manual deliberately does not describe how those protections are implemented. If you need that detail for a security review, ask Carely directly rather than looking for it in a customer document.

Where Carely's AI runs

The features that use a language model — reading an intake packet, ranking caregivers for a client, the auto-scheduler's choices, the morning briefing, Ask Carely — send only what each needs, never a database. Carely can run them two ways, and your platform operator chooses: through Amazon Bedrock, where the same Claude models run inside AWS under the Business Associate Agreement that already covers Carely's hosting and backups; or through the model vendor's own API. System Health names which is in use. An operator can also switch the features that would send client information to a model off entirely, and Carely then uses its own rules for each of them.

Carely records how every AI request went — the feature, how long it took, and whether Amazon's limit made it wait — and never its contents. When requests start waiting or failing, the platform operator is emailed with what to do; you do not need to report slowness yourself.

Your privacy responsibilities

Carely helps you keep records. It does not make you compliant with HIPAA or with your state's rules, and nothing here is legal advice. Your policies, your training and your own counsel do that. Signing the BAA at signup is the beginning of that relationship, not the end of it.

21. Troubleshooting

Organized by symptom, because that is what you know when something goes wrong.

Each article follows the same shape: what you are seeing, the likely causes, what to check first, how to fix it, how to confirm the fix, and what to do if it did not work.

Access and sign-in

My agency is signed up but says access is paused

What it means. Your agency exists and your data is safe, but Carely is waiting on a subscription before it opens the app. You see "Complete your subscription to start using Carely."

What to do. Start a subscription from Account → Billing. Nothing has been charged and nothing has been lost — everything is where you left it.

If you signed up recently and never got your free trial, that is on us and we will fix it: email support@carelyapp.care and we will start your 30 days. You are not asked to pay for time you could not use.

I cannot sign in

Most likely causes. Wrong username; account not yet approved; agency in the payment holding state; caps lock.

Check first. Does the error say your details are wrong, or does it let you in but show no apps? Those are completely different problems — the second is the holding state.

Fix

1. Try your email address instead of your username, and the reverse.
2. Use Forgot your username? Recover it if you are unsure which it is.
3. Use Forgot password? and follow the emailed link within 30 minutes.
4. If you recently signed up, ask your administrator whether your account has been approved — an unapproved account cannot sign in.

Verify. You reach your app home screen.

If that did not work. Your administrator can check Operations Hub → Users for your account and its status. If the only administrator is locked out, contact Carely support.

Sign-in says my account is paused or locked

Most likely cause. Several wrong passwords (or two-factor codes) in a row for this account — yours, or someone guessing at it.

Fix

1. Select Reset your password to unlock on the sign-in screen (or Forgot password?) and follow the emailed link. The new password unlocks the account immediately.
2. A 15-minute pause also lifts on its own; a lock (after 10 failures) does not.
3. No email on your account? Ask your administrator to open Operations Hub → Users and select Unlock on your account.

Verify. You can sign in with your password.

Carely asks me to confirm my email before continuing

Cause. The 7-day window to confirm an agency administrator's email address has passed.

Fix

1. Open the confirmation email from Carely and select the link. Check spam.
2. No email? Select Send me a new link on the confirm screen.
3. Wrong address? Select Wrong address? Change it, enter the right one and your password.
4. Then select I've confirmed it — continue.

Verify. Your apps open as usual.

The password reset email never arrived

Most likely causes. Wrong or missing email on the account; spam filter; the account does not exist.

Check first. Confirm with your administrator which email address is on the account.

Fix. Check spam. Have your administrator confirm the address in Users. If the account has no email address, it cannot be recovered by the user — an administrator or Carely support must intervene.

NOTEThe screen shows the same message whether or not the account exists. That is anti-enumeration by design, not a failure. It is not telling you the request worked. It does now point you at the next things to try — spam, another address you may have signed up with, or asking your administrator to send the link directly.

For administrators. You can send a reset link yourself from the agency's Users screen, which goes to the address actually on the account rather than the one somebody typed. If that arrives and the self-service one did not, the address they entered is not the one on file. Carely records an unmatched attempt in the audit log, so this can be confirmed rather than guessed at.

I have lost my phone and cannot get past two-factor

There are no backup codes and your administrator cannot reset 2FA. Only Carely support can clear a lost authenticator.

Contact support with your agency name and username. Expect to prove who you are.

Prevent it next time. Use an authenticator app that backs itself up, and never enable 2FA on your agency's only administrator account.

Everyone can sign in but no apps open

Cause. The agency is in the pay-before-access holding state — payment has not completed or has failed.

Your data is not deleted. It is waiting.

Fix

1. Complete payment in avatar menu → Account settings → billing.
2. Sign out and sign back in. Your session is stamped with your status when you sign in, so it will not change under you.

Verify. Your apps appear on the home screen.

A user cannot see a feature / a menu item is missing

Most likely cause. Their role. See the role-access matrix.

The three that come up most

• Administrators have no Family Connect. Use Chat.
• Caregivers have no Revenue Center. They have My Pay.
• Clinical screens are invisible unless the clinical module is on for the agency.

Fix. If they genuinely need administrator access, change their role in Operations Hub → Users. An administrator starts able to do everything; you can then take specific areas away from them. → Limiting what an administrator can do

My invitation does not work

Cause. Carely does not send invitations. There is nothing to expire.

Fix. Send them your organization code (avatar menu → Account settings → Copy) and the address carelyapp.care. They create their own account with it; you then approve them in Users.

People and records

A caregiver is missing from the list

Most likely causes. They were never added as a caregiver record; they were deactivated; you are looking at a filtered list.

Check first. Operations Hub → Caregivers, with any filters cleared.

Fix. If they are not there at all, add them. If they were deactivated, reactivate them from the same screen.

IMPORTANTA caregiver who signed up and was approved has a login, not a record. They will not appear in the caregiver list or any caregiver dropdown until you create the record. These are two separate steps.

A client is missing from the list

Most likely causes. Their status is Discharged, Deceased or Archived; a search filter; they were permanently deleted.

Fix. Clear the search. Check whether they show a terminal status — those clients remain but are treated as inactive. If they were deleted, their full record is in Operations Hub → Archived Records.

A caregiver cannot see the visit I scheduled

Work down this list — this covers almost every case:

1. Do they have a login, and has it been approved? Operations Hub → Users.
2. Does the name on the login exactly match the name on the caregiver record? Carely connects them by name. A middle initial on one and not the other separates them.
3. Is the visit actually assigned to them, or is it still an open shift?
4. Are they looking at Schedule → My shifts? Open shifts are a different tab.
5. Is the visit on the date they are looking at?

Verify. They can see the visit under My shifts with a Clock in button.

A family member logs in and sees nothing

Cause. They are approved but not linked to a client. Approval grants a login; linking grants access to a person.

Fix

1. Operations Hub → Clients & Matching, open the client.
2. Family portal access → Family member who can see this client → choose them.
3. Save.

Verify. Ask them to sign in; their loved one appears in My Loved Ones.

If they are not in the list. They have not been approved with the Family Member role yet. Do that in Users first.

A family says they messaged and nobody replied

Cause. Their message is in Chat, which nobody opened. The count on Chat in the sidebar shows it is waiting.

Fix. Read and reply there. Build a daily check into someone's routine.

Prevent. Tell families at onboarding that messages are not monitored around the clock, and that emergencies mean calling, not messaging.

Scheduling

Carely will not let me schedule this caregiver

Two different refusals with two different fixes. Read which message you got.

If the assignment is being refused and the person's schedule looks empty, it is credentials.

A caregiver cannot claim an open shift

Cause. Their own required credentials are not current, they are not credentialed for the visit type, or claiming it would double-book them.

They cannot override any of these. Only an administrator can assign an exception.

Fix. Read the message they received, fix the underlying cause — usually a credential — and have them try again, or assign them directly.

A caregiver is getting offered less work than expected

Cause — nearly always the availability trap. Until a caregiver saves availability once, they count as available. After the first save, every unmarked day counts as unavailable.

Fix. Have them open Schedule → My availability and fill in all 14 days, marking each Day, Evening, Overnight, Any or Off.

Verify. The confirmation states how many days are marked available. It should be the number they expect.

I approved leave but the shifts are still assigned

Cause. This is deliberate. Approving leave lists the visits that fall inside it and changes none of them — Carely will not move somebody's care on its own.

Fix. The list appears as soon as you approve. For each visit choose Reassign to…, Release as open, or Cancel visit. If you closed it, re-open the request to see the conflicts again. → Approving leave when visits are already booked

If the request says "Dates not recorded", it was made before Carely recorded leave dates and cannot be checked against the schedule. Select Record the dates beside the warning and give the first and last day off; any visits already inside those dates are listed as soon as you save. → Leave dates, and why they matter

The AI auto-schedule filled nothing

Most likely cause. Required credentials are not uploaded, and credential blocking is on (the default). A caregiver missing a required document is not eligible for any visit, so on a new agency the first run can come back completely unfilled.

Check first. Open one of the unfilled proposals. It names the constraint that stopped each candidate — missing or expired credential, already booked, has not marked the day available, on approved leave, would pass their weekly hour limit. That list is the diagnosis.

Fix

1. Upload the missing credentials with their expiry dates, in Operations Hub → Certifications, or switch blocking off there if that is your agency's policy.
2. Have caregivers fill in all 14 days of availability — one saved day makes every unmarked day count as unavailable.
3. Run it again.

The other common cause. There were no open shifts in the period. The auto-scheduler staffs visits that already exist and does not invent them, so a week with everything already assigned has nothing for it to do — the count reads 0 open shifts analyzed, which is the correct answer rather than a fault.

The auto-schedule refused to publish something

Cause. Somebody changed that visit after the schedule was generated — assigned it by hand, moved its time, or deleted it.

This is deliberate. The person who made the change wins; a schedule generated at 10:00 does not overwrite an edit made at 10:03. Carely names each refused proposal and why.

Fix. Publish the rest, then regenerate to pick up the current state of the calendar.

A recurring visit did not generate correctly

Cause. Repeat weekly creates the visits at the moment you save — separate visits, not a live series. There is no background job creating them later.

Fix. Check the confirmation you saw: "Creates [n] weekly visits, [first] through [last]." If you need more, create another run. If you need fewer, delete the extras individually.

Remember. Repeat covers one weekday. Monday, Wednesday and Friday is three runs.

A visit is missing

Check. Was it deleted? Was the client discharged — discharging cancels upcoming visits? Are you looking at the right week?

Note. Deleting a visit removes the evidence it existed. Check Security activity to see whether someone deleted it and when.

EVV and visits

A caregiver cannot clock in

Follow the EVV decision guide. The answer is usually step 3: they are on the wrong screen. Clock in lives on Schedule → My shifts.

A caregiver cannot clock out

See cannot clock out. Only the assigned caregiver can clock out — nobody can do it for them. Where they cannot, the office closes the visit through an EVV exception, which records a correction beside the original evidence rather than editing it. → Closing a visit when the evidence is missing

Location verification failed

What you will see. Not EVV verified, with a reason.

IMPORTANTThe visit still counts and is still payable. An unverified visit is a record with a flag on it, not a lost visit.

Every visit for one client shows "outside the client's location"

Cause. The client's location anchor is wrong — it was set from a first clock-in that happened somewhere slightly off, such as a parking lot.

Check first. Open the visit's Verify location panel — it shows each punch's distance and plots it against the client's location, which tells you at once whether the caregiver moved or the anchor is simply in the wrong place. → Verifying exactly where a visit happened

Fix. Open a visit where you know the caregiver was at the house, select Verify location, and choose Use as client's location on that punch. That sets the anchor from real data and confirms it in one step.

You can also set it from the client record: Operations Hub → Clients & Matching → the client's EVV location row. → Confirming a client's location

Do not discipline a caregiver over this pattern. When every visit for one client fails, the anchor is the cause, not the caregiver.

A visit is stuck In Progress

Cause. Somebody clocked in and never clocked out.

What you can do. Administrators still have no controls on the visit itself: no edit, no reassign, no delete, and only the assigned caregiver can clock out. What you can do is record a correction beside it.

Fix — the caregiver is reachable

1. Contact the caregiver and ask them to open Carely and clock out.
2. The visit completes and produces a timesheet with the real elapsed time — which may be enormous if they clocked out the next day.
3. Deny that timesheet with a clear reason and have them log the correct hours. → Correcting a timesheet

Fix — they have left the agency, or the phone is gone

Operations Hub → EVV Exceptions. The visit is listed there automatically after 16 hours. Raise an exception, say what happened, and enter the real times. Carely writes an administrative clock-out that is labeled as entered by you and carries no location — it never claims a device verified it. A change of more than 30 minutes waits for a second approver. → Closing a visit when the evidence is missing

You no longer need to contact Carely support for this.

Prevent. Train caregivers to clock out before leaving the house. Forgetting is the costliest caregiver habit in Carely.

A visit appears incomplete

Check. Was the visit report filed? Reports are completed at clock-out. If the caregiver clocked out without completing it, the visit is Completed but has no report — ask them to add one.

Time and money

A timesheet is wrong / hours are incorrect

Nobody can edit hours. Use the deny-and-relog path in Correcting a timesheet.

A visit is missing from billing

Work down this list, in order of likelihood:

1. The timesheet is not Approved. Operations Hub → Timesheets. This is nearly always the answer.
2. The visit was never clocked out, so there is no timesheet at all.
3. It is already on another invoice.
4. The client's billing period has not closed. It will be in Upcoming, not missing.
5. The client has no charge rate, so it appears at zero.

An invoice total looks wrong

Check, in this order

1. Is any client showing hours but no money? They have no charge rate. → Rates
2. Are there Pending timesheets for that period? Those hours are excluded.
3. Is there a visit with impossible hours? A late clock-out inflates one row and the total with it.
4. Was a rate changed mid-period? Each visit is priced at the rate that applied on its own date, so a period spanning a change legitimately mixes two rates. Open the client and choose Rate history to see which rate covered which days. → Changing a rate: it applies from a date

A caregiver says their pay is short

Fix. Have them compare hours submitted against hours approved on Revenue Center → My Pay. Pay is calculated from approved hours, including visits already paid in a locked pay run, which show as Paid. Pending hours are not counted yet. Mileage reimbursement is listed on its own line and added to take-home after tax.

Usually the hours are there and simply not approved yet. Tell caregivers to make this check before calling the office; it resolves most of these calls without you.

Payroll totals look far too high or too low

1. Check the pay period dates at the top of the Payroll screen. Everything below responds to them, and they are the first thing to get wrong.
2. Check Pending Approval is zero. Unapproved hours are excluded.
3. Look for a late clock-out inflating one visit.

We were invoiced or charged twice

Billed work is tied to its invoice on the server and Carely refuses to bill it again. If you are seeing a genuine duplicate, report it with both invoice numbers.

Our subscription bill went up and we did not change plans

Cause. Your active-client count crossed a tier boundary and stayed above the grace band for 14 consecutive days.

Check. Your notifications — Carely warns at 80% of capacity, when the streak starts, part-way through, when the change is scheduled, and when it applies.

If you have sub-units: the count is the whole family, parent plus every sub-unit.

To come back down, the count must stay low across a 60-day window. That asymmetry is intentional.

Compliance and documents

A credential still shows as expired after I uploaded it

Most likely causes. No expiry date entered; uploaded against the wrong person; the old expired record is still there.

Fix

1. Open the credential and confirm it has an Expiry date. Without one it counts as incomplete and still blocks scheduling.
2. Confirm it is on the right caregiver.
3. Confirm the new expiry date is in the future.

Verify. The Expired counter drops, and the caregiver can be assigned to visits.

A document will not upload

Check. Is it a PDF, PNG, JPG or SVG? Is it 5 MB or smaller?

Fix. Convert or compress it. A photograph of a certificate taken on a phone is usually fine and well under the limit; a high-resolution scan often is not.

A compliance warning will not clear

Compliance findings are computed from live data. The finding disappears when the underlying record is fixed — not when you acknowledge it. Fix the credential or the training record, then reopen the screen.

The application itself

The page will not load

1. Reload the page.
2. Check your internet connection.
3. Try a different browser or device to see whether it is local to you.
4. If everyone in the office is affected, report an outage.

Changes are not appearing

1. Reload the page.
2. Check whether you are looking at a filtered list or a different date range.
3. Confirm you are in the right agency if you administer more than one — see Multi-location.
4. If you just fixed a subscription problem, sign out and back in.

A dashboard number looks incorrect

Check. Almost every dashboard number is derived from your own records. Before reporting it, confirm the underlying data:

• Unapproved timesheets suppress revenue and payroll figures.
• Missing referral sources make Referral Performance look terrible.
• A brand-new agency shows "Not enough data yet" — correct, not broken.

If the underlying data is right and the number is still wrong, report it with the screen, the number, and what you expected.

An AI recommendation looks incorrect

1. Read the finding's stated reason — every finding says what it saw.
2. Check that data yourself on the underlying screen.
3. If the data was wrong, fix it; the finding will correct itself.
4. If the data was right, report it. Include the finding and why you disagree.

Nothing acts on your data without approval, so a wrong recommendation is a nuisance rather than a risk — unless you approve it without reading it.

An EVV or Medicaid connection failed

1. Which environment is it in? Nothing real transmits from sandbox — by design.
2. Has it passed a live test? Production transmission requires it.
3. What did the vendor's message say? Rejections come from your vendor, not Carely. Common causes: a member ID that does not match the state's record, a service code outside the authorization, or a visit outside authorized dates.
4. Are the credentials still valid? Vendor credentials expire and get rotated.

Before the next one, check Operations Hub → Authorizations. Carely warns at scheduling time when a visit falls outside the approved dates, uses the wrong service code, or would exceed the approved units — which is where these rejections come from. → Authorizations: what a payer has approved

A notification was not received

Check what Carely actually sends, because some of these are deliberate silences and some are real email:

So a missing digest or a missing schedule email is worth investigating; an absent instant alert on an open shift is not.

For email that should have arrived: check the address on the account, check spam, and confirm your agency's reply-to address is set correctly in Account settings.

22. Getting help

Who to ask, and when

IMPORTANTCaregivers and families should always go to the agency first. Carely support cannot see your agency's data and cannot answer "why was I not scheduled Tuesday?"

The two buttons, and the difference

Both sit in the bottom corner of every screen, and using the wrong one wastes your time.

• Ask Carely — "Ask how to do something…". A help assistant for finding your way around. Answers immediately. It cannot fix anything.
• Report a problem — "Describe the problem or idea…". Files a real report with Carely, with the context needed to investigate.

Asking the assistant to fix a bug does nothing. Filing a report to find a button means waiting for an answer you did not need.

What Ask Carely knows

ALIASES: ask carely, help assistant, chat help, how do I, where is, ask a question, help bubble

It is an AI assistant, and it says so at the start of every chat and under its name.

It answers out of this manual, and it tells you which part. Every reply ends with the section it came from — "Manual § 10. Timesheets → Approving timesheets" — as a link. Click it to read the whole section, which is worth doing when your situation has a wrinkle the short answer does not cover. Underneath, Related offers the two or three sections that nearly answered you, which is often where an unusual case is actually described.

Because it reads this manual rather than a summary of it, it can answer the specific questions: why a particular visit is missing from an invoice, what happens when a payer's approved units run out, what to do when a caregiver could not clock out. It is not recalling how home-care software generally works — it is quoting the page that describes how Carely works.

Use your own words. You do not have to know our vocabulary. "I'm locked out", "sign off hours", "backdate a rate" and "forgot to clock out" all find the right section, because the manual records what people actually call these things.

It says when it does not know. If your question is not covered, it says so and points at the nearest sections instead of answering anyway. A confident answer from the wrong page is worse than no answer, so it would rather admit the gap — if you get one, the question is a good one to send through Report a problem, because it usually means the manual is missing something.

It cannot see your data. It has no access to your clients, visits, invoices or staff, so it cannot tell you whether this timesheet was approved — only how approving one works. For anything about a specific person, ask your administrator.

WARNINGThe Report a problem window says it plainly: do not include patient names or health information. Describe the problem structurally — "a client whose billing period is monthly", not the person's name. The same applies to Ask Carely.

Checking the reply

Inside the Report a problem window there is a second tab, My reports, holding every report you have sent, its status, and Carely's written reply.

The answer comes back there — not by email. You will not miss it: when Carely has answered something you have not read, the support button changes to "Carely replied" with a marker on it, on every screen. Opening My reports is what clears it — the marker goes because you looked, not because time passed.

Replying to Carely

Every report has a Reply link under it. Use it rather than filing a second report: a new report arrives with none of the history, and the person reading it has to piece the two together.

If a report is marked Resolved and the problem is still happening, the link reads "Still happening? Reply". Replying reopens the report and puts it back in front of Carely. You never have to file a fresh one to say a fix did not work.

Statuses mean:

Reporting a suspected bug

Say what you did, what happened, and what you expected. "The schedule is broken" starts a slow conversation. This does not:

"I tried to reassign Tuesday's 9am visit for the Davis client to Sam Rivera. Carely said he was already booked, but his Tuesday looks empty in the week view. Chrome on Windows."

What to send with a support report

Include:

• Your agency name
• The role of the affected user (administrator, caregiver, family)
• Which client or caregiver is involved — by their initials or record position, not full name plus health details
• The date and time of the visit or action
• The exact error message, word for word
• Your browser or device
• What you already tried
• A screenshot, with any health information cropped or covered

Never send: passwords, 2FA codes, API keys or vendor credentials, full payment card details, or health information you were not asked for. Carely will never ask you for a password.

Reporting an outage

If Carely is unreachable for everyone in your office, say so explicitly in the report — "nobody in our office can load Carely since 9:15am" — so it is triaged as an outage and not as one person's browser.

Reporting a billing problem

Include the invoice number or the subscription period, what you expected, and what you were charged. For a client invoice dispute, check An invoice total looks wrong first — most are unapproved timesheets or a missing charge rate, which you can fix yourself in minutes.

Reporting a security or privacy concern

Report it immediately and mark it clearly as a security or privacy concern. Include what happened and when, but not the affected people's details in the first message. Do not post it in a shared channel or discuss it with colleagues who do not need to know.

Practice in the demo

There is a complete working agency at carelyapp.care/demo, with invented clients and caregivers. Nothing you do there touches your real account.

Use it for: training new office staff, your first invoice run, testing a migration file, and anything with the word "delete" in it.

23. Best practices

Habits that prevent most support calls, roughly in order of how much trouble they save.

Every day

1. Check Chat for family and caregiver messages. Treat it exactly like the phone.
2. Look at today's schedule for unassigned visits, and fill them by phone, not by hope.
3. Read the Growth Center briefing. It takes a minute and tells you what changed.

Every week

1. Approve timesheets before you bill or pay. Most money problems are this one thing.
2. Resolve EVV exceptions while people still remember the visit.
3. Check credentials due to expire. Renewing before the date takes five minutes; renewing after it means someone could not work that morning.
4. Verify next week's schedule before the week starts.
5. Check availability is filled in, especially for anyone whose hours have dropped.

Every billing cycle

1. Look for clients with hours but no money before you send anything.
2. Compare the total with last period and explain any large swing before sending.
3. Set the pay period first on the Payroll screen, every single time.

Every month

1. Review who has an Administrator account, and which areas each of them can reach.
2. Review Security activity.
3. Check contact details — client billing emails, caregiver phone numbers.
4. Take a backup you keep yourself.

Always

1. Clock out before leaving the house. Forgetting is the costliest caregiver habit in Carely.
2. Remove departing staff the same day — both the caregiver record and the login.
3. One person, one account. Shared logins destroy the audit trail.
4. Read the warning before dismissing it. Carely's messages usually name the exact problem and the exact fix.
5. Train caregivers on EVV before their first assignment, including "if location does not work, clock in anyway."
6. Take a backup before anything disruptive — a migration, a bulk change, a rate revision.
7. Try it in the demo first if you are unsure.

Training a new employee

Give them the Quick Start for their role, then have them do it in the demo while you watch. Twenty minutes of supervised practice prevents months of small mistakes.

24. Quick start guides

One page per role. Learn these, and you can do the job.

Owner / administrator — what to learn first

You will use Carely more broadly than anyone else. Start here:

1. Understand the chain. Scheduled → clocked → timesheet → approved → billable → invoiced. Nearly every question you get is a step in it. → The one rule
2. Add people. They sign up with your org code; you approve and set the role. → Adding a user
3. Add clients with rates. No charge rate means a zero invoice. → Creating a client
4. Build a schedule. Read the warnings rather than working around them. → Creating a visit
5. Approve timesheets weekly. The gate everything downstream waits on. → Approving timesheets
6. Run a billing cycle. Generate, review, send. → Generating and sending invoices
7. Keep credentials current. Expiry blocks scheduling. → Adding a credential
8. Read the daily briefing. → Agency Growth Center

Know these three things before you are surprised by them: an administrator starts able to do everything and is narrowed by removing areas; nobody can type over recorded hours — corrections go through an audited EVV exception that keeps the original; and approving leave lists the visits it collides with and changes none of them until you say so.

Office staff — the daily loop

1. Check Chat for family and caregiver messages, first thing.
2. Check today's schedule for unassigned visits.
3. Fill gaps — reassign, or post as an open shift. Caregivers see open shifts in the next daily digest, which is useful for next week and useless for this afternoon: for anything urgent, post it and pick up the phone. → Open shifts
4. Approve timesheets as they arrive.
5. Watch credentials due to expire.
6. Answer family questions, escalating to your manager when it is about money or care decisions.

Scheduler

1. Create visits — client, caregiver, date, times, repeat. → Creating a visit
2. Read the two refusals. Double-booked means change the time or person. Missing credential means Certifications. → Warnings
3. Use open shifts for anything you cannot assign, and call people for anything urgent. To staff several at once, Schedule → AI auto-schedule proposes a caregiver for each open shift with the reason it chose them, and assigns nobody until you publish. → AI auto-schedule
4. Understand availability. A caregiver who saved availability once is unavailable on every day they did not mark. → The availability trap
5. Handle call-outs: reassign, release, or delete — and write a note first, because there is no cancel button. → When a visit is not going to happen
6. After approving leave, work through the conflicts it shows you. Approval lists every booked visit inside the leave and changes none of them until you choose. → Approving leave when visits are already booked

Billing staff

1. Approve first. Only approved timesheets can be billed. → Approving timesheets
2. Check for zero-amount clients — a missing charge rate. → Rates
3. Trust ready versus upcoming. Upcoming is not late.
4. Generate, review, then send until you trust your setup. → Generating and sending invoices
5. Work the aging report oldest first. → Invoice aging
6. Set the pay period before reading any payroll number. → Payroll
7. Know why a visit is missing from an invoice — the list is short and ordered. → Why a visit is not on an invoice

Caregiver — everything you need

1. Create your account with the agency's organization code, then wait for approval.
2. Sign in at carelyapp.care.
3. Fill in your availability — Schedule → My availability, all 14 days. If you save only one day, you have told the office you cannot work the other thirteen. → The availability trap
4. Find your visits — Schedule → My shifts.
5. Pick up extra work — Schedule → Open shifts, and claim. If it refuses, your credentials need updating; ask the office.
6. Clock in on arrival. Allow location when your phone asks. If location does not work, clock in anyway — never skip the punch.
7. Clock out before you leave the house. Remembering this prevents most payroll problems.
8. File the visit report — mood, signature, and a note the family will read.
9. Check your pay — Revenue Center → My Pay. Compare submitted with approved hours before calling the office; unapproved hours are usually just waiting.
10. Keep your certifications current — Operations Hub → My Certifications. An expired one stops you from being scheduled.
11. Request leave — Operations Hub → Request Leave. Give the first and last day off. When it is approved, the office is shown every visit you already have inside it and decides what happens to each one — so check with them that yours have been covered.
12. Questions go to your agency, not to Carely.

Nurse (clinical agencies only)

Everything a caregiver does, plus:

1. Keep your license current on your profile — type, number, states, expiry, NPI.
2. Take the visits you are credentialed for. You cannot self-claim one you are not qualified for; a supervisor must assign it.
3. Document in the Clinical Workspace — MAR, notes, orders — and sign your notes.
4. Treat medication flags as prompts, not decisions. → Clinical

Client / family

1. Create your account with the agency's organization code, choosing Family Member.
2. Wait for the agency to approve you and link you to your relative. If you sign in and see nothing, that link has not been made — call the agency. → Family sees nothing
3. My Loved Ones — your relative's overview.
4. Care Reports — what happened at each visit.
5. Care Plan — the care being provided.
6. Messages — the care-team conversation. Not monitored around the clock — in an emergency, call.
7. Billing — your invoices.
8. Keep the emergency contact current. You maintain it, and caregivers see it on their visit screen.

25. Training paths

Structured, in order. Each module is a section of this manual plus something to do in the demo at carelyapp.care/demo.

New agency administrator

Module 11 is the one to actually do rather than read. Watching Carely refuse to schedule someone teaches the rule permanently.

New office staff

Modules 1, 3, 6, 7, 9 — then shadow a real day.

New caregiver — before their first shift

1. Read the Caregiver Quick Start.
2. Create their account and get approved.
3. Fill in all 14 days of availability.
4. In the demo: clock in, clock out, file a report.
5. Learn one sentence by heart: "If location does not work, clock in anyway."
6. Know who to call at the office.

New family member

Send them the Client / family Quick Start. Two minutes on the phone walking them through their first sign-in prevents most family confusion.

Ready to Operate Carely checklist

You are ready when you can honestly check every line:

• [ ] I can explain the chain from scheduled visit to paid invoice
• [ ] I can add a user and set their role
• [ ] I can add a caregiver, and I know a record and a login are different
• [ ] I can add a client with both rates and a billing contact
• [ ] I can link a family member to a client
• [ ] I can create a visit, a recurring run, and an open shift
• [ ] I know what to do when Carely refuses to schedule someone
• [ ] I can explain EVV to a caregiver, including what to do when location fails
• [ ] I know only the assigned caregiver can clock out, and what to do about a stuck visit
• [ ] I can approve timesheets, and correct a wrong one by denying and re-logging
• [ ] I can run a billing cycle and check it before sending
• [ ] I can set a pay period and export payroll
• [ ] I can find an expiring credential before it blocks anyone
• [ ] I know where the activity log and backups are
• [ ] I know how to report a problem and where the reply appears
• [ ] I know the three things Carely will not do that people assume it does

18. What has changed recently

Carely is tested continuously, including by deliberately trying to break it. This section records the changes that affect how you work — what behaves differently, and why. Nothing here needs any action from you; it is all live.

The website says what is different, and has pages of its own

carelyapp.care now opens on what makes Carely different — software you can see, price and leave — and it has real pages for the questions people ask before they trust a vendor: About (who makes Carely, how to reach a person, what happens to your data), EVV by state (the clearinghouses and aggregators Carely transmits to, by state), Switching (what moves from another system and what does not), and Pricing and Security at their own addresses. On a phone the site has a menu button. The sign-up screen and the installed app now carry the same name and line as the website. The Terms of Service (version 1.2, effective October 18, 2026) gained §8A, which writes down what the pricing page promises: plans by active client, what an active client is, the free trial with no card, cancellation from inside Carely, and notice before any price change; the contact address is support@carelyapp.care. Everyone sees the new Terms once at their next sign-in.

The Privacy Policy's contact address

Questions about data Carely itself holds now go to privacy@carelyapp.care. The Privacy Policy (version 1.8, effective October 18, 2026) names that address in §10 and §14 in place of the Gmail mailbox the earlier versions listed. Nothing else in the policy changed. Because the text changed, everyone sees the new version once at their next sign-in and acknowledges it; questions about a client's health information still go to the agency, as before.

Starting a free trial

Starting a trial asks for less up front. The home page has an email box beside Start free; the address you type there is already filled in on the sign-up form. The first step of sign-up is now just an email and a password, and your agency's name and your own come first on the second step, with the care type and the agreement. Nothing else changed: no account exists until you sign on step two. At the end of the demo's guided tour, the last card now gives the starting price and says that everything you set up during a trial stays when you subscribe.

The demo on a phone

On a phone, the demo at carelyapp.care/demo now opens on a Today screen: timesheets to approve, open visits to fill and unpaid invoices, each opening its own screen, plus buttons near the top to see the agency as a caregiver or a family member. A timesheet waiting for approval can be approved right there, and the screen says where the hours went. A caregiver’s phone opens on the next visit with one Clock in button — clock in, clock out, and the hours are on the timesheet (in the demo the location is set at the client’s home rather than read from your phone). The guided tour is a short strip above the bottom bar, long menus show the section you are in with All sections to open the rest, and Start free sits in the header. To view the demo as someone else on a phone, open the account menu (your initials, top right) — View Carely as. Your agency's own Carely, and the demo on a computer, are unchanged.

The demo's pay periods

The demo's two-week pay periods now match a real agency's. Each period counts from the same Monday the server uses, so on the second week of a period the demo shows "This pay period" with its visits, instead of starting a new, empty period every Monday. In the demo, the role switcher at the bottom no longer covers the chat's message box, and it stays put between tour steps.

Office-entered clock-outs go to the state as manual

A clock-out the office enters is no longer reported as a phone GPS punch. When the office closes a visit whose caregiver forgot to clock out (EVV Exceptions), Carely records that clock-out with no location, because nobody's phone was there. CareBridge, Netsmart, Sandata and HHAeXchange were then told it was a GPS punch, at the spot where the caregiver had clocked in. It now goes as a manual entry with its reason, the way an office-entered time should. A clock-out the caregiver's phone located is unchanged.

CareBridge in Arkansas

Arkansas visits follow CareBridge's Arkansas specification. Times are written as "YYYY-MM-DD HH:MM" in UTC, the check-in distance is in miles with a reason when it is past 1/8 mile, and diagnosis codes keep their dot. Each visit carries the caregiver's own Medicaid provider ID (a new field on the caregiver's profile), the modifiers on the client's authorization (a new field on each authorization), and, for home health visits, the attending provider. The member attestation says complete only when someone other than the caregiver signed at clock-out. → Connecting your state EVV aggregator

Deleting a sent visit cancels it at CareBridge, and a claim can be voided. A visit deleted in Carely after it was sent to CareBridge is now canceled there as well. Before, it stayed at CareBridge and could still be billed. An agency that bills through CareBridge can void one visit's claim with Void claim in EVV transmissions. Aggregators that do not take a cancellation from Carely raise a bell alert instead, so the visit is canceled in their portal.

When Carely AI helps choose a caregiver

You confirm the AI's caregiver pick before it is saved. On the Clients tab, when Carely's AI chose the caregiver, Auto-match now shows its pick as a suggestion with the reason, and the client keeps whoever they had until you select Use. Without the AI, Auto-match works as before. → Letting Carely choose the caregiver

Caregivers are told when AI helped choose them. A visit published from an AI auto-schedule run, and a client match made from an AI suggestion, now say so to the caregiver: "Carely AI helped suggest you… Your office reviewed and approved it." The note goes away if the office changes the assignment by hand.

Ask Carely says it is an AI assistant when each chat starts.

These follow the updated usage policy of Anthropic, whose models power Carely's AI, from November 12, 2026.

A clock punch saved on a phone

The sign-in screen no longer announces one person's unsent punch. A clock punch or visit report made without signal waits on the phone until the caregiver is signed in with a connection. That used to be a banner on the sign-in screen, which everybody on a shared tablet saw. Now the phone tells Carely what it is holding, and Carely tells the people who can act: the caregiver gets an email saying what is waiting and the three steps to send it, the office sees it in the bell and as a chip on the caregiver's card, and the caregiver's own home shows it on every device — including "saved on another device" when they are reading on a different one — until the records arrive. Nothing else clears it.

AutoPay

AutoPay is kept on the client record and runs every morning. Turning AutoPay on — from the family portal for your own loved one, or from the office — needs a verified bank on file, and the choice is stored on the client, so the office and the family see the same thing on every device. Each morning Carely debits every enrolled balance that has fallen due, through the same Stripe collection the Revenue Center's Collect uses, once per invoice per day. A refused debit (no bank, a closed account) leaves the invoice open and raises one reminder in the bell naming the family and the reason; it is retried the next morning once fixed. Previously enrollment lived only in the browser and nothing charged unless someone pressed Run AutoPay. The billing policy "Enroll clients in AutoPay once their bank is saved" is now an agency setting kept on the server, and it acts when a bank is saved — the first moment there is anything to debit.

The billing policy is kept on the agency. Net days, the days after a period closes before an invoice is issued, automatic reminders and the late fee used to be saved only in the browser and reverted to the defaults as soon as Save returned, so every invoice was Net-14 whatever was set. They are agency settings now: saved once, validated, and read back on every device. An invoice raised by the API or by a claim gets its due date from the same policy. → Collecting what an invoice is owed

Connecting a clearinghouse or EVV aggregator

Every clearinghouse in the list connects at the address its vendor publishes, with nothing but your username and password. Availity now goes through the Availity FTP mailbox (files.availity.com, SendFiles and ReceiveFiles) — its REST address never took a batch 837. Office Ally is at ftp10.officeally.com and claim files carry the 837P keyword their sheet requires; Waystar's host, folders and .CLP file name and TriZetto's Gateway EDI host and folders are filled in for you. All four now read their response folder the way Claim.MD's API is read: a 277CA marks each claim accepted or rejected with the payer's words, a 999 rejection raises an alert naming the fault, and every 835 is downloaded and reconciled — on Check now and every quarter hour. → Connecting your clearinghouse

HHAeXchange connects through its Aggregator API. The form asks for the Client ID and Client Secret HHAeXchange issues, your nine-digit tax id and the Payer ID from your state's specification; Sandbox uses HHAeXchange's Implementation environment and Production its live one, so nothing has to be typed. Each visit goes out in the aggregator's own record — member by Medicaid ID, caregiver by worker ID, UTC times, both calls with GPS — and the aggregator's verdict is read back onto the transmissions card.

Netsmart (Tellus / Mobile Caregiver+) connects through its Rendered Services file. Netsmart publishes its full third-party specification per payer, so Carely now writes each completed visit in that layout to the SFTP folder Netsmart assigns and reads the verdict back from the response folder — for Florida's Aetna, Anthem and UnitedHealthcare plans, Georgia, Montana, Nebraska's four payers and Virginia. Netsmart issues its account to the certified vendor, so Carely holds it: the form asks only for your state and payer, and the agency identity comes from Billing provider. Pennsylvania's Highmark Wholecare posts no specification and is not offered.

Vesta is no longer in the aggregator list. Vesta (DataLogic) is an EVV vendor, not a state aggregator — in Texas the aggregator is HHAeXchange under TMHP, which a Texas agency connects to directly.

CareBridge connects through its published state specifications. CareBridge publishes its third-party vendor contract per state, so Carely now writes each completed visit in that state's column layout to CareBridge's SFTP folder and reads the verdict back from the response folder — for Arkansas, Iowa, New Jersey, North Carolina, Tennessee and Wyoming. The account is Carely's, credentialed by Carely's SSH key; the form asks only for your state and health plan.

Security and privacy, tightened across Carely. A full audit found places where an account could reach more than its role should, and each is now closed:

• Only an administrator records a nursing license on their own account; a caregiver's license is recorded on their caregiver profile by the office, and a family login can never carry nursing authority.
• Caregivers' credential files (licenses, IDs, TB results) are seen by the office, by the caregiver themselves, and by a family only for the caregivers on their loved one's care team. Only the office or the caregiver themselves can add one.
• Filed state forms and the medication alerts are no longer readable outside the people allowed to see that patient.
• A change of role takes effect immediately, and an administrator who is demoted loses administrator access at once. Access to another agency (a sub-unit, or an enterprise location) is checked on every request, so revoking it takes effect immediately.
• A caregiver can update a family's phone number, but only the office changes the family's email (it receives their password resets).
• An agency can only set up an email sending domain that is its own.
• Canceling an account deletion puts the agency back exactly as it was before.
• Only caregivers claim open shifts, and approving an account only applies to accounts waiting for approval.
• A family's bills and loved ones are matched to their account, not just their name.

Delete your own account, from inside Carely. Caregivers, family members and administrators can now delete their own account under Account settings → Delete my account. Before, they had to ask the agency or email Carely. The records of care stay with the agency, and the agency is told. → Deleting your own account

Privacy Policy 1.7: phone notifications. The Carely phone apps can now send notifications, delivered by Google (Firebase Cloud Messaging) and Apple. A notification only says that something is waiting, such as "You have a new message in Carely." It never includes a name, a health detail or what a message says; you read that inside Carely. Everyone is asked to acknowledge the updated policy at their next sign-in.

Report a message, block a person. In Chat, anyone can report someone else's message to the office, with a reason, and block a person. Blocking hides that person's messages from you and stops one-to-one messages either way. Administrators review reports in Chat → Reported messages and remove the message or dismiss the report; the office is also told when someone is blocked. Chat also refuses slurs and abuse before they are sent. App stores require all three in any app where people message each other. → Reporting a message

Billing is view-only in the phone apps. In the Carely iPhone and Android apps, Account settings → Subscription & plan now shows your plan and its status, and the Enterprise Portal's Billing tab shows the enterprise subscription's status. Subscribing, switching between monthly and annual, updating the card, invoices and canceling are done in Carely on the web, where nothing has changed. App stores require their own payment system for anything bought inside an app, and Carely is sold to agencies directly.

Dashboard figures stay on one line on a phone. A large amount, such as revenue collected, used to break across two lines in the middle of the number on a phone. It now shrinks to fit its card.

Typing on a phone no longer zooms the screen. On an iPhone, tapping a box to type in it zoomed the whole page in, and it stayed zoomed with the right side cut off. Boxes on a phone or tablet now use slightly larger text, which iPhones leave alone. The username box also no longer starts with a capital letter or gets autocorrected, and your phone's password manager can fill in your sign-in.

A mistyped password no longer signs you out. Entering the wrong current password when changing it, deleting the agency, or turning two-factor sign-in on or off used to sign you out completely. Now Carely just says the password or code is not right.

Medicaid payments add up the way payers send them. Every payment line on a remittance is now kept for its claim, and the claim and its invoice follow the total:

• A payer's correction — taking a payment back and paying the right amount, often in a later remittance — is applied. The claim shows the corrected amount, the invoice shows what is still owing, and you are told.
• A payment taken back with nothing paid in its place reopens the invoice for what was taken back, rather than leaving it paid.
• A claim paid on appeal after a denial is paid, and its invoice is reopened and settled.
• The same remittance received twice is counted once.
• A payment beyond what the claim is for is flagged as an overpayment; expect the payer to take the difference back.

Smaller fixes. Signing out now starts a fresh page, so on a shared tablet nothing of the last person stays behind. A caregiver renamed by the office sees their own shifts under the new name without signing out. Tapping a chat notification opens that conversation, including a group chat. Carely reads its AI request limit from Amazon each hour, so a raised limit is used without anyone changing a setting.

Permissions and nursing scope, closed where they were open. A fifth pass of the audit checked who can reach what on the server:

• Taking Clients, Caregivers, Staff or Settings away from an administrator is now enforced on every route that holds them. Settings covers backups, the full data export, the email sending domain and Carely's own subscription. Before, the screens followed the restriction but several parts of the server did not.
• A nurse adds medications, records doses, writes notes and writes or signs orders only for the patients they are assigned to, the same patients they can read. An order must be for a patient on the roster.
• An order that a renewal replaced, or that was discontinued, can no longer be signed or renewed again.
• A caregiver's team is matched on their exact name, so unusual characters in a name can no longer match other people's teams.
• Renaming a caregiver renames that caregiver's own sign-in only, never a family member or administrator who shares the old name.
• Restoring a permanently deleted client from the archive brings back their authorizations, the hours used against them and their rate history. Before, it brought back only the record and contacts.
• Retry on a claim sends it again only when it failed to send or the clearinghouse rejected it. A claim still at the clearinghouse, or one the payer denied or reversed, is not resubmitted unchanged, and Carely says what to do instead.

Saving waits for Carely to confirm, and nothing you typed is lost. A fourth pass of the audit went through the everyday screens:

• Clock out and Clock in wait for the punch to go through, so a second tap can no longer send a second visit report to the family. The hours in "Visit completed" are counted from the real clock-in, including overnight visits, which used to read 0.
• A caregiver's own shifts, timesheets, pay and clients are found even when their name was saved with different capital letters or spacing.
• Recording a client as Deceased or Archived asks first and cancels their upcoming visits, as Discharge does.
• Adding a client, a caregiver or a shift, claiming an open shift, and importing EVV visits say "done" only once Carely has saved it. If it is refused, the form stays open with what you typed. A new caregiver's bank details are saved only once the caregiver is.
• A chat message that could not be sent goes back in the message box.
• My availability shows your saved days even when they load after the tab opens, and saves only the days you changed. It used to be able to mark every other day unavailable.
• Pay caregivers says how many payouts failed and why. It no longer says "Payouts sent" when none were.
• An administrator's own hourly rate is kept after saving; it used to return to $25.
• My Pay works out tax for the period shown (weekly or every two weeks). When the newest visit is not from this week, it says which week it shows.
• Editing one month of a recurring cost no longer moves the start of the whole series.
• A client's location cannot be confirmed with empty boxes; that used to save it as 0, 0.
• Deleting a visit from the schedule asks first.
• Large agencies see every client, caregiver, timesheet, invoice and certification. Lists used to stop at 500 without saying so.

Status screens show green only when it is. A third pass of the audit checked every place Carely reports that something is fine:

• A screen whose information could not be loaded now says so. It used to show its "all clear" state, such as "No backups yet" for an agency that had backups. The same applies to messages that were never shown: about thirty screens displayed "Nothing here yet" while loading, after a failed load, or instead of their own explanation. Each now shows its own message.
• Off-site backup says when the last encrypted copy actually arrived, or that it failed. It used to say "On" whenever off-site storage was set up.
• A Medicaid connection reads Healthy only on a successful test from the last seven days. Carely re-tests a live connection weekly by itself. The built-in simulator is labeled Simulator, never Live.
• Actions the Copilot agents prepared for your approval expire after seven days and leave the queue. They no longer stay listed when clicking them only says "expired". A reminder the mail service refused is not recorded as sent, and it is tried again on the next run.

A branch needs the parent's approval. Your agency code is the code every one of your caregivers and families signs up with. Entered as Sub-unit of another agency, it used to attach a branch to your agency at once — $19/month on your bill and its clients counted toward your plan — without asking you. It is now a request: you get a notice in the bell, and you approve or decline it under Operations Hub → Sub-units. Until you approve, the branch is locked, nothing is billed, and it is not covered by your Business Associate Agreement. → Sub-units

Pay runs and payments, made exact. The same audit went through every place money moves:

• A pay run can no longer be locked from a draft that has gone out of date. If a timesheet in it was denied, or its hours were corrected, after the draft was built, Lock says so and asks you to rebuild the draft, so a caregiver is paid for the hours that were approved.
• A pay adjustment made after a run was locked (an EVV correction, a back-pay amount) is paid on the next run, once. It used to be left out of every later run.
• A timesheet that has been paid can no longer be denied or sent back for resubmission, and hours a caregiver enters by hand are always marked as entered by hand.
• An EVV correction that needs a second approver: if the second person changes the times, it becomes a new proposal waiting for someone else, rather than being approved under the first person's name. A correction's pay adjustment is priced at the caregiver's own pay rate.
• Every bank debit — AutoPay, Collect and the family's Pay — is recorded in the payments ledger against its invoice. A debit the bank returns days later reopens the invoice and puts a "payment returned" notice in the bell; a second debit for the same invoice can never be mistaken for the first.
• The nightly money check and the automatic backups run agency by agency, so a problem in one agency no longer stops them for the others. An agency with caregivers or visits but no clients yet is now backed up too.

A family's Pay button now really pays. It used to mark the bill paid without collecting anything. It now debits the family's saved bank, and a declined debit leaves the bill open everywhere — AutoPay, Collect and Pay. → Collecting what an invoice is owed

Alerts and the error log are honest. An alert email the mail service refused is no longer counted as sent; it is tried again on the next pass. Errors from browser extensions and signed-out sessions are no longer logged as Carely's own.

No more leftover strip when you switch sections. Going from a long section scrolled part of the way down (Clients & Matching) to a short one (Caregivers, in an agency with a few caregivers) could leave a band of the previous screen showing at the bottom of the window, in Chrome and the desktop app. The page now keeps its place while it switches: you either stay exactly where you were, or, if you had scrolled past where the new section starts, it is brought up under the top bar.

The AI briefing says it is working. In the Agency Growth Center, AI briefing and Full report used to turn into "…" while Carely AI wrote, with nothing else on the page changing. Now the button shows a spinner and Writing…, the summary card says Carely AI is writing your briefing (and, after 20 seconds, that it is still working and nothing is wrong), and the new briefing replaces the old one when it is ready.

Every confirmation is now Carely's own. The last of the browser's plain pop-ups are gone: restoring or deleting a backup, signing out your other devices, removing a tax ID from your invoices, and using a clock-in's location as the client's location now ask in Carely's own dialog, with buttons that say what they do. A confirmation opened from inside Settings now covers the whole screen instead of sitting inside the Settings window, and closing it leaves Settings open.

Forms that were hard to read in dark mode. The boxes you type into on the EVV exception, leave dates, authorization and troubleshooting forms, and in a few places on Schedule, were styled with colors Carely does not define, so in dark mode they had no background of their own. They now use the same colors as every other form. Confirm this location on a client's EVV location, and deleting a website visitor's record, now ask in Carely's own dialog rather than the browser's plain pop-up, like every other confirmation in Carely.

Invoices are organized the same way. The invoice list now opens on Unpaid, grouped by when each invoice is due — overdue first, the most overdue at the top — with what is still owed totaled in each heading. Paid, Void and All go by the month the invoice was issued, newest first, with older months folded, and a search finds a client, family or invoice number. Families see To pay first in their Billing tab, and Payments are grouped by month. → How the invoice list is organized

Timesheets are organized. The Timesheets list was one long list, with what needed approving mixed into months of paid visits. It now opens on Needs approval, grouped by caregiver, with whoever has waited longest first, each caregiver's hours and pay totaled, and one Approve for all of a caregiver's visits. Visits from an earlier pay period are flagged. Everything else is grouped by pay period, newest first, with a search for a caregiver or client. Caregivers see My Timesheets by pay period and are told first when one is sent back. On My Pay, visits now run newest first. The Pending Approval count on Payroll counts only timesheets that are waiting; it used to count paid and denied ones too. → How the timesheet list is organized

Editing someone at the bottom of a list brings the form to you. Choosing Edit on a client, or opening a caregiver's profile and then Edit profile, used to open the form at the top of the page while you were still looking at the bottom. Carely now scrolls the form into view, just below the top bar, on a computer and on a phone.

Kentucky is a known state. An agency in Kentucky now starts from Kentucky's rules instead of the federal ones: EVV through the Therap aggregator, KARES and the state registries for background checks, Personal Services Agency certification renewed every year, the K-4 withholding form, and unemployment tax on the first $12,000. Overtime follows Kentucky's seventh-day rule. → Setting your own overtime rate

Carely watches its own AI. Every request to the model records how it went — which feature asked, how long it took, and whether Amazon's per-minute limit made it wait or turned it away — never what was in it. If that limit starts slowing people down or failing their requests, Carely emails the platform operator right away with the number to ask AWS for, so a slow document upload is caught without anyone having to report it. Records are kept for 30 days. → Where Carely's AI runs

Privacy Policy 1.6 and Terms of Service 1.1. The policy now says Carely's AI runs on Amazon Bedrock under the AWS Business Associate Addendum, and says exactly what each AI feature sends — including that reading an uploaded document sends that document's text. The terms add that a document Carely reads, and a caregiver it suggests, are for a person to check. Everyone is asked to accept the new terms and acknowledge the new policy the next time they sign in.

Carely's AI now runs on Amazon Bedrock. Reading intake packets, ranking caregivers, the auto-scheduler, the morning briefing and Ask Carely now run inside AWS, under the Business Associate Agreement that already covers Carely's hosting and backups, rather than through the model vendor's own service. While AWS opens its newest models to Carely's account, these features use the Claude models AWS has made available; nothing changes in how you use them. → Where Carely's AI runs

Edit a message you sent. Point at your own message and choose Edit message; Enter saves and Esc cancels. Everyone sees the new words marked Edited. Only the sender can edit, and Carely keeps the earlier wording for the agency's records. → Editing a message

Group chats, adding people, and deleting a message. New chat now starts a chat with the people you pick — one person, or a named group, optionally about a client. In a group, the people button adds people (you choose whether they see earlier messages), removes them, renames the chat, or leaves it. Add people on a care conversation starts a group with the people already in it. You can delete your own messages, and the office can delete any; the chat shows This message was deleted and Carely keeps a copy for the agency's records. Messages sent before encryption was added are now encrypted where they are stored too. → Group chats, and adding people · Deleting a message

Chat: every conversation in one place, laid out like Microsoft Teams. Messages used to live in four places with different layouts: Communications and Announcements in Operations Hub, and Messages and From the Office in Family Connect. They are now one app, Chat, for administrators, caregivers and families. Your chats are listed on the left with unread counts, the conversation is on the right, and Enter sends. New chat starts a conversation, and announcements are a pinned channel at the top. Who can read what is unchanged, and every chat says so at the top. When a family or caregiver writes, the office or caregiver on the other side now gets a phone notification, with no name or words in it. Message text and photos are now encrypted where they are stored, and a busy agency's chats always show the newest messages. The delivery log of emails Carely sent is under Operations Hub → Sent emails.

My Pay keeps counting a visit after it is paid. Locking a pay run marks its visits Paid, and My Pay counted only Approved visits, so a caregiver's period dropped to $0 the moment their pay went out. Paid visits now count, and are listed as Paid with their mileage.

Visits already paid to the caregiver are billed to the family. Locking a pay run marks its visits Paid, and the invoice worksheet used to skip them, so a visit paid before its invoice went out was never billed. Paid visits now appear on the worksheet like any approved visit.

Mileage checks. A trip over 200 miles, or a visit with over 400 miles of trips, is refused as a likely wrong place. Typed miles, or over 100 miles on one visit, hold the visit for a person to approve. A caregiver can change trips only while the visit is Pending or Denied, and nobody can once it has been paid. A rate change or mileage switch dated in the future now takes effect on its day without anyone saving the client again.

Privacy Policy 1.5. The policy now says that mileage trips a caregiver logs are kept with the visit, and lists Amazon Location Service, under the same AWS Business Associate Addendum, as where those places are looked up. Everyone is asked to acknowledge the new version the next time they sign in.

Mileage is the trips the caregiver drove, logged at clock-out. ZIP code to ZIP code is gone. When a client is billed mileage, the clock-out panel asks about driving during the visit, and the caregiver lists each trip From → To. Each place is picked from suggestions as it is typed, and Carely measures each trip in a straight line. The miles go on the family's invoice and are paid back to the caregiver at the IRS rate as a non-taxable reimbursement, kept separate from gross pay. The office sees every trip in a new Mileage column on Timesheets, and can review, edit or correct it while approving. → Mileage

Bill by the visit, and bill mileage. A client can now be billed — and its caregiver paid — per visit instead of by the hour, chosen on the client's Care rates card and switched on a date like any rate, so earlier visits keep the terms they were delivered under. Mileage can be switched on per client, from a date, and is priced at the IRS standard mileage rate for each visit's date, read from irs.gov daily. Invoices now list each visit and each mileage charge as its own line, and the family sees those lines. → Rates: charge and pay · Mileage

Availity's and Office Ally's own reports now reach the claim. Availity's IBR, EBR and DPR reports and Office Ally's File Summary and EDI Status reports are read line by line, so a claim either of them rejects shows rejected with the vendor's reason instead of only a note in the bell. New claims get shorter reference numbers (ten characters) so every report can name them in full.

Claims and EVV connections were checked against what every vendor publishes, and corrected. Each clearinghouse's companion guide, each aggregator's specification, and real sample files were compared with what Carely sends and reads. What changed for you:

• Billing provider now asks for the Medicaid payer ID from your clearinghouse's payer list, and every claim carries the client's home address and a diagnosis code.
• Claim files are addressed the way each clearinghouse asks, and each has its own control number. Every acknowledgement and status the clearinghouse returns is read; a file it throws back marks its claims rejected with the reason.
• Optum claims now come back with statuses and remittances, like the others.
• Sandata asks for your Account ID and state Medicaid provider ID, plus the payer and program under Advanced Setup, and sends caregivers and clients before their visits.
• Caregivers have a new State EVV worker details group: birth date, gender, hire date and the state license or registry number, which HHAeXchange requires.
• Send test works only on Sandbox. Visits completed before your aggregator is live wait, and go the moment it is.

The bottom of a phone screen is no longer under the help buttons. The floating "Ask Carely" and "Report a problem" buttons sit above the tab bar, and a page's last rows could end up beneath them, where a tap hit the button instead of the row. Every screen now scrolls clear of them.

A saved record on a shared device is nobody else's banner. When a phone or tablet held another caregiver's unsent punch, whoever signed in next saw a banner naming that person across the top of every screen, covering the avatar menu. It is gone: the caregiver it belongs to is told by email and on their own home, and the office by the bell. A caregiver's own "saved on this device" band now sits above the header instead of over it.

What happens after CareBridge accepts a visit now shows in Carely. Carely reads CareBridge's daily Appointment Status and Appointment Errors reports: a visit's claim status (paid, denied, and the steps between), the payer claim number and billed amount appear under the visit in EVV transmissions; an error CareBridge finds later turns the visit rejected and tells the office; and a rejected visit now has a Re-send button.

AuthentiCare and Therap say "certification pending". These aggregators issue their interface to an EVV vendor only under an integration agreement, which Carely has not completed. They used to accept an endpoint and show as connected while nothing they would accept was ever sent; now the form says so before you type, your credentials are kept encrypted, the test reports certification pending, and no visit is transmitted until Carely is certified with that aggregator. → Connecting your state EVV aggregator

Scheduling a night shift

Overnight visits can be booked as one visit. A waking night or live-in shift running 9:00 PM to 9:30 AM used to be refused outright — "End time must be after start time" — so the only way to record a night was two visits on two dates, which split the hours across two days and made overtime and leave awkward. Now you enter the evening start and the morning end, and Carely reads it as one visit crossing midnight, confirming the span under the form before you create it.

Everything downstream follows the night rather than the calendar day: the hours count the whole span, double-booking is caught across midnight in both directions, a handover at exactly the end time is still allowed, and nobody is chased for running late while the night is still in progress. Visits imported from an EVV vendor get the same treatment — a night imported from a vendor used to come in as zero hours and pay nothing. → Overnight visits

Times are set in AM and PM. Everywhere a time is typed, it is now three dropdowns — hour, minutes, and AM or PM — instead of a 24-hour box: booking and editing a visit, Log time worked, requesting part of a day off, recording the dates on someone's leave, and both timesheet period forms. Everywhere a time is shown reads the same way — the schedule, the week grid, the caregiver's own shift list, and the double-booking refusal, which used to come back reading "already booked 21:00–09:30".

The time box used to render in whatever format the computer's language settings chose, which on many machines is the 24-hour clock, so the same screen looked different on different desks. What Carely stores has not changed, so nothing you have already recorded or exported is affected.

Security

A sign-in is one browser tab's, not the whole browser's. Signing in used to sign in every tab of that browser: anyone opening carelyapp.care in a new tab on the same computer was inside your agency with no password asked. A sign-in now belongs to the tab it was made in — a reload keeps it, a new tab or window asks for your password. The installed desktop and phone apps are unchanged. Everyone's existing browser session was signed out once by this change. → A sign-in belongs to one browser tab

Small things that were wrong on screen

The Privacy Policy is at version 1.3. It was brought in line with what Carely does today. It now says where a clock-in location comes from on a computer, and that location is read only at clock-in and clock-out; lists the information an agency adds through its own intake questions, uploaded documents, messages and the clinical module; describes the record kept of each agreement you accept; says what the sign-up form and the demo record about a visit — which step and which fields, by name, never what was typed; and says where a password-reset link is sent. Everyone is asked to acknowledge the new version at their next sign-in. → /privacy

Version 1.2 of the Privacy Policy introduced the website's visitor identifier. The public website and live demo now keep a first-party visitor identifier so Carely can see how people find and explore it — which pages and sections are viewed, whether the demo is tried, whether a visit leads to an account. It is one cookie set by Carely's own server; there is no third-party analytics, no advertising cookie, no email tracking pixel, and nothing typed into the demo is recorded. A browser sending Global Privacy Control gets no persistent identifier, and "Forget my browser" in the site footer deletes the record. The policy's §2B, §7, §9 and §10 say exactly what is kept and for how long. Everyone is asked to acknowledge the new version at their next sign-in, as with any policy change. → /privacy

Carely opens even when the browser refuses to remember anything. A private window, a full storage quota, or a "block site data" setting made the app stop on its first screen with "Something went wrong" — a theme preference could not be saved, and the whole page gave up. Preferences that cannot be remembered are now simply not remembered; the app opens and works.

Every external system connects with the credentials your vendor gave you. Several vendor connections could never have worked: Sandata was addressed at a path that answers "not found", Optum's token was requested in the wrong format, an OAuth clearinghouse read its own valid token as "no API at that address", Office Ally, Waystar and TriZetto had invented web addresses when they take claims over SFTP, and Epic was asked for a secret it does not issue. Each adapter now speaks its vendor's documented interface — Sandata Alt-EVV v1.1 with the member's Medicaid ID and the worker's identifier on every visit, Optum's Medical Network, Availity's token server, HHAeXchange's published API, SFTP for the file clearinghouses, and SMART Backend Services for the EHRs (Epic's signed key, Oracle Health's and athenahealth's secret). The connect forms say where each credential comes from, ask up front for an address your vendor issues per customer, and test on save with the vendor's own words. An endpoint typed under Advanced Setup on the Medicaid tab used to be sent as a credential and ignored; it is sent as configuration now. → Connecting your clearinghouse · Connecting your state EVV aggregator · Clinical integrations

A nurse can add a medication. The medication record had no way to put a medication on it — the empty state pointed at a "clinical profile" that did not exist — and the patient pickers for notes and orders were built from the medication list, so a new agency had nobody to choose. Nurse Station now has Add medication (coded to RxNorm as you type, with hold and discontinue beside each medication), and every patient picker offers the active roster. Signed notes and added medications reach a connected EHR within seconds instead of at the next five-minute sweep, as do EVV visits and claims. → Adding a medication

City and state are picked, not typed, on the Medicaid billing provider. A city typed by hand arrived in four spellings and the payer's enrollment file holds one of them. Both are now dropdowns — the city list follows the state — with Other city… for a town the list does not carry, which is capitalized on save. The ZIP is checked against the state.

The live demo opens on the numbers, with a guided tour you can follow or leave. The practice agency at carelyapp.care/demo used to open on Operations Hub's full menu of sections, with the tour in a corner and the View Carely as bar across the middle. It now opens with the tour card across the top of the screen and the agency's figures directly under it. Each step of the tour names one thing to try — Try it: Press Timesheets… — and outlines that control for you; press it and the card confirms it.

While the tour is showing, a long menu on the screen it is describing is folded down to the section you are on, the one the tour suggests, and All sections. Select All sections to open the whole menu, or Explore on my own (or Skip) to dismiss the tour — the full menu and the View Carely as bar come straight back. During the tour the persona switcher is the small Demo button at the bottom of the screen; select it to switch between Admin, Caregiver, Family, Enterprise and Parent agency at any time. None of this appears in your own agency's Carely: it is the demo only.

The demo agency's month in progress is billed weekly. In the demo, Revenue Center → Expenses → This month used to show the practice agency thousands of dollars down for most of every month, because its sample invoices were all dated late while rent and wages landed early. The demo now bills the current month a week at a time — on the 1st, 8th, 15th and 22nd — so This month reads the way a healthy agency's does: at most a little under in the first week of a quarter, and in profit from the second week on. Completed months, the invoices that show as owed or overdue, and every other demo figure are unchanged.

The live demo has one cast of characters. Screens in the demo named caregivers and clients who were nowhere on the Caregivers or Clients tabs — a pay run for "Jamie Fox", a claim for visits by "Sarah Chen", contacts for "Robert Chen". Every person a demo screen names is now someone you can open on the roster, and the demo's registered nurse, Nancy Okafor, is on the Caregivers tab alongside the aides she works with. The sample Medicaid claim also follows the client you pick rather than always showing Mary Johnson's.

Quieter screens for caregivers and families. Every sign-in loaded a handful of screens' worth of data that only an administrator may see, and the server refused each request — no harm done, but wasted and noisy. A caregiver's or family's tab now asks only for what it can show. Two similar requests — an owner who is not a caregiver being checked for a payout account, and a non-clinical agency's family being asked for clinical records — no longer happen either.

The bell can be cleared. There was no way to put a notification away: most of the list is worked out from live data, so nothing could be marked read and the count only ever went up. Each item now has a ✕, and Clear all empties the list; a dismissed item returns only if what it says changes. And a standing condition — an open incident, a visit note still missing — is one item, not a fresh copy every morning, which is how a bell reached 22. → Notifications

Every state box is a list now. The State box on a client's service address and a caregiver's home address is the same list of states by name as everywhere else — the wrong state cannot be typed into it. The state and the ZIP must still agree.

Licensed states are picked, not typed. Clinical activation and the Add nurse form asked for states as "comma separated, e.g. FL, TX" — you had to know the postal codes, and anything that was not exactly two capital letters was dropped without a word. Both are now a list to choose from, one state at a time, each shown as a chip you can remove.

The Director of Nursing is a licensed person, not a name. Activation asked for a name and checked it against nothing. The DON is now picked from your nurse roster — their license on file becomes the license of record — or named with a nursing license number. An agency that has said it provides clinical care can add nurses before activating, so the DON can be on the roster when activation asks. → Turning the clinical module on

You can see which app you are in. The highlight on the current app in the left-hand list was a tint so close to the list's own color that on a bright screen there was no highlight at all — the only clue was the breadcrumb. The current app now has a filled row, a bar down its left edge and a bold name, in both light and dark mode. The tabs inside each app are marked the same way — filled, bold, with the edge bar — and on a phone the tab you are on is scrolled into view rather than left off the edge of the strip. The Week / List switch on the Schedule shows its choice in the same color instead of a faint shadow.

Every screen now hears about changes as they happen. Until now an open tab caught up with the agency on a timer — every minute and a half, and whenever you looked at the tab again. Now each signed-in tab keeps a connection to the office, and a change made by anyone in your agency — a clock-out, a message, a leave request, an approval — reaches every other open screen within a few seconds, without a reload. When the connection cannot be made, the timer is still there underneath. → Screens keep themselves current

Communications kept up to date. The office's conversation screen took one look at the messages when it opened and did not look again. You typed a reply, the box cleared, the toast said "Message sent" — and your reply was not in the thread until you reloaded the page, so it was easy to send it twice. A family's answer arriving while you had the screen open did not appear either. The screen now follows the conversation: your reply is there the moment it is sent, and new messages arrive on their own while you are looking. → Chat

Four places were showing a dot instead of an icon. The State Forms tab in the Operations Hub, the Sign action on a state form, the Edit billing details button, and the screen shown when the connection drops. Each was asking for an icon by a name that does not exist, and an unknown name draws a small dot — which looks deliberate rather than broken, so they sat there unnoticed. All four now show the icon they were asking for, and the build refuses any icon name that is not defined, so a typo cannot quietly become a dot again.

Your agency's own forms

You can upload the forms your agency files. State Forms now opens with Your agency's forms: upload a PDF or a scan of anything you have to file — a state document, a county addendum, a payer's packet — name it, optionally tag a jurisdiction, and it is kept encrypted on your agency's record to open or print whenever you need it. Previously the screen only offered forms Carely had pre-built, so anything else meant keeping it somewhere outside Carely.

The screen also no longer lists which other states Carely holds forms for. Being told that forms exist for somewhere you do not operate is not something you can act on. → State forms

Reading the pay run

The Pay run card now says which days it covers. It is built for the pay period that is closing, while the Payroll run panel below it is built for whatever dates you type in — so the two hour totals can legitimately differ, and with nothing naming the periods the only available reading was that one of them was wrong. Given the same dates, the two agree exactly; where they differ, it is because hours already paid in a locked run are not counted a second time.

You now decide where paid leave comes out of, per category. Operations Hub → Leave → Leave allowance & accrual has a setting under each of the five categories: paid leave can come out of that category's one allowance, as it always has, or out of a separate paid allowance you set alongside it — so "80 hours of paid annual leave, unpaid beyond that" is something you can actually run.

Where a category holds two, the pay decision picks the pot: unanswered or refused draws the general allowance, approved draws the paid one, and changing the answer moves the hours back. Changing the setting itself re-files leave you have already approved, and says how many caregivers it rebalanced. Every category starts on the shared allowance, so nothing moves until you change it. → Where paid leave comes out of

Leave is asked for in days now, and you set how long a day is. A full day is 8 hours unless you change it — Operations Hub → Leave → A full day of leave is __ hours. The request form gives caregivers day buttons and shows what it worked out, so nobody multiplies in their head. Part-day leave still records its own hours.

Approving pay past someone's allowance is now refused, with an override. It used to be allowed with a warning, and payroll would pay it. Carely now stops it, says what is available and by how much the request is over, and offers Approve the pay anyway — logged against your name. Their time off is unaffected, and a category with no allowance set is never refused.

Two things the balance warnings were getting wrong. Where a category has a separate paid allowance, the over-balance warning was measuring the general one — so a request that was about to empty an 8-hour paid pot could draw no warning at all if the general allowance had room. And the warning only showed while the request was awaiting its time-off decision, so it had disappeared by the time anyone pressed Approve pay — the click that actually commits money. The pay decision now shows what that allowance has left and how far past it you are going, and says so for as long as there is a decision left to make.

Leave

Caregivers can ask for leave to be paid, and you answer that separately. The request form now has Ask for this to be paid leave. In Operations Hub → Leave the request carries its own row with Approve pay and Deny pay, apart from approving the time off — so "take the week, but it comes out unpaid" is something you can actually record, which one Approved/Denied could never say. Refusing asks for a reason, and the caregiver sees it beside the decision in their own My requests list along with the time-off answer. → Paid leave — a second, separate answer

And approved paid leave is now actually paid. The hours go onto the pay run for the period the leave falls in — no adding them to a timesheet by hand. The caregiver's line shows the worked hours and the paid leave separately, both exports carry their own columns for it, and locking the run claims the leave so a later period cannot pay it twice. Leave that straddles two pay periods is split across both.

Paid leave is paid at their average hourly rate, and is not counted toward overtime — overtime is owed on hours worked, so 36 hours worked plus 8 hours of holiday is 44 hours at the normal rate, not 40 plus 4 at time-and-a-half. → How approved paid leave gets paid

"Record the dates" now shows the dates it recorded. Saving them always worked, but the Leave screen could not see what it had saved, so the card came back reading "Dates not recorded" exactly as before and the button looked broken. The dates appear the moment you press Save.

Leave balances are spent by the approval itself. The hours used to come off in the browser, in a separate step after the decision — so leave approved any other way never touched the balance, and if that second step failed the hours were never spent, with nothing recording that they were owed. It is now part of approving. Three things follow that were not true before: approving twice does not double-spend, taking an approval back returns the hours, and correcting an approved request moves the difference. → What approving does to the balance

Getting help

Ask Carely now answers out of this manual, and cites the section it used. It used to answer from about twenty short summaries written separately from the documentation, so it could tell you where the Timesheets screen was but not why a visit was missing from an invoice, or what happens when a payer's approved units run out — both of which this manual explains. It now searches the whole manual, quotes the passage that answers you, and links the section so you can read the rest. When your wording is not ours — "I'm locked out", "sign off hours" — it still finds the right page. And when the manual does not cover something, it says so instead of answering from the nearest page. → What Ask Carely knows

Knowing whether an email arrived

You can now test your email setup without emailing a family to find out. Account settings has a "Send a test email to myself" button: one message to your own address, and if the provider refuses it, its exact words are shown. Previously the first sign that sending was broken was a relative who never heard from you. → Testing your email without emailing a family

The "Email a family" compose box has been removed. It let you type a subject and body to one family, which duplicated the care-team conversation sitting directly above it — its own description told you to use the conversation for anything back-and-forth. Two places to write to a family means two places to look for what was said, and only the conversation keeps it with the client's record where the whole team can see it.

Nothing else changed: invoices and payment reminders still go out by email from Revenue Center, and Sent messages still logs every one with its delivery status.

A family email that was refused no longer says "Sent". The badge in Sent messages was showing whether email is set up, not whether that message left the building — so messages the provider had rejected appeared in green, in a log headed "every message sent to families". An agency reading Sent believes the family was told, which is worse than seeing an error, because nobody goes looking. The badge now reports what happened to that message, and a failure prints the provider's reason underneath it. → Checking an email actually arrived

Addresses, and the devices you can clock in on

An address is now typed in separate boxes and written one way. Street, city, state and ZIP are their own fields for caregivers as well as clients, and Carely standardizes what you type — "123 north main street" and "123 N Main St." are stored as the same address instead of as two different places. This matters beyond tidiness: EVV checks a caregiver against the client's address, and a claim carries it as the service address. A state spelled out in full is now accepted and shortened for you rather than refused. → Adding a client

And the client's address is now asked where you can see it. It used to sit inside a collapsed section headed "only needed if you bill Medicaid or an insurer", so an agency that does not bill Medicaid never opened it — while a separate intake question collected the same address as one free-text line. The address EVV anchors to was therefore missing on most clients. It is now four boxes under Where care is given, always shown, and the duplicate question has been removed. Any address already sitting in that old intake answer has been moved onto the client for you.

A state can be typed as a name, and lands as the right code. The State box used to keep the first two letters of whatever was typed. That is right for Florida by accident and wrong for most states — and two of them were wrong silently: "Arizona" became AR, which is Arkansas, and "New York" became NE, which is Nebraska. Both are real codes, so nothing questioned them and the record was filed in the wrong state. Type the code or the name; either lands correctly.

And an address whose state and ZIP disagree can no longer be saved. Every US ZIP belongs to one state, so Carely checks: a Phoenix ZIP with Arkansas in the State box is refused, and the message names the state the ZIP belongs to so you can see which of the two boxes is wrong. It only refuses a definite contradiction — a ZIP it does not recognize is accepted without comment, because a check that wrongly blocks a real address would be worse than the mistake it prevents.

Clocking in works on any device that knows where it is — phone, tablet or laptop, in a web browser. Devices without a satellite receiver are now asked a second way, for the network-based location they can actually give, which is what makes a laptop work. The installed desktop app is the exception and cannot report a location at all. → Clocking in works on any device that knows where it is

Carely fits a phone screen. Every screen is checked at handset width on each build, so a page cannot quietly become wider than the phone it is read on.

EVV and the state

A visit with no location is now held back from the state instead of being sent. If a device could not give a location — a clock-in from a computer, or a refused permission — that visit waits on the Medicaid screen marked Held, with the reason, until somebody closes an exception for it. Previously it was transmitted like any other visit. Location is what Electronic Visit Verification verifies, so a record without one is now a decision somebody makes rather than something that happens quietly. → Clocking in works on any device that knows where it is

And the visits that were sent now carry their coordinates. The transmission to the state was reading the location from the wrong place and sending an empty one every time, including for visits where the phone had captured a perfect fix. Both punches' coordinates, the accuracy, and any exception reason now go with the visit.

Carely warns before a clock-in that cannot be verified, rather than showing it afterwards under a punch that has already been made.

And a held visit asks to be dealt with. It raises a notification for your administrators rather than waiting quietly on a screen nobody opens on a good day, and the notification clears itself once the visit is released.

Visits

A caregiver with no signal can now clock in. The punch is saved on the phone with its GPS fix and sent when service returns — and the visit keeps the time it really happened, not the time it uploaded. A band across the screen shows anything still waiting, so nobody has to wonder whether their visit reached you. See Clocking in with no signal.

A visit worked with no signal is now kept whole. Three things could go wrong after a caregiver clocked a visit without service, and all three are fixed. A saved punch used to be discarded if the eight-hour sign-in ended before the phone found service; it now waits, and is sent — with its real time — when the caregiver signs in again. Opening the app in a dead zone used to sign the caregiver out; they now stay signed in, and the schedule loads when service returns. And the note, mood and checklist written at clock-out used to be lost when there was no connection; they are now held and sent with the punches. If Carely ever refuses a saved punch, the caregiver is told on screen and asked to tell the office. See Clocking in with no signal.

The care plan shows each signature at the top. The administrator's and the responsible party's signatures now open the plan — with the name, how it was signed and the date — where they used to sit below its last line.

The plan of care is written and signed when you add the client. The Add client form now ends with the plan of care and a place for two signatures — yours and the responsible party's. A plan signed by both is shown as Signed and approved on the client's card, to the caregiver and to the family; one that is not says who it is waiting for. Changing a plan afterwards removes the signatures on the old wording until it is signed again. See The care plan.

A caregiver can open Carely with no signal and still find today's visits. Their own visits for the days around today are kept on the phone — the client's name, the times and the checklist, nothing more — and removed when they sign out. A notice says the list came from the phone and when it was last updated.

Signed out mid-visit, a caregiver can still clock out. The fifteen-minute sign-out is unchanged, but the sign-in screen now offers Clock out for a visit started on that phone, with an optional note. It is saved there and sent when they sign in, and the visit ends at the time they pressed it rather than whenever they next found service.

A visit report can only be written by someone on that client's care team, a caregiver who has a visit with them, or the office. Previously any caregiver login could file one against any client by name.

Only the assigned caregiver, or the office, can clock a visit. The screens have always worked this way; the server now enforces it as well, so a visit cannot be clocked in or out by another caregiver's login or by a family login.

A visit can be clocked in once and clocked out once. Tapping the button again is refused with a message. Previously a second clock-out created a second timesheet, so one six-hour visit could be billed and paid as eighteen; a second clock-in reset the start time, so a full day could record as zero hours. Corrections now go through the timesheet, where they leave a trail.

Visits can only be booked for people on your roster. A name that matches nobody is refused when you schedule it, rather than becoming a visit that is worked, paid and exported to your payroll provider for someone who does not exist. Discharged clients and deactivated caregivers cannot be scheduled either.

The AI auto-scheduler now runs on Carely's server rather than in your browser. It staffs the open shifts already on your calendar and applies every rule the schedule screen applies: already booked, day not marked available, missing or expired credential, the license a clinical visit type requires, approved leave, and the caregiver's own weekly hour limit. Previously it knew only four of those, so it could hand you a full week that Carely then refused one visit at a time — while reporting that it had published them all. Each proposal now says why that person was chosen, nothing is assigned until you publish, and if somebody edits a visit while you are reviewing, the edit wins and Carely says which proposal it refused. → AI auto-schedule

Deleting a visit now tells the caregiver it is off. Reassigning a visit emailed them and moving its time emailed them, but deleting — the change most likely to send somebody to a client's house for a visit that is not happening — told them nothing.

The schedule shows a working window — roughly the last four months and the next eight — rather than the oldest visits on file. On an agency with a year of history the calendar previously filled up with old visits and hid everything upcoming.

Money

The screen after your trial ends now tells the truth. It read "Start your free trial — $0 today" to every agency that reached it, while the payment page behind it charged the full price at once (there is never a second trial). It now says the trial ended, on which day, and exactly what is charged today. And after paying, the app used to sit on "Loading Carely…" until you reloaded; you are now in your agency within seconds of the card clearing. → Your free trial

In-app payments now work — and say so honestly. The Payments screen read "Stripe is connected — payments are processed for real" as soon as the platform had a Stripe key, while every collection, payout and reimbursement was quietly recorded as simulated: no family could save a bank account, no caregiver could connect a payout account, and a refund had nothing to refund. All three now exist — a family's bank is saved on Stripe's secure form (from the client's card or Family Connect → Billing), a caregiver connects their payout account from My Pay, and a reimbursement is a refund of a real collection. The banner reflects your agency's connection, the lists say who has a bank or a payout account, and a payment that cannot be made is refused with the reason on the card rather than written down as if it had happened. Statuses now settle: Processing → Completed or Failed, with the bank's reason. → Taking payments in Carely

Changing a rate over work already done now asks you first. Rates run by the day, so a rate starting today covered all of today — a visit that finished this morning at $30 quietly became $45 when the office raised the rate that afternoon, and the family's invoice moved by $60. Carely only checked for disturbance when the date was in the past, so the commonest change of all was the one with no warning. It now stops before saving anything, says how many visits it would move and what the difference is worth, and lets you choose: cover the work already done, or start from tomorrow. Cancel now saves nothing at all — it used to leave the new rate on the client anyway. → Changing a rate: it applies from a date

A visit corrected after it was invoiced now reaches the client. Correcting a visit only ever changed the hours: a visit billed at four hours and put right to five stayed marked invoiced, so the fifth hour was never charged — while payroll paid the caregiver for it. The agency paid out and never billed, quietly, every time. Carely now records how many hours of each visit have actually been billed and charges the difference. Corrected downward, the family is credited instead, netted against their next invoice — and never as an invoice for a negative amount, which is not something to hand to AutoPay. → The billing worksheet

A visit that recorded no time is no longer billed as if it were nothing. When a clock-in and clock-out landed seconds apart, a four-hour visit closed worth zero hours and arrived on Timesheets as an ordinary line reading "0 · $0", with nothing to say what had gone missing. Billing it charged nothing — and marked the visit invoiced, which meant that when the office later corrected those hours the visit never came back to be billed: the caregiver was paid and the client never charged. These visits now appear in EVV Exceptions, are named on the billing worksheet as held back, and bill normally once corrected. Genuinely short visits are untouched. → Closing a visit when the evidence is missing

An invoice no longer bills work whose period is still open. The billing worksheet took every approved, unbilled visit regardless of its date and then labeled the invoice with the last closed period. A visit worked on 10 September was billed immediately on an invoice headed "Aug 1–31" — a bill for a month that did not contain the work — and because those visits were then marked invoiced, September's own invoice came up short by the same amount. Visits now wait for their own period to close, and the invoice states the period the work is really in. → The billing worksheet

You can record a payment that arrived as a check or cash. Every unpaid invoice now has a Mark paid button. Until now the only thing that could mark an invoice paid was a family paying through the portal, or AutoPay — so an agency banking a check had no way to close it off, the balance stayed outstanding for ever, and Carely kept sending reminders to a family that had already paid. → Payments

Caregivers and families no longer see the agency's other rate. Both the charge rate and the pay rate travelled on every client record to everyone who could see it, so a caregiver could read what you bill for their hours and a family could read what you pay the caregiver. Each side now sees only their own: families the charge rate, caregivers the pay rate. → Rates: charge and pay

"Start free" opens the new-agency form. It used to open the sign-in screen — "Welcome back", with a password field — so somebody starting an agency had to find "Create an account" and then switch a tab before reaching the form they had asked for.

The setup wizard knows your agency's name. It asked for it again, with an empty box, immediately after you had typed it at signup.

Invoices have a number you can read. They used to show the internal record id — inv-mtqb1d8d-40-ryxj — on the emailed invoice, the printed copy and the family's portal. Nobody can read that over the phone or write it on a check. Every invoice now has a plain number, starting at 1001 and counting up per agency. Invoices raised before this were numbered in the order they were issued and keep those numbers.

Sending an invoice now sends it. The Send button stamped the invoice "emailed" and put nothing on the wire — no email, no record — and then showed a green confirmation. "Bill & send due now" and "Email unsent" did the same. An invoice is now only marked sent when a provider has accepted it; if it is refused you are told why, on the spot, and the invoice stays in your list to try again.

Late-approved work is no longer lost. A timesheet approved after its pay run was locked used to sit unpaid forever, in no run at all. It now carries into the next run, and the run says which hours came from an earlier period.

The same invoice cannot be raised twice by a double click. An identical invoice within a few minutes is refused and points at the one that exists. A deliberate second invoice still works — confirm it.

Rates must be real numbers. $25.00, 1,234.56 and 12.34.56 are refused rather than guessed at, because a rate guessed wrong becomes somebody's pay. A rate that was not a number used to be stored as typed and quietly billed as $0.00.

The payroll export carries scheduled hours as well as verified hours. The manual has always told you to sort by the difference to find problem visits; the export only ever had one hours column, so that could not be done. Both are now there, with the columns named.

The Expenses screen opens on the last 30 days rather than the calendar month. On the 1st of a month "this month" is one day long: nothing invoiced yet, no timesheets approved yet, but the monthly overhead has already landed — so the screen opened on zero revenue and a loss, every month, for anyone who looked early. "This month" is still one click away.

Carely now works out what your agency actually spends. There is a new Expenses tab in the Revenue Center. Caregiver wages and employer payroll taxes are calculated from your own timesheets and pay rates; you record your overhead; Carely adds them up and shows your operating profit and margin for whichever period you pick.

Profit & Tax now uses your real figures. It previously took every invoice you had ever raised, multiplied it by twelve to make an "annual" revenue, and estimated your expenses at 62% of that. Neither number was yours, the 62% was never disclosed anywhere on screen, and the revenue figure got further from the truth the longer you used Carely. It now uses what you invoiced in the tax year you select, and the expenses Carely has calculated and you have recorded. You can still type different numbers in to model a scenario — Carely will tell you when you are looking at a scenario rather than at your own figures.

Withholding is calculated, not estimated. Federal and state income tax now go through the real 2026 tax brackets, standard deductions and exemptions for all fifty states and DC, using each worker's W-4. Carely previously applied a flat 22% federal rate — the IRS rate for bonuses, not wages — and each state's top tax rate to every dollar earned. In California that took 6.6% from a caregiver whose real rate is about 1.3%. Local city and county tax is now withheld for Maryland (all 24 jurisdictions), Indiana (all 92 counties) and New York (New York City and Yonkers). Each caregiver has a tax locality on their record, because local tax depends on where they live rather than where the agency is. States Carely cannot yet calculate — Pennsylvania and Ohio among them — are named on screen instead of being quietly withheld at zero.

The cash-flow forecast costs your payroll properly. The 30-day forecast in the Growth Center used to price caregiver hours at an assumed 62% of the charge rate. It now uses each client's real pay rate, your overtime rules, your employer payroll taxes and your recorded overhead.

Time

Carely now works on your agency's clock, not UTC. Days, pay periods and "today" begin at midnight where you are. Previously a visit worked on a Sunday evening fell into the following pay period and was paid up to two weeks late — four hours' worth of window on the east coast, ten in Hawaii. Daylight saving is handled: a week containing a clock change is still seven days.

Carely uses the time zone of your licensed state unless you set one explicitly in Account settings.

The screens now agree with the server about what day it is. The server had learned your agency's clock; the screens had not. Every date a screen filled in for you — the New shift form, the "today" column in the week grid, the pay period on the Payroll tab, the date a rate change takes effect, the date on an incident — came from the UTC clock, so from 8:00 PM in New York (7:00 PM in Chicago, 5:00 PM in Los Angeles) they all opened on tomorrow. A visit booked in the evening for "today" landed a day late; an evening rate change took effect a day later than intended. Dates a screen fills in now come from your own clock.

State forms and Medicaid

State forms are built in. Forms a state defines — Virginia's DMAS-99, DMAS-225 and DMAS-97 — can be completed against a client's record and signed, with the signature bound to exactly what it was signed against. Carely records the answers and tracks the dates these forms depend on; where the state requires its own document, that is still what you file. See State forms.

A form Carely does not hold cannot be opened. DMAS-98R is listed and marked pending rather than guessed at, because a regulatory form with invented fields looks official, gets signed, gets filed, and is wrong.

Medicaid claims are filed from the screen, and answered on it. The Medicaid screen could connect a clearinghouse and then told you claims were "API-only" — no agency could actually file one. Now the screen says where a Claim.MD Account Key lives, takes your agency's NPI, tax ID and address once, and fills a claim in from the client's record and their approved visits — one T1019 line per visit at the charge rate — for you to review and file. Nothing came back before unless you pasted an 835 by hand; now each claim's status (accepted, rejected, denied, paid) and every 835 remittance are pulled from Claim.MD every fifteen minutes or on Check now, and paid claims mark themselves paid. A visit already on a claim in flight cannot be billed again. → Medicaid billing

The Medicaid screen no longer claims what it cannot do. Every clearinghouse was shown with ticks against eligibility checks, remittance retrieval and claim-status lookup. Only what is built is shown: claims for every clearinghouse, status and remittances for Claim.MD.

Clinical

Clinical work follows the license, not the job title. Administering medication, writing clinical notes and signing physician orders require a current nursing license on record. Administrators see everything and can ask a nurse about any record, but do not perform clinical work. If you own the agency and are a nurse, record your license in Account settings → My clinical license. An expired license stops clinical actions on the day it lapses.

Nurses have their own app — Nurse Station — holding the medication record, clinical documentation and physician orders.

A discontinued medication cannot be administered. Nothing can be recorded against it at all. A held medication cannot be given, but the hold can still be documented.

Renaming a patient carries their whole chart — medications, the eMAR history, orders and notes. Who gave a dose is deliberately not rewritten: that records who performed an act at the time.

Email

Links in Carely's emails work. Password resets, the staff-sent reset link and the welcome manual all carried a broken address — the whole list of permitted web addresses pasted together instead of one of them. The token was always fine; only the part a person clicks was wrong, which is why nothing noticed.

The invoice email has a link in it, and says the same thing once. It told the reader to sign in without giving them an address, and warned twice, in two different wordings, that the mailbox does not take replies.

A notice is emailed once. Two overlapping delivery passes could send the same message twice a minute apart. Each message is now claimed before it is sent.

Nobody is told about a subscription they do not have. A new agency that had not subscribed to anything could receive a yearly renewal reminder. Reminders now go only to agencies with a live subscription.

No email Carely sends accepts a reply. Every message now comes from a mailbox that receives nothing and says so at the bottom. This is deliberate: an email a family can answer is an email a family will answer with detail — "she fell this morning, her sugar was 3.2" — and that reply would sit in a mail provider's store, outside Carely's encryption and outside its audit trail. Replies belong in the care-team conversation, where the whole team sees them and the record is kept with the client.

A family is told what kind of thing is waiting, and nothing else. A message, a care update, a visit report, an invoice, a payment reminder, an incident, a change to the plan of care — named, with no client name, no health detail and no amount. Those are one sign-in away. Where a reference is unavoidable, the client's ID is used.

Caregivers are emailed when their schedule changes — a visit added, moved, taken off them, or deleted — carrying the day and time only, and telling them to open Carely for the client, the address and the care plan.

Open shifts go out as one digest a day, listing what is open in the next two weeks, never the client's name, and nothing at all on a day with nothing open.

A visit still running an hour after it was due to end emails the caregiver and raises it in the office's bell, because a clock-out that never happens has to be corrected later by a person attesting to times nobody witnessed.

Accounts and access

The free trial is a trial. A new agency used to be locked out of every screen until a subscription was completed — a card form in front of a product you had not been allowed to see. You now get 30 days of full access with no card, and access closes only when the trial has actually run out. See Your free trial.

If you sign up and get distracted, Carely follows up — once after a couple of days and once more about a week later, then stops. It only writes to agencies that have not added anyone yet.

Renaming a caregiver renames the account they sign in with, so their schedule, clock-in, pay details and license keep working. Previously the records moved and the login did not, leaving them with an empty schedule.

Joining an agency requires its organization code. Signing up without one is refused, so nobody lands in your agency by accident.

Messages

A family can now reply to the office and to their caregivers separately. There used to be one room per client — every admin, every assigned caregiver and the family in the same thread — so "reply" meant reply to everyone, and a family with a concern about their caregiver had nowhere to raise it that the caregiver could not read. See Who you are messaging.

Caregivers can message the office about a client without the family reading it, which is where a gate code or a scheduling problem belongs.

The messages you already had have been sorted into those conversations. They arrived in "Earlier messages" — one undifferentiated pile — which meant the new conversations started empty and the history stayed behind. Carely has now placed each one by who wrote it, without showing anything to anyone who could not already read it and without deleting anything — Earlier messages is empty as a result and no longer appears.

A caregiver's conversation with the office is now their own. It was held per client, so every caregiver attending that client read all of it — under a heading that says "You and the office. The family cannot see this conversation." A caregiver raising a safety concern, or saying they were struggling, was writing to their colleagues without being told. Each caregiver now has their own private line to the office, and when the office replies it names the caregiver it is answering. In Operations Hub, Caregiver ↔ office is listed by caregiver rather than by family. → Who you are messaging

Family Connect now reads correctly for a caregiver, and leaves out the family's billing. Caregivers open the same screen as the family, and every word on it was written for the family: a caregiver was greeted with "Your loved one's care", given a tab called "My Loved Ones", and — because the list was built from family links rather than caregiver assignments — told "No loved ones linked yet" while their messages were working. It now names their clients. Billing and Quality no longer appear for caregivers: the first is what the family owes, and the second is the family's own rating of the agency. The plan-of-care signature and the emergency-contact editor are the family's too, so a caregiver now reads both rather than being shown controls the system would refuse. → What caregivers see

"Message family" in Operations Hub reaches the family again. It named no conversation, so after conversations were separated it was addressed to the family–office one, which caregivers may not write to. It now goes to the conversation the caregiver and family share.

The "Earlier messages" tab is gone. Everything that was in it has been sorted into the real conversations, so it was a permanently empty box sitting beside three live ones. Nothing was deleted.

Every conversation now carries its own unread count. A red bubble sits on Messages, on each loved one, and on each conversation — and in Operations Hub, on each conversation type and each family in the list. The number on top is exactly the total of the ones underneath it, so you can see not just that something is waiting but which conversation it is in. Opening one clears only that one, and nothing is marked read until it is actually on screen. → Unread counts

Signed agreements

Downloading a signed agreement now gives you the agreement. It produced a file of record ids, hashes and timestamps — correct for a computer and no use to the auditor, attorney or agency owner who asked. You now get the document: the Business Associate Agreement in full, then every other document the agency's people accepted, each followed by who accepted it, in what role, how, when and from which address. It opens in a browser and prints to PDF.

It says whether the text can still be trusted. Each acceptance recorded a fingerprint of the exact wording on screen at the time. The document re-checks it and states plainly whether the text shown is the text that was signed, whether the published wording has changed since, or whether that version can no longer be reproduced. A signature record that cannot say which of those it is proves nothing.

View shows the agreement, not a summary of it. Selecting View used to fetch the record and draw a few lines underneath the agency list — below the bottom of the window, so it looked like nothing had happened while Download worked. It now opens the same document in place, and scrolls to it. One person's record opens the same way.

The machine-readable record is still available for anyone reconciling programmatically.

The desktop app

On a Mac, Carely reopens after its window has been closed. Closing the window leaves the app running in the Dock; clicking the Dock icon then showed "A JavaScript error occurred in the main process" and no window, and the only way back was to quit. It now opens a window, every time. The app updates itself: open it once and the fix is installed the next time you quit.

If the desktop app ever hits a problem of its own, it says so plainly. Instead of a technical error box it asks "Carely ran into a problem" with Restart Carely and Keep working, and the problem is reported to Carely automatically — the error and the app version, nothing from your screen.

Buttons that ask you for a reason now work in the desktop app. Retrieving or downloading signed agreements, exporting an agency for a legal hold, break-glass retrieval, signing or voiding a state form, recording a UAT session, approving a release — and the emergency stop for all agents — all asked for their reason using a browser pop-up the desktop app does not support. Clicking them did nothing at all: no box, no error, no clue. They now ask inside Carely, in a proper dialog that shows what will happen and what is recorded.

Nothing was ever wrongly exported or approved by this — the actions simply never ran.

Reading the screen

Carely keeps itself up to date while you have it open. It used to load your agency's data once, when you signed in, and never again — so a tab left open all day went on showing what was true at breakfast. A caregiver could clock out and your Timesheets screen would still say "No timesheets yet" until you knew to reload the page. A day of visits could look like a day where nothing happened, which meant nothing was approved and nothing was invoiced.

It now refreshes when you come back to the tab, and quietly on a timer while you are looking at it. Nothing you are typing is disturbed — a half-filled form stays exactly as you left it — and a refresh that fails because the internet hiccupped says nothing, because you did not ask for it and nothing was lost. Anything you do that fails still tells you, as it always did.

Every screen is now checked for readability in both light and dark mode. Carely walks each app and each tab in both themes and measures the contrast of every piece of text on the screen against the accessibility standard (WCAG AA). A screen with text that is too faint to read no longer ships.

That check found and fixed a long list of places, including:

• Avatar initials — the colored circles beside names. The initials were painted the wrong color for the circle behind them, so they were invisible in light mode and again in dark mode.
• Alerts and status banners across the Copilot, Clinical, Medicaid, Migration and Enterprise screens. These painted a pale background but let the text follow the theme, so in dark mode they were pale text on a pale panel.
• Leave balances on the Operations Hub. An over-balance figure was dark red on a dark red chip.
• Status chips — the small uppercase labels such as CRITICAL, RISK and DUE. The white label did not read on the lighter fills (amber, green and grey).
• Buttons that are switched off. A disabled button used to fade out entirely, so you could not read what it would do once it became available.
• Money and percentage figures shown in a status color, and the value on every summary tile.
• The current app in the sidebar was the hardest label in the list to read in dark mode.

Color still means what it always did — red is still a problem, green is still fine. What changed is that the shade now follows the theme you are in, so the same information is legible on a light screen and a dark one.

Growth percentages are rounded. A change in client count could display as "33.33333333333333%".

Dropdowns and text boxes follow the theme. A dropdown such as the agency picker on Enterprise → Compliance was showing black text on a dark panel, because form controls do not inherit their color from the page the way ordinary text does — they use the browser's own black on white until told otherwise. Every input, dropdown and text box in Carely now takes the theme's text and background together.

The readability check now runs as each kind of user — administrator, caregiver, family member, enterprise owner and parent agency — because each one sees a different set of apps. Checking only the administrator's view left the Nurse Station, Family Connect and the Enterprise Portal unchecked, and unreadable text was still sitting on the enterprise screens. It now covers 133 screens in both themes.

Getting Carely on your computer

On Windows, Carely runs in your browser, and the Windows download has been removed. Nothing about Carely has changed for Windows users — every feature works in Edge or Chrome exactly as before, and the manual describes the same screens.

What changed is that we no longer offer a Windows program to install. The installer was not signed by a certificate authority, so Windows greeted anyone who ran it with a "Windows protected your PC — unknown publisher" warning. Clearing that warning costs several hundred dollars a year, and we would rather point you at the browser version, which is the same Carely, always current, and never asks you to click past a security warning.

On a Mac, the desktop app is now signed by Apple and notarized, so it opens without any warning. It remains optional — the browser version is identical.

If you already installed the Windows app, it keeps working and keeps updating itself. → Where Carely runs

Subscription billing

What counts toward your plan is now measured from your activity, not from a status field. A billable client is one whose record generated a visit, an EVV clock-in, a visit note or an invoice in the last 30 days.

Two things change in practice. On Hold and Hospitalized clients now count — the manual always said they did, and now the billing engine agrees. And archiving or discharging a client no longer removes care they already received from the current measurement; it ages out over the following 30 days instead of disappearing the moment the status changes.

Nothing here bills you for clients you no longer serve. A client who genuinely leaves stops counting once they have had no activity for 30 days, with no action needed from you, and the same is true whether you archive them, discharge them, or do nothing. → What counts as a billable client

Upgrades now take a full 14 days. The sustained-usage counter was advancing once per billing run rather than once per day, and the run happens twice daily — so the "14 consecutive days" you are shown before a plan moves up was arriving in about seven, and a deployment could shorten it further. It now counts days.

Your daily briefing

Carely now tells you which engine wrote your executive summary. Above the summary in the Agency Growth Center you will see either "written by Carely AI" or "from Carely's analysis engine", and if you asked for the AI version and it was unavailable, it says so rather than quietly showing you the other one.

Only the wording is affected. The Business Health Score, the risks, the opportunities and every figure on the screen are calculated identically either way — the AI writes the summary, it does not do the analysis. → The written analysis