The teammate who never clocks out.
ADA is an agent inside your inbox, your calendar and your patient list — not a chatbot bolted onto the website. It does the work a very good front desk would do if it had infinite hours, and it waits for your approval before it speaks for you.
Ask it about your own clinic
ADA has a working set of tools, not a canned script. It can look up availability, book and reschedule appointments, register a patient, search your patient list, count appointments over a period, find and update leads, and put together a daily briefing. Ask in plain language and it uses the right one.
When a question needs a choice — which patient, which slot — it renders a real interactive card in the thread rather than making you retype anything.
Held for 10 minutes. Confirm to write it into ModMed and send the intake form.
A website chat that can
actually book something
A chat widget on your own site. It answers from a knowledge base you control, lists your real services and providers, checks genuine open times, holds a slot while someone finishes, and takes the booking — including a deposit if you ask for one.
It runs on a separate, restricted profile with no access to staff tools, and you decide when it goes live on your site.
You decide what it's allowed to say.
An agent that writes to patients on your behalf needs brakes. These are the ones ADA has.
Approval-first by default
Replies from your inbox wait for a person before they send. You decide, per conversation type, where ADA may answer on its own — appointment confirmations and the website chat are the usual first two — and everything else stays under review.
It can't see more than you can
ADA doesn't have its own database access. It calls the same API your browser does, carrying the signed-in staff member's credentials, so it inherits exactly their permissions and clinic scope — no more.
It hands off rather than guesses
Complaints and medical questions go to your team with the whole thread attached. The website widget has an explicit escalation tool for exactly this, and ADA uses it instead of improvising.
Versioned instructions
The instructions ADA works from are editable and versioned. Publish a change, see who published what and when, and roll back by publishing an earlier version. Its knowledge base is yours to edit, so you control exactly what it answers from.
Silence it per conversation
Any thread can be taken off ADA entirely. It stops drafting there and the conversation is yours.
It learns your clinic, not your patients
ADA answers from your knowledge base, your service list and your live calendar — content your team writes and can change in a moment. Grow Theory does not train any model on your patients' conversations.
It tells you what's
about to be forgotten
ADA watches conversations for the ones drifting — a patient who asked a real question and hasn't had a real answer, a thread handed to someone who's been in clinic all day. Each nudge comes with the reason it was raised and a reply already drafted.
Acknowledge it, snooze it, or dismiss it. Nudges live in their own inbox tab with an open count, so nobody has to go hunting.
Patient asked an operational question about insurance verification and hasn't had a substantive answer.
"Apologies for the wait — yes, we verify insurance before your first appointment. I've asked Kate to confirm the specifics for your plan and she'll come back to you today."
Front-desk capacity,
without the hire
The inquiries that arrive after hours are the ones most likely to go to whichever practice replies first. ADA answers them in seconds, with your real availability, and books them into your calendar — so your team walks in to confirmed appointments rather than a backlog.
During clinic hours it does the drafting, the chasing and the flagging, and leaves the judgment to your people.
Watch it answer
a real inquiry.
Bring a message you'd normally reply to yourself and see what it drafts.