AI for Clinics: AI Receptionist, Bookings and Reminders

The phone rings while you’re taking payment from one patient, another is at the desk asking to move Thursday’s appointment, and WhatsApp is filling up with “how much is a hygiene appointment?”. More messages arrive overnight that nobody answers until morning, and by 10 a.m. there are two gaps in the diary from people who didn’t turn up. This page explains which parts of that work AI can take on in a dental practice, physiotherapy clinic, aesthetics clinic, vet practice or small medical practice, and which parts should stay with your team.

The short answer

What you can automate: answering treatment, price and opening-hours questions on WhatsApp, your website or the phone; booking, moving and cancelling appointments in your practice management software; reminders and confirmations; offering cancelled slots to your waiting list; pre-visit intake forms; post-visit follow-up and review requests.
What it must not do: diagnose, recommend treatment to a specific patient or decide whether something is urgent. Symptoms, emergencies and clinical questions go to a person on your team.
What it saves: hours of phone and messaging time at the front desk every week, after-hours enquiries that no longer go cold, and fewer empty slots because every cancellation is offered straight away to someone who wants it.
What it costs: an assistant that answers with your clinic’s information from around €1,000; an agent that books into your diary and sends reminders, €2,000–5,000; multi-system or phone (voice) from €5,000. Usage is extra, typically cents per conversation. Details in our guide to what an AI agent costs.
How to start: with bookings and reminders on WhatsApp or SMS, connected to the diary you already use. Then the waiting list and out-of-hours calls.

What you can automate

Process	What AI or automation does	Who benefits
Treatment, price and opening-hours questions	Answers on WhatsApp or web chat using the information you give it (fees, what each treatment includes, directions, parking, insurers you work with)	Front desk, new patients
Booking, rescheduling, cancelling	Checks real availability in your practice management software and books, moves or cancels the appointment	Front desk, patients
Reminders and confirmations	Sends the reminder, reads the reply (“confirmed”, “can’t make it”) and updates the diary	Front desk, clinicians
Waiting list	When someone cancels, offers the slot to waitlisted patients in order until one accepts	Clinicians, waiting patients
Pre-visit intake	Sends a secure form for contact details, consents and, where relevant, medical history, straight into the patient record	Clinicians, front desk
Follow-up and reviews	A short message after the visit, recall reminders and, if the experience was good, a Google review request	Owner, marketing
Out-of-hours calls	A voice agent answers, gives information, handles bookings and leaves the front desk a note on anything it can’t resolve	Patients, front desk
Back-office work	Tidying the clinician’s dictated session notes, drafting insurance paperwork, sorting the inbox	Clinicians, admin

An AI receptionist on WhatsApp, web chat and the phone

Before: a patient messages on Sunday night: “hi, do you have anything for a check-up this week?”. On Monday the front desk reads it between calls, opens the diary and offers two times. The patient replies at lunchtime that neither works, and it drags on until Tuesday. Meanwhile another practice answered first.

After: the agent replies straight away, says it’s the clinic’s AI assistant, asks what the patient needs, checks real availability for that appointment type and clinician, offers two or three options and books the one they choose. It confirms the appointment with the address and anything they need to bring. If the patient asks something that isn’t in the clinic’s information, or mentions pain, swelling or any symptom, the agent doesn’t improvise: it flags the conversation for the front desk with a summary.

What makes this work is the connection to your practice management software (Dentrix, Open Dental, Dentally, Cliniko, Jane or whatever you use). The agent doesn’t guess at free slots: it reads them from the system and writes the booking back, just as your receptionist would.

Appointment reminders that cut no-shows

Before: the day before, someone at the desk calls or messages tomorrow’s patients one by one, or the practice software sends an SMS nobody replies to.

After: the reminder goes out with two buttons: confirm or reschedule. A confirmation is marked in the diary. If the patient can’t make it, the agent offers another slot in the same conversation and frees up the original one. If they don’t reply, the front desk gets a short list of who to call.

We won’t quote you a no-show reduction figure: it depends on your clinic, your patients and what you already do today. What we can measure with you is your own no-show rate before and after, from your diary data.

One important detail: the reminder shouldn’t name the treatment. “You have an appointment tomorrow at 10:00 at X Clinic” is enough. Anyone can read a lock screen.

Filling cancelled slots from the waiting list

Before: someone cancels at 9 a.m. for a 12 o’clock slot. The front desk has no time to ring round, and the slot stays empty.

After: patients who asked to be told if something earlier comes up sit on a waiting list. When a matching slot opens (same appointment type, same clinician, a time that suits them), the agent offers it in order, and the first to accept gets it. If nobody answers within a time you set, it moves on to the next person. In physiotherapy, with short sessions and lots of patients mid-treatment, this tends to matter a lot.

Intake forms before the first visit

Before: new patients arrive ten minutes early to fill in their details, a privacy notice and sometimes their medical history on paper. Someone types it into the system later.

After: when they book, they get a link to a secure form that goes straight into their record. This is the most important rule on this page: health information is collected in the clinic’s form, not in the chat. The WhatsApp agent sends the link; it doesn’t ask about allergies, medication or history.

Follow-up, recalls and reviews

Before: six-month recalls depend on someone remembering to call. Google reviews arrive on their own, and it’s usually the unhappy ones.

After: a short message after the appointment asks how the patient is doing. If the answer is positive, they’re invited to leave a review. If they report a problem, there’s no review request: the clinic is alerted so a person can call them. When the recall is due, the agent offers a date. Note that care follow-up and marketing messages (offers, reactivating lapsed patients) have different legal bases; check the second with whoever handles your data protection.

Out-of-hours calls with a voice agent

Plenty of patients would rather phone, especially older ones. A voice agent can pick up when the clinic is closed or every line is busy: it gives information, handles bookings in your diary and, for anything it can’t resolve, takes a note for the front desk next morning. If someone calls with an emergency, the agent doesn’t assess it: it gives the clinic’s emergency number if you have one, and the national emergency number (112 in the EU, 999 in the UK) if it’s an emergency.

Back office: notes and insurance paperwork

Away from patient contact, AI helps clinicians turn dictated notes into a clean write-up, draft reports and the paperwork insurers ask for, and summarise a patient’s history before a visit. Always as a draft the clinician reviews and signs off. Data protection requirements here are at their highest, which brings us to the risks.

What a typical rollout looks like

We look at how you work. Which practice management software you use, which channels patients use, how many enquiries you get, which questions repeat, and which cases must always reach a person.
First process: bookings and reminders, connected to your diary. At first the agent can propose and the front desk approve; as it proves it gets things right, it gains autonomy.
We write down your clinic’s information: treatments, prices or price ranges, what the first visit includes, insurers, cancellation policy. The agent answers only from that.
We measure: enquiries resolved without staff, bookings made by the agent, conversations handed to a person and why, no-shows before and after.
Next: waiting list, intake forms, follow-up and reviews, and the voice agent if phone calls are your bottleneck.

What still needs a person: any clinical question, quotes for complex treatment (an orthodontic or implant plan doesn’t get closed on WhatsApp), complaints, patients who ask to speak to someone, and anything the agent doesn’t recognise. The point is to take the repetitive work off the front desk so they have time for exactly that.

If your diary software can’t be connected (not all of them offer an API, the door that lets another program read and write appointments, and some only open it to partners), we find out before we start. There are workarounds, but they change the scope and the price.

The risks of doing it yourself

In a clinic, a badly built agent doesn’t just give a wrong answer: it can expose your patients’ health data. These are the things that have to be solved:

Health data is a special category. Article 9 of the GDPR prohibits processing it except in specific cases, such as healthcare provided by professionals bound by confidentiality, or the patient’s explicit consent. The mere fact that someone has an appointment at your dental practice says something about their health. A chatbot wired to an AI model outside the EU, on a trial account, with no data processing agreement, is exactly what mustn’t happen. (Outside Europe the rules differ, for example HIPAA in the US, and the provider choices change accordingly.)
Minimal data, kept out of the chat. To book an appointment the agent needs a name, a phone number and the appointment type. It doesn’t need diagnoses, medication or history. If a patient types those anyway, the system shouldn’t keep them longer than necessary or reuse them. And reminders don’t name the treatment.
Processing in Europe, with no retention. Which model processes the conversations, where, and whether the provider stores them or trains on them. We cover it in our guide to EU data residency for AI models.
Saying it’s an AI. Since 2 August 2026, Article 50 of the EU AI Act requires people to be told they’re talking to an AI. More in our EU AI Act guide for businesses.
No triage. Deciding whether a symptom is urgent is a clinical decision. On top of that, Annex III of the AI Act lists emergency healthcare patient triage systems as high-risk. A booking agent shouldn’t rank anyone by severity: if there’s a symptom, it hands over to a person.
Wrong answers about treatments. A language model can state something false with full confidence: a price that isn’t yours, that a treatment “doesn’t hurt at all”, that something clears up in one session. The agent must answer only from information the clinic has approved, and say “the clinician will confirm that” when it doesn’t have it.
The diary integration. Double bookings, appointments created with the wrong clinician, or a silent failure when the practice software doesn’t respond. If nobody is monitoring it, you find out when a patient turns up and isn’t in the diary.
WhatsApp done the wrong way. For clinic reminders, the official WhatsApp Business API is almost always the right choice: it’s the route Meta supports, and it protects the number your patients know from being blocked. The differences are in our Evolution API vs WhatsApp Business API comparison.

A partner who has already put agents into production designs all of this in from the start: which data goes in and which doesn’t, where it’s processed, what gets answered and what gets handed over, and how a failure is caught before a patient notices it.

Want an agent for your clinic without these risks? Tell us how your clinic works on a free call.

How we do it at Aitomatik

We’re an AI and automation agency based in Spain, working in English and Spanish with clinics and businesses in Spain, across Europe and beyond. For a clinic, this is how we work:

Data in Europe: for European clients, the models we use are processed in the EU, with zero data retention and no training, plus a backup model from a different provider, also in the EU, in case the main one fails. Clients outside the EU aren’t limited to EU models.
We save and evaluate every conversation the agent has, and test every change against real conversations before it ships. That’s how we know what it gets wrong and fix it with data.
A person can always take over. The agent hands over to the front desk with a summary of the conversation. If you want, we add an optional dashboard so your team can see conversations and step into any of them instantly (the agent stops); it costs more, but pays off at high volume.
What goes out automatically and what a person reviews. To decide whether an AI-drafted reply can be sent on its own or needs review, we use decision models such as Jev, from TypeSafe, which returns probabilities instead of text. We run it in production to classify messages from our own contact form. In a clinic: “what time do you open on Saturday?” goes out automatically; a complaint, a question about a symptom or a billing dispute goes to a person.
WhatsApp the way that fits: the official API for clinic reminders and patient messaging; for other uses we also work with Evolution API, as in our WhatsApp AI agent for a domestic staffing agency.
You usually own the code, as agreed per project.

Agents that answer customers and look up bookings already run in production for us: in a phone and chat support project we cut human work by 80%, with a voice agent in four languages that looks up the customer’s booking in the system and transfers the call to a person when needed. For reviews, see our review and reputation management service.

For the basics behind all of this, read what an AI agent is. Our AI customer service covers web chat, WhatsApp and voice, and back-office work (notes, reports, insurance paperwork) falls under document automation. More in our guides and all our services.

Frequently asked questions

How much does an AI receptionist cost? Our projects start at around €1,000 for an assistant that answers with your clinic’s information, €2,000–5,000 for one that books into your diary and sends reminders, and from €5,000 with voice or several systems. Usage is extra: cents per conversation, plus WhatsApp or phone charges.

Is AI replacing receptionists? In a small clinic, it usually takes the repetitive part (the same questions, moving appointments, reminders) so your receptionist has time for the patient in front of them, the difficult calls and treatment quotes.

Can AI give medical advice or diagnose patients? It shouldn’t. A patient-facing agent gives administrative information (prices, hours, appointments, general visit preparation you’ve written down) and passes any clinical question to a clinician. Tools that support clinical decisions are a different category, with medical device requirements.

Is there a free AI receptionist? Many practice management systems include a basic assistant, and if that covers you, use it. A custom agent makes sense when you want it to answer with your own information, work on WhatsApp and the phone, run your waiting list and hand over properly to your team, connected to the software you already use.

Is it legal to use an AI chatbot with patients? Yes, done properly: a clear notice that it’s an AI, a data processing agreement with every provider, processing in the EU for European clinics, and no clinical data in the chat. Check with your data protection officer whether you need a data protection impact assessment.

Does it work for physiotherapy, vet and aesthetics clinics too? Yes, the pattern is the same: questions, bookings, reminders, waiting list and follow-up. In physio, courses of treatment and the waiting list carry the most weight. In vet practices, the record belongs to the pet, and vaccinations and check-ups make reminders especially useful. In aesthetics, packages, sessions and campaigns matter more, and campaigns have their own consent rules.

Shall we talk about your clinic?

Tell us which diary software you use, how patients get in touch and what takes up most of your front desk’s time. We’ll tell you what we’d automate first, how we’d handle health data properly and what it would cost. Book a free call.

Guide by Aitomatik. This isn’t legal advice. Last reviewed: 15 October 2026.

AI for clinics: an AI receptionist that books, reminds and fills cancellations