AI Receptionist for Clinics in India: What to Hand Over

It should answer anything your clinic can print on a notice board, such as timings, address and fees. It should take the details of requests that need a person's decision, and it should transfer everything clinical or urgent at once. The line is simple: book, route, record, never assess.

The Telemedicine Practice Guidelines, clause 5.4, say technology platforms based on artificial intelligence or machine learning are not allowed to counsel patients or prescribe medicines. Only a registered medical practitioner (RMP) may do that, and the RMP must communicate directly with the patient. That clause is written for telemedicine platforms. Our recommendation is to apply it to every phone AI in a clinic, whatever the call is about.

Write the approved text yourself, in short sentences, and keep it under your control. Anything the AI says that you did not write is a risk.

Log every inbound call for seven days and work out your missed-call rate, which is missed calls divided by total inbound calls. This takes one sheet of paper or your phone's call history. It tells you whether you have a problem worth paying to fix.

For each day, note:

1. Total inbound calls. Pass: the count matches your phone's call history.
2. Missed or abandoned calls. These are calls that rang out or that the caller ended before anyone answered. Pass: each has a number written down.
3. After-hours calls. Pass: counted separately from daytime calls.
4. Calls that only needed an approved answer. These are timings, address and fees. Pass: you can name them from the call history or a quick check with the caller.
5. Average call length in minutes. Time five typical calls. Pass: you have a figure you can use in the cost formula below.

Then call back every missed number within the hour and note how many were new patients. Our recommendation, not a measured benchmark: five or more missed calls in the week, or two or more at the same hour on different days, justifies a trial. Below that, set a callback rule and wait. Our missed call callback workflow shows how to turn each miss into a task.

If most misses fall after closing time, set the AI receptionist to take the caller's name, number and reason for a morning callback, and to play a fixed message for emergencies. It should not offer advice or promise a slot.

Providers charge either per minute of calls or per doctor, so the bill follows your call volume or your headcount. As of 10 October 2026, ConnectAI lists ₹800 a month plus ₹4 a minute on its pricing page. Patient Square lists its AI receptionist in its Autopilot tier at ₹4,999 per doctor a month, or ₹3,999 per doctor a month billed annually, plus 18% GST (Goods and Services Tax).

Patient Square's three published tiers are Assist at ₹1,999 (₹1,599 billed annually), Copilot at ₹2,999 (₹2,399 billed annually) and Autopilot at ₹4,999 (₹3,999 billed annually), all per doctor a month. The AI receptionist sits in Autopilot only. ConnectAI also lists ₹499 a month if you add the receptionist to its clinic suite, and a one-time ₹1,000 onboarding fee for new clinics.

Here is the formula for a per-minute plan. It is our model, not a quote.

Monthly cost = monthly fee + (calls per day × average minutes × working days × rate per minute)

We assumed 2.5 minutes per call and 26 working days, and used ConnectAI's ₹800 and ₹4. The table leaves out the onboarding fee. The last column adds GST at 18%, the rate Patient Square states, which is our assumption for any provider that shows a price without GST.

A per-doctor plan scales with your headcount instead. At Patient Square's published Autopilot rates, a three-doctor clinic pays ₹14,997 a month on monthly billing (3 × ₹4,999) or ₹11,997 a month on annual billing (3 × ₹3,999), before GST. With 18% GST that is about ₹17,696 and ₹14,156.

For comparison, we price our own plans on a different model: a monthly subscription plus talktime, the minutes of calling you pay for. Our Pro plan is ₹999 + GST a month with three hours of talktime a day, and extra talktime costs ₹49 + GST an hour. AI Receptionist pricing on a clinic number is quoted by our expert.

We publish this page and sell an AI Receptionist ourselves. The two providers priced above are other companies, and their figures are their own published rates.

Two things follow from the formula. First, calls per day and minutes per call carry equal weight, so every extra 30 seconds on 30 calls a day adds ₹1,560 a month before GST (390 minutes × ₹4). Keep the greeting and questions short. Second, the cost is only worth it if it recovers visits. Divide the monthly cost by your average first-visit fee. With a ₹500 fee (replace it with yours), the 30-call clinic's ₹8,600 breaks even at 18 recovered visits a month (8,600 ÷ 500 = 17.2, rounded up), ignoring GST.

Our AI Receptionist answers incoming calls, works out why the patient is calling, answers common questions, captures name, number and requirement, then transfers the call to your team or arranges a callback. Your front desk confirms the appointment slot, which keeps the booking decision with a person.

This is how we set it up for a clinic:

Paid plans include unlimited users and contacts, call recording and history, team management and analytics. Pro excludes integrations. Connecting calls to your own customer relationship management (CRM) software through custom integration is part of our Enterprise plan, which is quote-only. You can cancel anytime, with no long-term lock-in. Our guide to AI receptionists in India covers the wider picture.

Not reliably until you test it. Names, phone numbers and mixed-language sentences are where phone AI fails first, so run ten scripted calls and score them. Hinglish is Hindi and English mixed in one sentence. Test any provider this way, ours included.

Our bar, a recommendation and not a measured benchmark: at least six of calls 1 to 7 fully correct, and all of 8 to 10 correct. A single failure on 8, 9 or 10 is a stop. Use a noisy room for a few calls, since a clinic is rarely quiet. More detail is in our Hinglish AI phone agent test and in how to capture names, numbers and PIN codes.

The clinic is the data fiduciary, the party that decides why and how patient data is used, so the duties are yours even when software takes the call. The Digital Personal Data Protection (DPDP) Rules, 2025, were notified on 14 November 2025 with an eighteen-month phased compliance period. That puts the main duties around May 2027 (our arithmetic: 14 November 2025 plus 18 months). As of October 2026 they are not yet in force.

The Press Information Bureau's summary of the Rules sets out what they ask of a clinic:

• A clear consent notice stating the specific purpose for which data is collected.
• Breach notice. Affected people must be told without delay what happened and what to do.
• A contact point for data queries.
• Requests answered within ninety days. These cover access, correction, update and erasure.
• Penalties. Up to ₹250 crore for failing to keep reasonable security safeguards, up to ₹200 crore for not reporting a breach and up to ₹50 crore for other violations. These are ceilings, not typical fines.
• Children. Verifiable parental consent is required, unless the processing relates to essential services such as healthcare.

Before you sign any provider, ask five questions and get the answers in writing:

1. Where are recordings and transcripts stored? Pass: a named location and retention period.
2. Is there a recording announcement at the start of each call? Pass: you hear it on a test call.
3. Can you delete one patient's recording on request? Pass: they show you how.
4. Who owns the transcripts if you leave? Pass: the contract says you do.
5. Who do you call at 10 am on a Saturday if it goes wrong? Pass: a named contact and number.

This is general guidance, not legal advice. Ask your adviser how the Rules apply to your clinic.

Start with a seven-day audit and ten scripted calls, then try the AI Receptionist on your clinic's own number. Talk to Our Expert.