The most interrupted desk in the building
Whoever runs a clinic reception does three jobs at once: greets arrivals, manages the waiting room and the delays, and answers the phone. When all three collide - and they always collide first thing and at the end of the day - something gives, and it is almost always the phone, because it is the only one not looking you in the eye.
The result is familiar. Calls ringing out at lunchtime. New patients who ring three clinics and stay with the first that answers. And a capable person spending the day taking bookings instead of looking after the people in front of her.
A virtual medical secretary does not replace that post. It takes the phone off it, which is the part that makes it impossible.
What Clara handles in a clinic
Clara is the PulsifyAI voice assistant. In a practice she handles the administrative layer of the call:
- Booking, rescheduling and cancelling in the real diary, with each clinician's genuine availability and the right duration per appointment type;
- Answering new patients at any hour, including evenings and weekends, which is when many people decide to look for a practice;
- Answering what repeats - opening hours, address and parking, what to bring to a first appointment, which insurers are accepted;
- Confirming the day before and offering to rebook within the same call, which is where no-shows are recovered;
- Filling cancellations by calling whoever is on the waiting list;
- Capturing the administrative essentials before the appointment, so the patient arrives with the paperwork done.
The intake detail is in AI patient intake by phone.
The line that is never crossed: clinical advice
This is the rule that shapes everything else, and it is non-negotiable. A virtual secretary does not give clinical advice. It does not say whether a symptom is serious, does not suggest medication, does not interpret test results and does not decide whether someone should go to A&E.
What it does is recognise when the conversation has left the administrative and route it straight to the clinical team, with what it has already captured. If the caller pushes, the answer is always the same and it is the right one: that is a question for a clinician, and the conversation is now booked.
Setting that boundary before go-live is not a limitation. It is what makes the tool usable without risk, and it is how a serious deployment is told apart from a good demo.
Health data: what is captured and what is never touched
A clinic handles special-category data under a stricter regime than ordinary personal data. The rule we apply is minimisation by design: capture the minimum needed to book an appointment - name, contact, appointment type, clinician - and do not capture symptom descriptions or history.
That means deciding up front what is logged from each call, for how long, who can access it and under which processing agreement. If calls are recorded, further requirements apply, including a notice at the start.
Groundwork for the start, not after a complaint. The essentials are in health data on AI-answered calls and GDPR and AI voice assistants. PulsifyAI is a Portuguese company, so the data and the contract sit under EU law rather than someone else's.
No-shows are where the return shows first
In a clinic, the empty chair is the most expensive loss there is: the cost has already been incurred and no revenue arrives. And the day-before confirmation is exactly the kind of repetitive, fixed-time work that reception stops doing the moment the day gets busy.
An automatic confirmation the evening before, with the option to rebook within that same call, recovers a large share of those. The difference from the usual SMS is that the patient can respond and settle it there - and a rebooked appointment is worth far more than a politely announced no-show.
For dental practices we have a dedicated page at AI for dental clinics, and the diary mechanics in AI appointment booking by phone.