A clinic came to us with a front-desk problem: clinic appointment requests that arrived at all hours and rarely complete. They ran a practice whose front desk opened each morning to messages left overnight: do you take walk-ins on Saturday, is the doctor in on Thursday, can I book for my mother, how much is a consultation. Many were appointment requests missing half the details, and each turned into a back-and-forth the receptionist handled between patients, copying dates by hand.
What they told us
The clinic wanted two things. Practical questions — hours, services, which doctor is in on which day — answered from what the clinic had already written. And appointment requests collected with every detail the front desk needed, so a request could be confirmed without chasing the patient.
They were equally clear about what the assistant must never do: give medical advice, invent a service or a price, or tell a patient an appointment was confirmed when nobody had confirmed it.
How we approached it
A clinic is a place where a chatbot's mistakes have consequences, so we started from the limits and built inward. We planned around these requirements:
- Information comes only from what the clinic has written. A question it does not cover gets an honest "I don't have that", and the conversation can go to a person.
- A request is collected in the clinic's own terms, with the questions the clinic chooses, and stored where the front desk can review it.
- The assistant collects requests. It does not book into the clinic's calendar, and it never claims a booking it cannot see.
- The clinic, not the model, decides which disclaimers appear.
How it works
Requests, collected properly. With Data Forms, the clinic writes the questions it needs answered — the patient's name, the service, a preferred date and time — and the assistant asks them in the chat. Date and time questions use pickers, and a choice question can list the sessions the clinic offers, such as morning or afternoon. Every request lands as a row the front desk can browse and export, and the front desk confirms each one with the patient.
Information, not advice. The assistant answers from the clinic's own knowledge and nothing else.
The clinic decides on disclaimers. AI models tend to append their own "consult a professional" lines. A clinic can switch off the disclaimers the model writes on its own, while every disclaimer in its own knowledge stays exactly as written.
Its patients' languages. The questions are asked in the language the patient writes in, with no per-language setup.
Where it leaves the front desk
Overnight questions are answered before the doors open, and appointment requests arrive with a name, a service and a preferred time. The receptionist still confirms every one with the patient. The assistant collects requests; the clinic decides on them.
Frequently asked questions
Can a chatbot take clinic appointment requests?
Yes. With Data Forms the assistant collects the details you need — name, service, preferred date and time — and each request becomes a row your front desk can review. It collects requests; your team confirms them.
Will the chatbot give medical advice?
No. It answers only from the information the clinic provides, and says when a question is not covered.
Does the chatbot confirm the appointment?
No. It collects the request with the details you asked for; your front desk confirms it with the patient. It never claims a booking it cannot see.