A clinic front desk that stops shouting into the phone
Three receptionists, a jammed switchboard, patient forms retyped by hand. We automated booking, reminders, and request routing. The front desk can finally breathe.
Details anonymized and adapted for confidentiality.
The problem
A group practice, eight doctors, three people at the front desk. The phone never stops, around 140 calls a day, more than half just to book or move an appointment. Meanwhile patients stand waiting at the counter, and intake forms arrive on paper or as badly scanned PDFs, retyped one by one into the patient software.
The concrete cost: roughly 18 % of slots lost to no-shows for lack of a reliable reminder, data-entry errors on medical history, and two receptionists out of three spending the day firefighting instead of welcoming people.
What we built
Self-service booking - A web and phone module wired into the existing calendar. The patient picks a slot by reason and practitioner without going through the desk. Real-time sync, zero double-booking.
Automatic multichannel reminders - SMS and email sent at D-3 and D-1, with one-click confirm or cancel. A freed slot reopens for booking instantly.
AI-structured intake forms - The patient fills in an online form. A language model extracts the useful fields (history, medication, reason) and turns them into a structured record ready to validate, no more manual retyping.
Smart routing of incoming requests - Emails and messages auto-classified (urgent, admin, results, prescription) and routed to the right person, with a draft reply suggested for standard cases.
The result
- No-shows more than halved, from 18 % to around 7 % in three months thanks to reminders and automatic slot reopening.
- 40 % fewer calls at the desk, with most booking moving to self-service.
- Patient records ready in 2 minutes instead of 12, with no retyping and no transcription errors.
- One receptionist fully redeployed to patient support and care coordination.
- Everything stays auditable: every automated action is logged and can be validated by the team.
The stack
- Web booking module connected by API to the existing calendar and patient record
- LLM pipeline for extracting and structuring intake forms
- Language-model classification and routing of incoming messages, with a human in the loop
- Automated SMS/email reminders and real-time cancellation handling
- European hosting, encrypted health data, GDPR compliant