ProductUse casesPricingBlogContact
Dashboard Sign in Start free
Industry Applications

AI voicebots in healthcare: a website assistant for clinics

AI voicebots in healthcare: a website assistant for clinics

AI voicebots in healthcare work best at a clinic’s front door: they answer how to book or cancel an appointment, when the clinic is open and how to prepare for a test, and they hand everything medical to staff. The hard part is not the answers but the limits. This guide covers which patient questions to automate, which health data the bot must not collect, and the routing rules that send patients to a person at the right moment.

Picture a patient on your clinic’s website late in the evening. Her blood test is tomorrow morning and she is not sure whether she may drink coffee before it. Reception is closed. If the website can answer her out loud in a few seconds, from instructions your staff approved, she arrives prepared and the test does not have to be repeated.

Which clinic questions can a voicebot answer?

A clinic voicebot can answer administrative questions that have one approved answer for every patient. Medical questions, which depend on the person, stay with medical staff. In practice the safe list looks like this:

  • Appointments: how to book, change or cancel, which specialist sees which kind of case, what the cancellation rules are, how to get a referral.
  • Opening hours: regular hours, holiday closures, hours of the blood collection point, where to go when the clinic is closed.
  • Preparation instructions: fasting before tests, what to bring to a first visit, how early to arrive, which documents are needed.
  • Practical details: address, parking, public transport, accessibility, how to collect results, how invoices work.

Preparation instructions deserve special care. Write them once, have them approved by medical staff and give each one a review date. A bot repeats whatever is in its sources, so an outdated fasting rule becomes an outdated answer for every patient who asks.

Phone line or website: where the voice assistant works

AI voicebots in healthcare come in two forms: phone voicebots connected to the clinic’s telephone system, and voice assistants on the clinic’s website. Phone systems need telephony integration and usually a link to scheduling software; the article on bots answering clinic calls describes what changes when the caller is a patient.

A website voice assistant is simpler to run. The patient opens the clinic’s page, presses a button and speaks; the assistant answers by voice in the browser, from the clinic’s own materials. It does not replace the phone line, but it takes the questions that never needed a phone call in the first place.

Health data is special category data: what not to collect

Treat anything a patient says about their health as sensitive. Article 9 of the General Data Protection Regulation (GDPR) lists data concerning health among the special categories of personal data, whose processing is prohibited unless one of the exceptions in that article applies. For an information bot the simplest rule follows: design it so it does not need health data at all.

  • Do not ask about symptoms. The bot answers “how do I prepare for a blood test”, not “tell me what is wrong”.
  • Do not ask for identifiers. No national ID numbers, insurance numbers or dates of birth; an administrative answer works without knowing who asks.
  • Do not put patient records in the knowledge base. Upload procedures, price lists, opening hours and FAQs, never medical documentation, appointment lists or test results.
  • Warn patients gently. The welcome message can say that the assistant answers general questions and that personal health details should go to staff.
  • Settle what happens to conversations. Decide with your data protection officer how long conversation logs are kept and who can read them.

Patients will still sometimes mention a diagnosis or a medicine. Plan for it: the bot should not build on that detail, it should point to the right contact. Your chatbot privacy notice should tell patients that a bot answers, what happens to the conversation and how to reach a person instead. For a wider list of what must stay out of the sources, see the guide on what never belongs in a bot’s sources.

When should AI voicebots in healthcare route patients to staff?

A clinic bot should route a patient to a person whenever the question depends on that patient’s health, needs a decision or is not covered by approved materials. Write these rules down before launch, so the bot does not improvise.

Situation What the bot does
Signs of an emergency, such as chest pain or breathing problems Tells the patient to call the emergency number at once, without asking further questions
Questions about symptoms, medicines or test results Explains that a doctor, nurse or pharmacist must answer and gives the contact
Changing or cancelling a specific appointment Explains the rules and points to the registration phone or online booking
Complaints or a patient who is upset Gives the contact of the person who handles complaints
A question the materials do not cover Says it does not know and gives the reception contact

The last row matters more than it looks. A bot that says it does not know is far safer in a clinic than one that fills a gap with a plausible guess. Test every routing rule with staff, including emotional and awkward questions, before patients use the assistant.

How to start with a voicebot in a clinic

Start with administrative questions only and widen the scope when the answers prove reliable. A small, clearly described scope is easier to check and safer for patients.

  1. Collect the questions reception hears most often over a few weeks.
  2. Write approved answers together with the person responsible for patient information, and have medical staff approve every preparation instruction.
  3. Write down the routing rules from the table above, with real contacts.
  4. Tell patients in the first sentence that they are talking to a virtual assistant.
  5. Test with staff, then launch on the website and review unanswered questions every week.

How Botino can support a clinic website

Botino adds a chat assistant and, on the Professional plan, a voice assistant to a clinic’s website with one line of code. It answers only from the documents, FAQs and pages you provide, in the languages you configure, and you set its scope so that medical topics are routed to staff. It works in the browser on your website: it does not answer phone calls, book appointments or connect to medical systems, which keeps its role narrow and easy to check.

FAQ

Can a voicebot give medical advice?

It should not. A clinic voicebot can explain procedures, opening hours and preparation, but questions about symptoms, treatment or results must go to a medical professional.

Is it legal to use a voicebot when patients may mention health details?

Health data is a special category under GDPR, so design the bot not to ask for it, keep patient records out of its sources and agree the setup and log retention with your data protection officer before launch.

Which questions should a clinic automate first?

Opening hours, booking and cancellation rules, preparation for common tests and directions. They are frequent, non-medical and have one correct answer for every patient.

A good first step is a list of the questions your reception answers most often, with the purely administrative ones marked. Turn those into approved answers, add the routing rules, and test AI voicebots in healthcare on that narrow scope before you widen it.