AI assistant for a clinic
It books the right specialist, sends the preparation notes and answers about address, hours and documents. Everything before the consulting room, nothing inside it.
- Buildfrom $1,700
- Timeline2–4 weeks
- Supportfrom $570/mo
Who this suits: Clinics where an administrator repeats the same answer about preparing for a test all day, and a patient who cannot get through books across the road.
What is in the first version
- services and prices
- doctor booking
- preparation notes
- reminders and rescheduling
The figures above are the package this sits on: AI agent or process automation. Anything past its edges the calculator adds before the work starts, not after.
How the path goes
- 01
The patient writes rather than calls
A message at any hour is matched against what the assistant may answer: your services, prices, hours and specialities.
- 02
A slot, not a request
It shows the time that specialist actually has free and holds the slot at the moment of confirmation.
- 03
The preparation arrives before the visit
The day before, a reminder carries what to do in advance and a one-tap way to move the appointment.
What decides whether this works or annoys people
A symptom is a route
Any answer with substance here is a medical consultation, so the assistant turns the description into a speciality and offers a time.
The schedule is not the record
The assistant works with the schedule and the price list; diagnoses and results are another category and never reach a chat.
The wrong doctor
A wrong speciality costs more than no booking, so the mapping is agreed with you and doubtful cases go to an administrator.
What to measure once it is live
- questions closed without an administrator
- bookings outside working hours
- no-show and late cancellation rate
- bookings with the speciality changed
Why this comes out faster
The shape of this one is known: the states, the edge cases and the things that usually go wrong have been decided before. Nothing here is a template, and the saving is not in your half of the work. It goes into your process, your content and the systems this has to talk to, which is the part nobody can have solved in advance.
A likely stack for this
Picked against the task when we scope it, not decided in advance. This is the shelf it usually comes off.
- Assistant and modelsPythonFastAPIOpenAIAnthropicpgvector
- Schedule and dataPostgreSQLRedisDocker
- Channels and monitoringTelegram Bot APIn8nGrafana
Price it yourself, right here
Five steps, and you can see the number without leaving a contact. The estimate accounts for the kind of work, what you already have and what it will need inside.
These bills do not come from us
- AI model calls
- hosting that meets your data rules
- the clinic system subscription and its API
- SMS reminders
Asked before the first call
- Will it start diagnosing people?
- No: symptoms only choose a speciality, and the wordings on which it must fall silent, show the emergency number and call a person are approved by you.
- Where are the conversations with patients stored?
- In your own infrastructure and in the jurisdiction you name; we keep no copy. What the assistant may see is limited before the build: the schedule and services yes, the record no.
- We already run a clinic information system, will it connect?
- If it has an API, the schedule is read from it and the appointment is created in it, so there is still one schedule. If it has none, we say so before the estimate.