Home › Blog › AI in a doctor's clinic

AI in a doctor's clinic in India: what is safe to hand over and what is not

Akash Gupta Founder, AI Gurukul Last reviewed 23 October 2026 10 minute read

A clinic's front desk is one of the busiest places I know. The phone rings while a patient is paying. Someone wants to know if the doctor is in on Sunday. Three people need a reminder for tomorrow. A follow-up visit from last month was never booked. None of it is medicine. All of it takes hours.

My agency, PhantasyLabs, makes videos for doctors in India, so I spend a lot of time around clinics. I also run my own company with AI Employees every day, each one checked by a person, and I do not write code. This article puts those two things together: which clinic jobs AI can do safely, and where the line is.

Please read this as a starting point, not as legal or medical advice. I am not a lawyer or a doctor. The rules below are what I found in public sources on 23 October 2026, and they are still changing. Before you set anything up, speak to your own lawyer and check with your medical council.

If the idea of an AI Employee is new to you, read what an AI Employee is first. In short: an AI set up for one job, with clear written instructions, and a person checking its work.

What can a clinic safely hand over to AI?

The safe jobs have three things in common. They are about running the clinic, not treating the patient. They repeat every day. And a staff member can check the work before it reaches anyone.

1. Appointment reminders

An AI Employee can read tomorrow's appointment list and draft a reminder for each patient: the date, the time, the doctor's name, the address, and what to bring. Your front desk checks the list and sends them. This alone cuts down missed appointments and phone calls asking "what time was it again?"

Keep the message to the appointment only. No diagnosis, no test results, and nothing about the condition. A reminder that says "your diabetes review" can be read by anyone who picks up the phone.

2. Follow-up reminders

Many patients are told to come back in four weeks and simply forget. An AI Employee can look at the list of patients due for a follow-up, and draft a gentle reminder to book. The doctor decides who needs a follow-up and when. The AI only helps make sure the reminder actually goes out.

3. Answers to common questions

Timings, fees, parking, which days a specialist visits, how to book, what documents to bring, whether you accept a certain insurance. Write down your real answers once. An AI Employee can then draft replies from that document only.

The rule that keeps this safe: if the question is about symptoms, medicine or treatment, the AI does not answer it. It replies that a member of staff will call back, and passes it to a person. If someone describes an emergency, the reply should tell them to call emergency services or go to the nearest emergency department straight away, and alert your staff.

4. First drafts of health education content

An AI Employee can draft a post or a video script explaining a common condition in plain words. The doctor then reads every line, corrects it, and approves it. The doctor's name goes on it, so the doctor's eyes go over it.

Be careful here, because the rules on what a doctor may publish are strict. The NMC issued guidelines on ethical advertising and public communication on 6 October 2026. As reported by the press and in the notice itself, they bar AI generated promotional campaigns for commercial interest, require permitted AI generated promotional content to carry a mark saying it was made by AI, and forbid using AI to fake a patient, a result, a testimonial or an endorsement. They also ban patient testimonials, before and after pictures, and claims like "guaranteed cure" or "best doctor". Health education is still allowed. For the full picture, read can a doctor advertise in India.

What must a clinic never hand over to AI?

• Diagnosis. AI must not tell a patient what they have.
• Medical advice. No "take this", "stop that", "it is probably nothing", or "you should see a specialist". Those words carry weight, and they must come from a doctor.
• Prescriptions. AI must not prescribe or change medicine.
• Test results. Results and what they mean go from the doctor to the patient, not from a bot.
• Emergencies. A bot must never be the thing standing between a patient in danger and help.
• Anything sent without a person checking. Especially in the first months.

There is a clear rule behind the first three. India's Telemedicine Practice Guidelines, issued on 25 March 2020 and made part of the 2002 Code of Medical Ethics, say that AI and machine learning may not counsel patients or prescribe medicine. AI may help a doctor think, but the doctor gives the advice and signs the prescription. I could not find any change to that position as at 23 October 2026.

What does the DPDP Act mean for a clinic using AI?

The Digital Personal Data Protection Act, 2023 is India's main law on personal data. Patient names, phone numbers, appointment details and health information are all personal data under it.

Here is where it stands, as I found it on 23 October 2026:

• The rules were notified in November 2025. The Digital Personal Data Protection Rules, 2025 bring the Act into force in stages over about 18 months.
• Some parts are already live, mainly those setting up the Data Protection Board.
• Consent managers come in on 13 November 2026.
• The main duties apply from 13 May 2027, as reported by firms including Grant Thornton and Hogan Lovells. These include clear notice and consent, keeping data safe, reporting a breach, and deleting data you no longer need. Penalties can reach up to 250 crore rupees.

Do not wait for May 2027. Patients already expect their information to be kept private, and doctors already owe them confidentiality under the ethics code. The safe habits are simple:

• Ask first. Get the patient's clear consent before sending reminders on WhatsApp or SMS, and tell them what you will use their number for.
• Share the least. The AI only sees what the job needs. A reminder needs a name, a time and a phone number. It does not need the patient's history.
• Use tools you can trust with data. Know where the data goes, who can see it, and whether it is used to train the tool. Prefer business plans with clear data terms.
• Never paste patient records into a free personal AI account.
• Keep a record of what data the AI touches and why, so you can answer a patient who asks.

How should a clinic start?

What are the honest limits?

• AI can be wrong and sound sure. In a clinic, that is the main danger. This is why it answers only from your written document and never on medical matters.
• Patients may not know they are talking to AI. Tell them. A simple line such as "this is an automated reminder from the clinic" is honest and avoids confusion.
• The law is still settling. The DPDP duties start in May 2027, the NMC guidelines are weeks old, and how both apply to AI tools in practice is not fully clear. Expect changes and check again.
• Not every patient uses WhatsApp. Keep a phone line open to a real person.
• It saves time, not judgement. AI can cut the front desk load. It does not make a single medical decision safer, and it should not try.

For more on what AI Employees cost compared with extra staff, see what an AI Employee costs compared with hiring someone in India. For the difference between a simple chatbot and an agent that follows your rules, read AI agent or chatbot: what is the difference.

Akash Gupta, Founder, AI Gurukul

Akash studied at IIM Ahmedabad and built Peepoye, a Hindi YouTube channel. He does not write code. He runs his own company, Peepoye Network, with AI Employees he built himself with Claude, and teaches the same method in AI Gurukul. His story.

28.2 croreviews on his own channel

12.8 lakhsubscribers

1.4 crorehours watched