A WhatsApp chatbot for clinics: how it answers patients and books without staff, and when it must stay quiet
What a clinic WhatsApp chatbot can safely answer and book, what must go to a person, and how to stop it inventing prices and dates.
We run WhatsApp bots for clinics, on our own platform and on custom systems, and the first thing we learned is that the bot that "answers everything" is the most dangerous bot you can install in a clinic.
A patient asks "is my case suitable for the surgery?" and the bot answers confidently. Or asks "how much?" and the bot quotes last year's number. Or confirms an appointment that does not exist. Each of these costs more than the bot saved in a month.
This post is about how a clinic WhatsApp chatbot works properly: what it does, what it does not, and how to build it so it stays quiet when it should.
What the bot does extremely well
The repeated questions with a fixed answer
Opening hours, location with a map link, whether a given insurer is accepted, which services you offer, how long a session takes, how to prepare for a visit. These make up most of a clinic's messages, and the answers fit on one sheet. The bot answers instantly and in the patient's language, Arabic or English, in the dialect people actually write in rather than formal Arabic.
Booking and confirmation
It collects what the clinic needs (name, visit type, doctor, preferred day), shows the times that are genuinely free from the schedule, books, and confirms with a reference. Most importantly: the times come from the system, not from the bot. The bot never invents a time. We covered this in a clinic booking system in Saudi Arabia.
Reminders and follow-up
The day before: a reminder. The patient replies "confirm" or "cancel". After the visit: a follow-up message or a reminder of the next review. All through Meta-approved templates.
Handover to a person, with a summary
When a question falls outside what it knows, or there is a complaint, or something urgent, the bot hands over to an employee with a summary of the conversation. The employee does not start from zero, and the patient does not repeat themselves.
What it must never do
- Diagnose or decide the treatment. "Am I suitable for LASIK?" has one answer: "the consultant decides after the examination", then handover.
- Promise an outcome or a success rate. Strictly forbidden, even when the patient pushes.
- Quote unconfirmed prices. If the price is not in the bot's current information source, it gives no number. It hands over.
- Confirm an appointment that is not in the schedule. Every confirmation comes from the system, not from conversation.
- Discounts, hardship cases and complaints. A human. Always.
- Sensitive medical topics. Any message describing severe pain or an emergency: immediate handover, with direction to emergency services if warranted.
The rule we write into every clinic bot: if you do not have confirmed information, hand over. A wrong price or a wrong date costs the clinic.
How to stop it making things up
This is the technical part, and it is what separates a bot that helps from one that harms:
- One up-to-date source of information. The bot answers from the clinic's own file (services, prices, hours, policies) that the clinic itself updates. Not from the internet, not from "memory".
- Booking through tools, not through conversation. The bot asks the booking system for free times and books through it. If the tool says "nothing free", the bot says nothing is free; it does not suggest something of its own.
- An "I do not know" signal. The bot is instructed to say "a colleague will be with you shortly" instead of guessing, and to actually hand over.
- A reply cap when the employee is busy. If a conversation is with an employee who has not replied, the bot covers with a limited number of replies, then goes quiet and raises a flag, rather than turning into a fully automated conversation.
- A weekly review of conversations. In the first month, someone at the clinic reads a sample and corrects the replies that did not land.
Dialect matters
A bot replying in formal Arabic to a patient who wrote "hey, I want to book with the doctor tomorrow" feels like talking to a machine. Short replies in dialect, with no emoji and no repeated sign-off phrases, keep the conversation natural. This is set in the bot's instructions from the start, not fixed later.
What you need to begin
- The official WhatsApp Business API on the clinic's number. The regular app does not support bots. We explained the difference in WhatsApp API or the WhatsApp Business app.
- An up-to-date clinic information sheet: services, prices, hours, doctors, cancellation policy, insurers.
- A booking system the bot can read from and book into. If you do not have one, that is the first thing to build.
- A responsible employee who receives handovers and reviews conversations at the start.
How long, and how much
Verification at Meta is what takes time, and we start it on day one. Building the replies and connecting the booking is faster. Cost has two parts: building and integration, and Meta's per-message charges by message category and your volume, which we work out on your numbers rather than a generic example. We do not print numbers here because they change.
The short version
A clinic WhatsApp chatbot earns its place when it answers the repeated questions, books from the real schedule, reminds, and hands over to staff with a summary. It does not when it diagnoses, promises, or invents a price or a date. The difference is not intelligence, it is boundaries.
See the WhatsApp API and chatbot page and clinic solutions, or book a free consultation and we will look at your clinic's messages and tell you which ones a bot can cover.