Meet Atlas
Atlas is the AI built into Ready Practice. It answers questions grounded in your organization's real data and in the product documentation, and it can do work for you — draft a protocol, set up a booking link, chase an unpaid claim — always showing you exactly what it plans to change before anything happens. Think of Atlas as a capable new team member who has read every guide, can see your calendar and ledger, and never takes an action without your sign-off.
This page explains what Atlas is, what it can and can't do, how its approval model works, and how to get the best answers out of it.
Where you'll meet Atlas
Atlas shows up in two forms, for two different audiences:
- The staff copilot lives inside the Provider web portal — open it with the Ask Atlas button in the bottom corner of any screen. This is the Atlas your team talks to: it knows your clients, schedule, billing, labs, and settings, and it can stage changes for you to approve.
- The client Health Assistant lives inside your client-facing app. It helps your clients understand their own data — their labs, metrics, protocols, and upcoming appointments — in plain language. It only sees that one client's information, and you control whether it's enabled at all (under Settings → Client Apps).
The staff copilot works for your team; the Health Assistant answers a client about their own data when the client asks. Atlas never reaches out to a patient on its own — more on that below.
What Atlas can do
Atlas spans four jobs. You can ask it anything in these areas, in ordinary language.
Answer from your data. Atlas can read a client's labs, wearable metrics, visit notes, account balance, insurance claims, and chat history, plus your org-wide schedule and revenue picture. Ask things like "what's my revenue this month?", "who hasn't rebooked in 60 days?", "which claims are unpaid?", or — with a client selected — "summarize her last two lab panels."
Answer how-to questions. Atlas has read this entire documentation set, so "how do memberships work?" or "how do I set up insurance billing?" gets a grounded answer with links to the full guide — no digging through docs yourself.
Do setup work. Ask Atlas to "create an intake form for new clients," "set our brand color to #0E8C8B," or "create a booking link for new-client consults." It drafts the change and shows you the result before it goes live.
Assist clinically. Atlas can summarize a client's labs, flag out-of-range trends, and draft protocols and notes for a clinician to review. What each staff member can see through Atlas follows their role's permissions — a staff member without health-data access can't get health data out of Atlas either.
- AI Scribe — Atlas's note-taking arm: it joins or records your sessions and auto-drafts the visit note for you to review and sign.
The approval model: nothing happens without you
The most important thing to understand about Atlas is how it makes changes. Every change Atlas proposes becomes a card — a small summary showing exactly what would be created, updated, or sent. Nothing is written, sent, or charged until a human clicks Approve on that card. Decline the card and the change simply never happens.
This applies to everything: a drafted client message, a schedule change, a new form, a billing adjustment. Atlas prepares; you approve. Cards can be approved from the web portal, and if your organization has connected Slack, the same cards arrive there too — so an owner can approve Atlas's work from their phone without opening the portal.
Healthcare operations don't get "undo" on a sent message or a posted charge. The card model gives you Atlas's speed with a human decision on every consequential step. There is no setting that lets Atlas skip approval.
The bright lines
Two rules are absolute, in every configuration, for every organization:
- Atlas never messages patients on its own. Every outbound client communication Atlas drafts — chat message, email, reminder — waits as a card until a staff member approves it. The client-side Health Assistant only ever responds when a client starts the conversation.
- Atlas never initiates a charge. It can prepare an invoice, flag an unpaid balance, or stage a payment for review, but no money moves on Atlas's decision. A human approves every charge.
No feature, integration, or setting overrides these two lines. If you ever see behavior that looks like Atlas contacted a patient or moved money unprompted, report it immediately (see below) — that's a bug, not a feature.
Report bugs and request features through Atlas
Atlas is also your direct line to the product team. If something is broken or missing, tell Atlas in plain language:
"Report a bug: the calendar shows canceled appointments twice on Tuesdays."
Atlas files it straight into the product team's tracking system — with the context it can see, like your organization and the screen involved — and it's automatically checked against known issues so duplicates get linked rather than lost. Feature requests work the same way: "it would be great if booking links supported gift cards" becomes a tracked request.
When Atlas can't do something you ask, that's not a dead end — it records the gap so the team sees what users actually need. Asking for something Atlas can't do yet is genuinely useful; it's how the roadmap gets shaped.
How to ask good questions
Atlas is at its best when it knows exactly what you're pointing at. A few habits make a big difference:
- Select the client first for client questions. Open the client's profile (or name them explicitly), then ask "summarize the latest labs." Asking about "the labs" with no client in context forces Atlas to guess.
- Be specific about scope and timeframe. "Revenue this month vs last month" beats "how's revenue?" — the more precise the question, the more precise the answer.
- Say what you want done, not how. "Create a booking link that only offers the Intake Visit and charges up front" is a complete instruction; Atlas handles the clicks and shows you the card.
- Ask follow-ups. Atlas keeps the thread's context, so "now show just the ones with insurance" works after a client-list question.
The fastest way to learn what Atlas can do is to ask it: "what can you help me with?" — it answers with what's available to your role and organization.
Next steps
- Let Atlas draft your visit notes → set up AI Scribe so sessions turn into ready-to-sign notes.
- Decide what clients get → enable or disable the client Health Assistant under Settings → Client Apps.
- Put Atlas to work on setup → wherever a setup guide feels heavy, ask Atlas to do the step and approve its card instead.