Medical practice: reduce unanswered calls without hiring

In a practice, calls are not spread neatly across the day. They can bunch up at opening time, just after the lunch break or late in the afternoon, in other words when the waiting room is full and reception is checking patients in. The person at the desk then has to choose between the patient standing in front of them and the one waiting on the line.

The doctor, for their part, cannot pick up during a consultation without breaking the confidential one-to-one with the patient. When they practise alone, without a secretary, every ring is a dilemma: interrupt the examination or let it go. Either way, someone is poorly served.

Before looking for a tool, it is worth listening to what actually comes in on the phone for a week. These calls can be grouped into four types: booking, moving or cancelling an appointment; practical questions about opening hours, the address, access or documents to bring; requests that need an answer from the doctor, such as a test result to discuss or a prescription to renew; and finally, situations that worry the caller.

These types do not call for the same level of involvement. The first two can be handled without taking up a clinician’s time. The third only requires a clear message to reach the right person. The fourth sometimes has nothing to do with the practice at all, and the right response is to direct the caller elsewhere immediately.

The question “are you open on Saturday morning?” takes thirty seconds of a secretary’s time, but it can come up many times a week and interrupts whatever task is in progress each time. A reliable automatic answer, based on information the practice provides itself, removes these interruptions without degrading the service: the caller gets an answer straight away instead of listening to the phone ring.

The same logic applies to appointments, on one condition: the practice’s calendar must be connectable to the phone assistant. Alyas, for example, connects to Google Calendar or Cal.com, but not to medical appointment software. Where that connection exists, an elderly patient who does not use the internet can book a slot by voice, just as her grandson does on his smartphone.

An automated system must never assess how serious a symptom is, offer reassurance or recommend a treatment. Its instruction must be explicit: as soon as the caller mentions an urgent situation, it reminds them that 15 is the number of the SAMU, France’s emergency medical service, and that 112 can be used in every European Union country, then, if the practice has provided for it, offers a transfer to a member of the team.

For other clinical requests, its role is limited to noting the reason for the call in a few words, the caller’s identity and a callback number. The doctor then decides, with the patient’s file in hand, what to do next. This division of roles protects patient and practitioner alike.

As soon as a call is transcribed, sensitive information can end up in a summary. The best protection is not to ask for it: the script asks for the general reason, not detailed symptoms, and invites the patient to keep the specifics for the consultation. The caller must also be told from the outset that the conversation is handled by an AI and recorded.

Before any rollout, the practice would do well to check with its data protection officer, or with its professional bodies, which requirements apply to its activity: hosting location, retention period, who has access to the transcripts. This groundwork is simpler than correcting a system that is already in place.

The safest approach is not to switch everything over at once. Start with the times when the phone currently goes unanswered (lunch break, evenings, closing days), then extend to daytime overflow once the team has gained confidence. Each morning, the secretary goes through the summaries and handles callbacks in order of priority, instead of listening back through a full voicemail box.

This is a use that Alyas lends itself to: a voice agent that answers at any hour, announces itself as an AI, answers from the practice’s information, transfers to the chosen number and leaves a written summary of every call. The ten-day free trial lets you test it on uncovered time slots before deciding.