AI agent for a medical practice: automated scheduling and patient reminders
How a three-doctor practice halved its no-show rate and won back 12 hours a week with an agent that manages patient appointments.
Names and organisations are anonymised.
- Same day: a slot cancelled in the morning is offered again within the day
- Zero calls: confirmations and reminders go out without anyone dialling
Before and after
The starting point
A medical practice with three general practitioners gets 80 to 120 calls a day. The secretary handles the calendar, prescription renewals, test result requests and patient reminders. In practice, she does not have time to do it all well.
Three major friction points came back every week:
- Patients who don't cancel: about 15% of appointments are no-shows. A lost 20-minute slot, times three doctors and five days, is close to 5 hours of consultations gone every week.
- Calls to confirm an appointment: many patients call the day before to check their time. The secretary answers the same question 40 to 50 times a day.
- Handling freed slots: when a patient cancels at the last minute, nobody is told fast enough to fill the slot. It stays empty.
"The secretary was swamped. We were losing slots every day. And she no longer had time for real emergencies because she was answering basic questions."
What was deployed
The AI agent does not replace the secretary. It takes on the repetitive tasks that need no human judgement, so she can focus on what does.
Automatic reminders 48 hours before the appointment
The patient gets a text or a WhatsApp message two days before the appointment. They can confirm in one word or cancel directly. If they cancel, the agent frees the slot in the calendar in real time and tells the secretary. The slot can then be offered to the next patient on the waiting list.
Automatic answers to standard requests
When a patient sends a message to check opening hours, ask whether the practice is open, or request a repeat prescription, the agent answers with the right information. Questions that need a medical decision are automatically passed to the secretary or the doctor.
Managing the waiting list
When a slot opens up, the agent automatically contacts waiting patients in order of priority. The first to say yes gets the slot. What took 20 minutes on the phone now takes under 5 minutes, with no human involved.
You stay in controlThe practice decides: no appointment is moved without the front desk agreeing, and any medical question goes to the secretary or the doctor.
What changed day to day
The secretary got about 2 hours a day back. She is no longer flooded with confirmation calls. She can focus on emergencies, complex files and welcoming patients.
The no-show rate dropped from 15% to 7%. That is not a detail: for three doctors, it is the equivalent of half a day of consultations won back every week.
The practice did not change its management software. The agent plugged into the existing calendar through an API. The secretary sees everything in her usual tool.
What it does not do
The agent gives no medical advice. It does not handle emergencies. It does not interpret symptoms. It does organisation, not medicine.
This matters in healthcare: the tool must be very clearly scoped so patients are never confused. The whole setup is built around that scope.
Which practices is it for?
This kind of solution suits any practice with a high volume of appointments and an overloaded front desk: GPs, physiotherapists, dentists, ophthalmologists, nursing practices. The logic is the same: repetitive calls that eat up time, lost slots, a waiting list that is hard to manage.
Setup usually takes two to three days. The return on investment shows from the first week.