Software with a spine.
Most Indian hospitals run on paper, patchwork software, and memory. DivyaJivan is one system with opinions: one record, one ledger, one login — and rules the software enforces so people don’t have to.
The record belongs to the patient.
Most hospital software owns the patient. The file opens at your front desk, it lives inside your building, and when that person walks into a different hospital it starts again from a blank page — the same allergies re-asked, the same scan re-taken, three years of history reconstructed from memory in a corridor.
This is built the other way round. A person has one lifelong record, and a hospital reaches it because it is treating them: an appointment, a check-in, a referral, an admission. That relationship is what the server checks on every read, and when there is no relationship there is no access — not a warning, not an audit finding afterwards, no access.
For the hospital, the same system runs the whole day it is bought for: reception to pharmacy, lab to ledger, consulting room to the patient’s phone. Thirty modules switched on as you grow, five editions from one room to three hundred beds, one login per person however many jobs they hold. Your own business data — your tariffs, your collections, your staff — is yours, and is never shared with another hospital.
Four things that do not get traded away.
One record.
One person, one lifelong record, however many hospitals it passes through. Everything else in the product is downstream of that decision — the care-relationship check, the audit trail, the way a lookup across hospitals is masked and counted. A record copied into every building is not one record. It is several, and at least one of them is out of date.
Enforced, not remembered.
A safety rule that lives in a circular pinned to a noticeboard is obeyed on a quiet Tuesday and forgotten at two in the morning on a Saturday. The rules this product cares about sit in the way instead: the server refuses, and it refuses identically at two in the morning. That is less convenient than a reminder. It is meant to be.
The roadmap is public.
The front page of this site carries a list of things the product does not do yet — online payments, OTP sign-up, push notifications, video consultations, drug-interaction checking, ABDM. It is published rather than buried, because a hospital finds out either way, and finding out during the demo costs you the hospital. Nothing unbuilt is written here as though it were finished.
Built for India first.
Prescriptions that print in Gujarati as well as English. A ledger in rupees, held as whole paise from the counter to the day book. Family accounts, because a household arrives as a household. Registration that works for a walk-in with a phone number and nothing else. None of this is localisation bolted on at the end; it is the shape the software was drawn around.
Rules it holds, and why it holds them.
Every one of these makes somebody’s day slightly slower. Each is here because the faster version fails in a way that is expensive, unprovable, or dangerous — and because a rule the software keeps is the only kind that is still kept on a bad day.
A signed note locks. Corrections are dated additions.
A record that can be quietly rewritten is not a record. The original text stays, and so do the signing date, the correction date and how many corrections there have been — so a chart can still be explained a year later by the person who wrote it.
A refund can never exceed what was collected.
The alternative is a hole in the ledger that nobody notices until the month closes. The cap is arithmetic in the one place money moves, not a rule the counter staff are asked to keep in their heads on a busy afternoon.
A concession carries a reason and a named approver.
A discount with nobody’s name on it is not a discount. Recorded against the exact charge it reduced, so the gap between what the tariff says and what the patient paid is always answerable.
A part-dispense bills only what crossed the counter.
Two of the five strips on a prescription are in stock, so two are what the patient is charged for. The alternative — billing the prescription and settling up later — is how a pharmacy ends the month owing money it cannot account for.
A lookup into another hospital is masked, logged and capped.
A patient can be found by mobile, card or membership number from any hospital on the platform — that is the whole point of one record. So the phone number comes back masked, the search is written to the audit trail, and each account gets forty an hour. A search box over everybody is a search box over everybody.
Consent is three decisions, not one checkbox.
Terms, privacy and health-data consent are agreed separately and stored separately, each against the version of the document that was on the screen. Marketing is optional and never pre-ticked. Republish a document and everyone is asked again.
The server decides what you can see, on every request.
Permissions are read from the person’s own staff record each time, never from anything the browser sends — down to which of the thirteen tabs of a doctor’s screen an assistant is allowed to fill in.
Money is whole paise, end to end.
No floating-point arithmetic anywhere near a bill. The figure at the counter, the figure on the receipt and the figure in the day book agree digit for digit, because they are the same integer.
The things we said no to.
A feature list is easy to grow. These are four places where the product stops on purpose, and the reason in each case.
No no-code schema builder.
An administrator can add extra fields in the few places a hospital genuinely differs. They cannot redesign the clinical record. A field one hospital invents on a Thursday is a field no other hospital’s software knows how to read — and identity and clinical data are precisely where that ends badly. The core stays strict and validated; the flexibility is scoped to where it is safe to be flexible.
No clinical decision support.
There is no drug-interaction checking, no allergy cross-check and no dose-range warning, and no page on this site will imply one. A hospital can put its own caution labels on its own item master; they are notes, they never block anything, and no physician has signed them off. Software that looks like it is watching for a dangerous combination, and is not, is more dangerous than software that plainly is not.
No compliance badge.
Nothing here claims HIPAA, GDPR, DPDP, NABH or ISO. Those are certifications, and none is held. What is true is narrower and checkable: access is decided by the server, cross-hospital lookups are masked and rate-limited, and record access is written to an audit trail. Where the law is genuinely unsettled — data residency most of all — it stays an open question in our own documents rather than becoming a claim on our website.
No proof we have not earned.
No customer counts, no logos, no testimonials, no uptime figure, no store badge, no award. This is early software. The live counter on the front page stays dark until there is something true to put in it, and the roadmap section says out loud what is not finished. You would find all of this out anyway; better here than in the demo.
See it running on your OPD's busiest morning.
A thirty-minute walkthrough on your own cases — a token queue, a part-dispense, a refund, a concession that needs an approver. Bring your hardest workflow.