Most clinical AI is built for hospitals, because that is where the money is. A hospital charges in lakhs and can carry the cost of software built for institutions. A small standalone clinic charges a few hundred rupees a visit, runs a couple of hours in the morning and a couple in the evening, and cannot.
That gap is the whole reason Chikitsa exists.
What a small clinic actually needs
A founding clinician we work with described the person a small clinic really wants: a multi-purpose health worker. Someone who can take notes, book an appointment, help with a procedure, keep the records, and handle the money at the desk. Not a specialist in any one of those. A generalist who can do all of them well enough.
The problem is not finding hands. It is that the intellectual part of that role, the clinically-correct note, the honest account, the tidy record, is hard to hire for at a price a small clinic can bear for a few hours a day. That is the unmet need. Not a replacement for staff. An add-on to them.
What we are building, and what we are still arguing about
We are not trying to do everything at once. The clinician gave us a priority order, and we are following it.
First, the clinical assistant: ambient notes and clinically-correct reporting from the consultation. This has to be right before anything else earns trust. It is the part nobody else is building for a clinic this size, in the languages actually spoken in the room, and it is where our engineering belongs.
After that it gets more interesting, and we will be honest that it is unresolved. Doctors who have seen it ask next for the accounting, the stock, the appointment book. Those are real needs. They are also a different product, one that a dozen companies already sell, and a small team that tries to build both usually ends up with two mediocre things.
So the question we are actually sitting with is not what could we add. It is what should we build ourselves, and what should we connect to something that already works. Our instinct is to build the part that is genuinely ours, the ambient clinical documentation, and to integrate the rest rather than reinvent it. We have prototyped enough of the surrounding pieces to know what they cost, including reading a supplier’s receipt to keep stock updated, which we are still trying rather than claiming.
We would rather publish that open question than a roadmap we might not keep. If you run a clinic and have a strong view on it, we would genuinely like to hear it.
The economics, said plainly
The question a doctor asks is simple: is this worth paying for every month? The honest answer depends entirely on how genuinely useful it is on an everyday basis, and that is exactly what the pilot is measuring. We would rather find the loopholes now, with one demanding clinician, than sell a promise.
Chikitsa runs on our own hardware in India. The clinician approves every word that enters a record. It is a private pilot today, not a public product, and it will stay that way until it has earned the right to be otherwise.