WhatsApp chatbot for clinics in English and 18 Indian languages

A multilingual clinic chatbot has to speak Tanglish, not just Tamil

Patients do not message clinics in textbook language. A Tamil-speaking patient types "Doctor inniku evlo manika varuvar?" in Roman letters. A Hindi-speaking patient types "abhi number kab aayega?". An elderly patient types in Tamil script. A young professional writes in English. Often the same patient switches mid-conversation. A chatbot built on a menu of English buttons fails every one of these people, and the ones it fails are exactly the ones who then phone the desk.

TOKENN reads the message as written, works out what the patient wants, and replies in the same language and style. Tamil script gets Tamil script. Tanglish gets Tanglish. If you would rather every patient hear from the clinic in one language, you can fix the reply language in settings; otherwise it follows the patient.

That holds for all 19 languages, not just the two most clinics ask about first. A Bengali message gets Bengali, an Odia voice note gets Odia, a Kannada patient in a Chennai clinic gets Kannada while the patient before her got Tamil. The clinic does not choose a language per patient or set anything up per language; the assistant works it out from each message and remembers it, so the reminders later in the day arrive in the same language.

The 19 languages the clinic chatbot replies in

These are the languages with a complete set of written replies: every booking confirmation, status answer, cancel, reminder and button. The assistant understands more than these, but a patient writing in one of these 19 gets every reply in it, not a first line in their language and the rest in English.

The same status reply, in the patient's language

A few words stay in English in every language: Token, Reception, Clinic, Status, Appointment and Doctor. That is deliberate. They are the words printed on the clinic's signs and on the QR at the door, and the words patients already use when they speak, so a Tamil reply says "Token" and "Reception" in English letters rather than a formal translation nobody says aloud. Everything else is in the patient's own language and script, as in this reply to "status" for token A007, with two patients ahead and A005 being seen:

A WhatsApp bot for doctor appointments: what patients type, and what it does

The job is narrow on purpose. Here is the whole of it.

A clinic chatbot in Tamil: the status reply, in Tamil script

When a Tamil-script patient asks where they are in the queue, the reply is Tamil, not a translation bolted on afterwards: "உங்கள் Token A007. உங்களுக்கு முன் 2 பேர், சுமார் 10 நிமிடம். இப்போது A005 பார்க்கப்படுகிறது." That is their token, two people ahead, about ten minutes, A005 being seen now. The English patient gets the same information in English. Neither has to ask twice.

The 2-ahead reminder, the your-turn alert and the appointment reminder follow the patient too. If the patient has messaged the clinic in the last 24 hours, which is almost everyone who booked that day, the alert arrives in their language, with the same buttons. If they have not, WhatsApp only allows a pre-approved template, and that goes in English, so the alert still arrives however long ago they wrote.

What the chatbot will never do

It never diagnoses. It never suggests a medicine or a dose. It never tells a patient whether their symptom is serious. Those questions go to a human, and the chatbot says so plainly. This is a hard boundary, not a setting, because a clinic's WhatsApp number is a clinical channel and the assistant is a receptionist, not a doctor.

Every conversation is visible to your staff on the dashboard, every token can be changed or cancelled by reception with one tap, and the receptionist can always step in on the same WhatsApp thread.

A finished clinic chatbot, not a chatbot toolkit

WhatsApp business platforms sell the messaging pipe and a flow builder. You still have to design the queue: token numbering, wait estimation, the 2-ahead trigger, the display, walk-ins, cancellations, the daily limit. That is months of work and it is the part that goes wrong. TOKENN is the finished clinic assistant: the queue logic, the languages, the dashboard and the TV display are already built, and the chatbot is the front of it.

It runs on Meta's official WhatsApp Cloud API on your clinic's own number. If you already have a WhatsApp Business account, connecting takes about ten minutes; if not, we do the Meta setup with you on a call. The setup guide covers both paths.

What a clinic does configure is short and takes about twenty minutes: the hours each queue opens and closes, the days it is shut, how far ahead patients may book, how many patients ahead should trigger the reminder, and whether replies follow the patient's language or stay fixed in one. Departments come later, if a second doctor or a lab counter needs its own line. There is no flow to draw, no intents to train and no script to write, because the conversation a clinic queue needs is the same conversation in every clinic.

Where the chatbot hands over to a human

Three places. A medical question, always. An unclear request after one clarifying question. And anything the clinic has not told it, such as a doctor's leave day it does not know about. In each case the patient is told to contact reception, and the receptionist sees the thread. The point is not to replace the front desk; it is to take the repetitive four-fifths of messages off it so the desk can handle the rest properly.

For patients who will never message, nothing changes: reception issues a walk-in token in two taps and it joins the same queue.

A Hindi WhatsApp chatbot for doctors, tested in your patients' shorthand

Before going live, send the chatbot the exact phrasing your own patients use, not textbook sentences. In a Tamil clinic that means "doctor irukkangala inniku?", "vera doctor irukaangala?" and a bare "token venum" with no other context. In a Hindi clinic, "doctor available hain kya", "mera number kab aayega" and "cancel karna hai" on its own. A chatbot that handles the textbook form but stumbles on the shorthand quietly frustrates exactly the patients it was built for, so five minutes of testing in real phrasing before the first OPD is worth more than any demo.

Misspelled, one-finger, three-language messages

Owners often ask about the patient who types with one finger, misspells half the words and mixes three languages in one sentence. For an Indian clinic that is the everyday case, not the edge case, and it is what the assistant is built for: a garbled but recognisable "docter irukanga evlo neram" is understood the same as a cleanly typed one. If a message genuinely cannot be classified, the assistant asks one short question rather than guessing, so a confusing message never silently books the wrong thing.

What to tell reception before launch

The most common launch mistake is not telling staff what changes for them. Reception should know three things before the first patient books: they can still issue a walk-in token exactly as before; every WhatsApp booking appears on the dashboard the instant it happens, so there is no separate list to check; and when the chatbot hands a message over, it appears flagged on the dashboard, not buried in a chat log. Clinics that skip this five-minute briefing have a rocky first day, not because the chatbot fails but because staff do not yet trust it and double-check every booking by hand. By the end of the first week, after a few hundred correct tokens, the habit fades on its own.

What a clinic WhatsApp chatbot costs

The chatbot comes with every plan of booking credits. WhatsApp messages are included in each booking; there is no separate message bill. There is no per-language charge and no hardware. Start with 50 free booking credits, then 150 for ₹149; credits never expire, so you can watch it handle a real morning of Tamil, Hindi and English before paying anything. Details are on the pricing page.

See how it fits a pediatric practice, where parents are the ones messaging.

Frequently asked questions

Try TOKENN for your clinic

50 free booking credits · plans from ₹149 · credits never expire · no hardware · set up in 10 minutes. See plans.

Start with ₹149 · See a demo on WhatsApp

See also

All features →

Polyclinic software for multi-doctor clinics and group practices: a live queue per doctor →

Token system vs appointment system: which one fits each doctor →

Gynecology clinic queue management on WhatsApp: numbers, never names →