Kinea
Clinic software that runs a whole practice from one screen.
- clinics live
- 6
- appointments booked
- 1.5k+
- of bookings made by the assistant
- 65%
The problem
Physiotherapy clinics were running on notebooks and a WhatsApp group: schedules in someone's head, prepaid packages on paper, cash in two currencies, and a receipt to send before the patient reached the door.
What we built
The whole product: scheduling, patient records, clinical notes, payments and reporting, plus an AI assistant that runs the front desk in plain language. Bilingual English and French on every screen, from day one.
What it does
- AI front-desk assistant that books, checks in, takes payments and writes notes from a sentence
- Dual-currency ledger with methods and reversals
- Branded payment receipts sent straight to WhatsApp
- SOAP clinical notes that lock the same day, with addenda
- Prepaid session packages that count down automatically
- One-click export of every patient, note and payment
Built with: Next.js · TypeScript · PostgreSQL · AI assistant · i18n
A front desk that runs off one sentence.
Type “book Sara for a follow-up tomorrow at 3” and Kinea books it. The same assistant checks patients in, takes the payment in either currency, sends the receipt to WhatsApp and drafts the clinical note. It runs inside the product, against the clinic's own schedule and records, not in a chat window that knows nothing about the business.
Typed the way a receptionist talks
Plain English or French, the way a person at the desk would say it.
Real writes, not suggestions
Every action lands in the schedule, the ledger or the chart.
It shows, then waits
The booking and the amount are shown back before anything happens.
A shortcut, not a lock-in
Staff can still do all of it by hand.
How the assistant is kept honest.
The first question every clinic asks is how it avoids making things up. The answer is mechanics, not reassurance.
Grounded, then cited
Answers are retrieved from the clinic's own data before they are written. No source, no answer: it says it does not know.
A short list of tools
The model cannot do anything the product cannot do. It calls a fixed set of functions with typed arguments, under the signed-in user's own permissions.
Humans confirm the writes
Reading is free. Booking, charging, cancelling and sending are shown back and wait for a tap.
Tested before it ships
A set of real cases runs against every change, so a prompt edit cannot quietly break an answer that used to be right.
Want one built like this?
Tell us what you're building. You get a written scope and a fixed price for the first milestone.
Book a 20-min call