Why we built hospital software with no AI touching your data

Most hospital software sold in India today has an AI feature in it somewhere. A summariser for case notes. A chatbot for the front desk. Something that reads what you typed and suggests what to do next.

We built Amma Health without any of it, on purpose. This is the reasoning, written down, because "we don't do that" is easy to say and worth very little unless you explain what you turned down.

What "no AI" actually means here

It is a narrow, checkable claim, so it is worth being precise about it.

Nothing your hospital enters into Amma Health is sent to an AI system. There is no chat feature reading your records. Nothing summarises a patient's history. Nothing analyses the contents of a document you upload. And none of it is used to train a model — not ours, not anyone else's.

That covers everything you put in: patient details, appointments, notes, prescriptions, bills, and the files you attach to a record.

Why we decided against it

Patient information should not become someone else's training data. Once text leaves your system and reaches a model, you cannot get it back and you cannot audit what happened to it. The promise a vendor makes about that is a policy, and policies change with funding rounds. The only version of this promise that survives a change of ownership is the one where the data never went anywhere.

An AI feature would change what this software is. Amma Health is hospital management software: it handles the administration of running a hospital — the front desk, appointments, billing, staff, branches. It is not an EMR, not a clinical system, and it does not offer medical advice or decision support. The moment software starts interpreting clinical text and suggesting conclusions, that line moves, and it moves into territory with a great deal more regulation attached. We would rather do the administrative job properly.

Most of it does not survive contact with a real front desk. The work at a reception counter is not waiting on a cleverer suggestion. It is registering a patient without retyping their details, finding a booking while someone waits, and printing a bill on your letterhead. Those are ordinary problems, and they are solved by software that is fast and does not need to be argued with.

What you get instead

The things we did build in place of it are unglamorous and they are the reason people keep using the software:

• Register a patient once, and their details carry to every desk that needs them
• Book, move and cancel appointments without phoning another department to check
• Raise an itemised bill on your own letterhead without retyping anything
• Run several branches from one login, with each branch's data kept separate
• Give each role only the screens their job needs

The commitments this sits inside

No AI is one of four promises, and the other three are what make it mean something:

You own your data. It is yours, not ours. We make no claim of ownership over anything you put in. We do not sell it, rent it or trade it. We run no advertising and no tracking.

You can export it whenever you want. Ask for a full export at any time and you get it. Not as a retention tactic, not after a phone call with someone whose job is to keep you.

There is a real exit plan. If we ever shut down: sixty days' notice, payments stopped that day, and all your data handed over before we close. That is written into our terms rather than left to goodwill.

And with document storage, files sit in cloud storage you own and pay for directly. We keep no copy — not even in our backups — which means there is nothing on our side to hand over, analyse, or lose.

If this matters to you

It does not matter to every hospital, and that is a reasonable position to hold. Plenty of administrators would take a good summariser over a promise.

But if you have ever wondered where a patient's information actually goes after you type it in, this is the answer for our software: it stays in your hospital's records, it is read by the people you gave access to, and nothing else looks at it.

If you would like to see how the rest of it works, we will walk through the parts your hospital would actually use, on your own workflow rather than a canned demo.