SuiteMedica: an AI agent that books the appointments reception cannot get to
One agent handles six channels (WhatsApp, phone, Instagram, Messenger, email and web chat) and puts the appointment in the clinic’s calendar. It is in production and under active development.
- Sector
- Healthcare
- Role
- CTO
- My part
- Architecture, leading the team and development: I still write code.
- Period
- March 2026 to present
- Status
- In production, under active development
- Team
- Three people
System blueprint
The flow, in text
- The patient writes or calls on any of six channels: WhatsApp, phone, Instagram, Messenger, email or web chat.
- Every channel reaches one AI agent, with the same context.
- The agent checks real availability in the calendar.
- The agent puts the appointment in the calendar. Underneath, the platform: a NestJS API, PostgreSQL and a Next.js web app, with tests on every change.
Platform: NestJS API · PostgreSQL · Next.js web app · tests on every change · Diagram. Not live data.
Product in use
Problem and constraints
Patients ask for an appointment in a WhatsApp message at eleven at night, in a call nobody picks up, or in an Instagram comment. By the time reception sees it, the patient has booked somewhere else.
The clinics we work with also had their information scattered: the calendar in one system, the clinical record in another, billing in a third, and reminders done by hand.
Constraints
- Every clinic is a tenant: authorization is the most delicate part.
- One agent for all six channels, with the same context.
Decisions and architecture
I chose one agent for every channel, instead of one bot per channel. The patient writes or calls wherever they like and the conversation lands in the same place, with the same context.
The agent books the appointment. It checks real availability in the clinic’s calendar and prioritises by the patient’s insurance and condition, as a good receptionist would.
We built it on a single multi-tenant platform for the whole operation, so the agent works on the same data the clinic’s staff use: scheduling, clinical record, orders, consents, billing and point of sale, statutory care guarantees (GES), surveys and reminders.
It is a monorepo: a NestJS API on PostgreSQL (TypeORM) and a Next.js web app with Supabase SSR, with WebSockets for real time. The agent lives in its own package, agente-citas, and reaches language models through OpenRouter, so the model can change without touching the rest. Channels come in through Twilio and WhatsApp.
Because every clinic is a tenant, authorization is the most delicate part. That is why continuous integration runs dedicated end-to-end authorization tests, on top of build, type-check, lint and unit tests, before any change is accepted.
Results
SuiteMedica has been in production and under active development since March 2026, with a team of three. The public site is suitmedica.cl.
The integrations with IMED electronic insurance vouchers and with Haulmer/TUU payment terminals are in production.
- 6channels handled by one agent: WhatsApp, phone, Instagram, Messenger, email and web chat.
- 100test files. Every change goes through build, type-check, lint, unit tests and end-to-end authorization tests.
- 3integrations in production: Twilio and WhatsApp, IMED electronic insurance vouchers and Haulmer/TUU point-of-sale terminals.
Limits
- This case does not publish evaluation, latency or cost metrics for the agent, or usage figures per clinic.
Do you have a system that cannot be touched?
Tell me the problemWant the full career history?
View the CV