InsightsCustom automations

What an AI receptionist actually does in a dental practice

Not a chatbot on the website. A member of front-of-house staff that answers the phone at 8am, fills the cancellation at 11:20 and never forgets to send the forms.

6 min readBenjamin Yang

Most dental practices we speak to have the same Tuesday. The phone rings while the receptionist is checking in a patient. A cancellation opens a slot that nobody fills. Someone arrives without having done their medical history, so the appointment starts ten minutes late and the day never quite recovers. None of this is a technology problem, exactly. It is a capacity problem: one person cannot be on the phone, at the desk and in the diary at the same time.

That is the job an AI receptionist does. Not a chatbot bolted onto the website, but a colleague who takes the overflow and the repetitive parts of front-of-house so the person at the desk can look after the patient in front of them.

The call at 08:52

A new patient rings with a cracked tooth. The agent answers on the second ring, says it is the practice's AI assistant, and asks what has happened. It captures the symptoms, whether the patient is registered, how they are paying, and when they can come in. It knows the practice's rules: emergencies are seen the same day where possible, new patients need a consultation before treatment, and anyone in severe pain who cannot be seen is given the out-of-hours number.

What it does not do is diagnose. The line between triage and clinical advice is one the practice draws in the agent's instructions, and the agent stays on the right side of it. If a caller asks whether they need a root canal, the answer is that the dentist will assess that on the day.

The slot at 11:20

The booking agent watches the diary. When a cancellation lands, it knows who is waiting and who asked to be moved earlier. The patient from 08:52 is offered 11:20 with the dentist who handles emergencies on Tuesdays. The confirmation goes out by text with directions, parking notes and what to bring. If the patient does not confirm within an hour, the slot is offered to the next person on the list.

Empty slots are where the money leaks. Across the practices we have worked with, the share of cancellations that get refilled is the single number that moves most in the first month.

The forms at 09:30

Between booking and arrival, the intake agent sends a link to the medical history and consent forms. It follows up once if they are not completed. When they are, it files them against the patient record in the practice management system, flags anything the dentist should see before the appointment (a new blood thinner, a penicillin allergy), and leaves a note for the desk that everything is in order.

The patient walks in at 11:15 and is in the chair at 11:20. That is the whole point.

Payment, then the follow-up

After treatment the dentist logs what was done. The billing agent raises the invoice, takes the card payment or applies the plan, and prices the next visit so the patient leaves knowing what comes next. Two days later the recall agent asks how the appointment went. A happy patient is invited to leave a Google review. An unhappy one is routed to the practice manager before anything goes public, which is the kind of judgement you want a receptionist to have.

What the practice gives up

Very little, but not nothing. The agent needs the practice's rules written down, which takes an afternoon with the practice manager and usually surfaces a few rules that were never agreed in the first place. It needs to be connected to the phone system, the diary and the practice management software, which we do through their official integrations. And it needs a fortnight of running in shadow mode, where it listens and drafts but does not act, so the team can see what it would have done and correct it.

It also needs boundaries the practice is comfortable with. Every agent we build identifies itself as an AI, hands over to a person the moment someone asks, and never does anything irreversible (a refund, a cancellation of a treatment plan) without a human saying yes. For a CQC-registered practice, patient data stays in the UK, is never used to train models, and is deleted when the practice leaves. Those are not features. They are the terms on which a practice should be willing to let software talk to its patients at all.

What changes

The receptionist is still there. She is just no longer the bottleneck. The phone gets answered, the diary stays full, the forms are done before the patient arrives, and the practice manager finds out about a bad experience before Google does.

Share post

More insights

AI roadmap

Stop buying AI tools before you know the job

The average small business we meet is paying for four AI subscriptions and using one. The problem is not the tools. It is buying them without a job in mind.

5 min readBenjamin Yang

Talk to us about your Tuesday.

A free 45-minute call. You tell us where the week goes; we tell you honestly whether an agent would help and roughly what it would save.

Book a call