AI for medical and dental clinics: appointments, calls and GDPR

The phone rings during a consultation. Reception is busy with a patient who is paying. Someone else messages on WhatsApp at 10 pm wanting to book a scale and polish. By morning the list of unread messages is long, and some of those patients have already called the clinic across the road. If you recognise the scene, this article is for you.

Below you will find, concretely, where AI makes sense in a medical clinic or dental practice, where it has no place, how to handle health data and what the cost logic looks like. No big promises, just what works in practice.

What AI in a medical clinic really means

When we say "AI for medical clinics", we are not talking about software that makes diagnoses. We are talking about a digital employee for the admin side: it answers the phone, takes bookings, sends confirmations, answers repetitive questions about prices, opening hours and how to prepare for a test, then hands everything involving a medical decision to a person.

At Neurotech we build these systems on Claude, the AI model developed by Anthropic. The agent speaks naturally in the patient's language, understands what they want and works with your real data: the doctors' calendars, your list of services, prices and clinic rules.

Where it actually helps: 6 everyday situations

1. Calls missed during opening hours

An AI voice agent answers every call, including when reception is busy or after closing. It checks availability in the calendar, offers two or three slots and confirms the appointment. If the caller explicitly asks for a person or the situation is sensitive, the call is transferred or the patient is called back.

As a real benchmark: for a chain of cake shops in Sibiu, the voice agent we built recovers around 20 calls a day that used to be lost. The sector is different, but the mechanism is identical: every unanswered call is a customer who rings somewhere else.

2. Bookings and rescheduling on WhatsApp

Many patients prefer to write. A WhatsApp chatbot takes the request, asks which service the patient wants, with which doctor and at what time, then books directly. Rescheduling and cancellations work the same way, without reception rewriting the calendar by hand.

3. Confirmations and reminders

The day before, the patient receives a confirmation message. If they reply that they can no longer come, the slot is freed immediately and can be offered to someone on the waiting list. This cuts gaps in the schedule without staff making a single phone call.

4. Repetitive questions

"How much is a panoramic X-ray?", "Do you have parking?", "Are you open on Saturdays?", "Do I need to fast before my blood test?". The answers are the same every time, and the agent gives them exactly as you have approved them, from the clinic's own documents.

5. Recall for regular check-ups

In a dental practice, six-monthly check-ups and regular hygiene appointments bring in a significant share of revenue. An automation flow sends reminders at the right time and books positive replies directly.

6. Order in the back office

Beyond patients, there are service records, treatment plan quotes, instalment payments and stocks of consumables. An internal app can bring all of this into one place, instead of five Excel spreadsheets and a notebook.

What AI must NOT do in a clinic

This is the most important part. A well-built system has clear limits, written into the agent's instructions and tested before launch:

• It does not give medical advice. It does not interpret symptoms, recommend treatments or say "it's probably nothing serious".
• It does not do clinical triage. If the patient describes an emergency, the agent tells them to call 112 and alerts staff immediately.
• It does not interpret results from tests or imaging. It can only pass them to the doctor through the agreed channel.
• It does not promise treatment outcomes or compare doctors with one another.

The simple rule: AI handles the logistics, the doctor handles the medicine. When a conversation crosses the line, the agent says openly that it cannot answer and hands the case to a person.

Health data and GDPR: what you need to know

The General Data Protection Regulation (GDPR) treats health data as special categories of data, under Article 9. Processing it is prohibited as a general rule and only allowed under the conditions the regulation explicitly sets out. For a clinic, this means the AI system has to be designed with data protection in mind from the start, not bolted on at the end.

In practice, here is what we look at in every implementation:

• Data minimisation. A booking needs a name, phone number, service and time slot. The agent does not ask for or store descriptions of symptoms unless necessary.
• Storage in the EU. The database runs on Supabase, on servers in the European Union, with role-based access.
• Restricted access. Reception sees the appointments, the doctor sees what they need, and the conversation history is only accessible to authorised people.
• Transparent information. The patient learns from the outset that they are talking to an automated assistant and where to find the privacy policy.
• Documentation. You work with your data protection officer or a lawyer on the lawful basis, processing agreements and impact assessment, where applicable.

You can read more about how we handle data on our data security page. One thing needs saying clearly: we build the technical system; the clinic's legal compliance remains the responsibility of the clinic and its advisers.

What an implementation looks like, step by step

1. Initial call. We find out how many calls and messages you receive, which booking system you use and where most time is lost.
2. Conversation map. Together we write down the types of request: new booking, rescheduling, cancellation, price question, emergency, document request.
3. Knowledge base. We gather the list of services, prices, doctors, opening hours, address and preparation rules. You approve every answer.
4. Calendar integration. The agent reads and writes directly in your booking system, or in a new app if the current one does not support integration.
5. Internal testing. Your team calls and writes in as difficult patients: rushed, undecided, asking for medical advice. We adjust until the answers are right.
6. Gradual launch. We usually start with out-of-hours calls, then expand.
7. Monitoring. We read the real conversations from the first few weeks and fix whatever is not working.

What AI for a clinic costs

Starting prices for implementation, excluding VAT, paid in instalments tied to deliverables:

Monthly costs (infrastructure, usage, maintenance) depend on volume: how many call minutes, how many messages, how many users. We work them out from your real figures, not general estimates.

The cost logic is simple. Write down how many calls you miss a week and the average value of an appointment. If a single lost consultation or piece of dental work covers a large part of the monthly cost, the sums are quick to do.

Common mistakes to avoid

• An agent without clear limits. An assistant that tries to be "helpful" with medical questions is a risk. Limits are written down and tested.
• Out-of-date prices. If the price list changes and nobody updates the agent, patients get wrong information. Decide who is responsible for this.
• No route to a person. Patients must be able to ask for a person at any time. Otherwise frustration builds.
• Collecting too much data. Every extra field is an extra responsibility.
• Everything at once. Start with one clear problem, usually missed calls, and expand once you see results.

A quick checklist before you start

• Do you know how many calls and messages you get a day, and how many go unanswered?
• Do you have a list of services and prices you can approve?
• Does your booking system support integration, or do you need a new one?
• Have you decided who receives the cases the agent hands over?
• Have you spoken to your data protection officer about the new channel?

Next step

If you would like to see how an AI assistant would work with your clinic's processes, book a free 30-minute call. We will tell you honestly where it makes sense to start and where the investment is not worth it.