AI receptionist for clinics and salons in India: what it can and cannot do

An AI receptionist can reliably do one thing for a clinic or salon: answer the calls nobody picks up, and book, move or confirm an appointment against the calendar you already use. It should not be trusted with anything that needs judgement, such as whether a symptom is urgent, a price negotiation or a complaint. So the useful question when buying one is not how human it sounds. It is how quickly it recognises a call that belongs with a person.

Which calls is a clinic or salon actually losing?

Every business that takes bookings by phone has a number it never sees: the calls that came in while the line was busy, while the front desk was with a patient or a client, over lunch, or after closing. Most of those callers do not leave a voicemail. They ring the next clinic on the list. In India some of them are deliberate missed calls from people expecting a callback, and those count too.

Measure this before buying anything. Most phone systems and mobile carriers keep a record of missed calls. Count them for two weeks and note the time of day. If they cluster at lunchtime and after seven in the evening, that is the job. If there are hardly any, you do not need an AI receptionist, and anybody who sells you one anyway did not ask.

What can an AI receptionist do well?

• Book, reschedule and cancel against the calendar or booking system the business already runs on, so a slot it offers is a slot that exists.
• Answer the routine questions around a booking: timings, the address, which doctor or stylist is in on which day, whether a service needs an appointment.
• Send the confirmation on WhatsApp after the call, since that is where most patients and clients in India actually read their messages. (The same logic drives how a website chatbot should hand over to WhatsApp.)
• Pick up at any hour, including the hours when the desk is closed.

What should it never do?

• Triage. An automated agent on a clinic line must not decide whether a symptom is serious. If a caller describes anything that sounds like an emergency, it should tell them to contact emergency services and put them through to a person straight away.
• Give medical or treatment advice, or quote a price it has not been given.
• Argue. A caller who asks for a person, wants to negotiate or wants to complain should reach a person quickly, without the agent trying to talk them out of it.

This is why we are careful with the word receptionist. A good receptionist uses judgement all day long. What an AI agent can responsibly take on is the booking and the questions around it, and then hand everything else over.

Will it understand Hindi and Hinglish callers?

Test this yourself rather than taking it on trust. Speech recognition for Indian English, Hindi and mixed Hinglish has improved a great deal, but quality still varies by vendor, by accent and by how much noise is around the caller. Names of doctors, treatments and localities are where most mistakes happen.

Ask for a trial on your own number, then call it the way your patients or clients do: in the language actually spoken at your desk, from a busy street, asking for a doctor or stylist by name. Five real calls will tell you more than any demo video.

What happens when it cannot help?

The handover is the part to check most carefully. Ask three things.

1. During working hours, does it transfer the call to a person, and what happens if nobody answers?
2. After hours, does it take a message with the caller's name, number and reason, and send it somewhere the team will see first thing, such as WhatsApp?
3. When a person picks up, can they see what the caller has already said?

What decides the cost?

There is no honest single figure, because the cost is built from parts that differ by business.

• Call volume and length. Phone lines and speech processing are usually billed by the minute.
• The number itself, and how your existing calls are forwarded to it.
• The booking system. If it already connects to your calendar or practice software, that is quick. If the connection does not exist yet, it has to be built.
• Languages, and how much testing each one needs.
• WhatsApp confirmations, which run on Meta's per-message pricing.
• Upkeep. Timings, doctors, stylists and services change, and the agent has to change with them.

What about patient data?

A clinic's calls carry health information. India's Digital Personal Data Protection Act, 2023 sets the rules for personal data collected digitally, and call recordings and transcripts are exactly that kind of data. Before buying, ask the vendor where recordings are stored, how long they are kept, who can listen to them, and how callers are told they are speaking to an automated system. This is not legal advice, and a clinic should check its own obligations with its own adviser.

Where this studio stands

We build the voice agent for exactly this job: the calls that would otherwise ring out, booked against the calendar you already use, and handed to a person the moment the call needs one. We do not sell it as a receptionist, and that page explains why.