Status and roadmap
Every feature carries one of four statuses. This page lists what is proposed and, more importantly, what is not built. It is the section that actually recruits.
Works in the product
Engine built and tested, screen not wired
Designed or schema only
Proposed, not started
Global caveat
Nothing persists to a database yet. The whole product runs on in-memory state. Persistence is the current milestone.
Proposed · 1 of 3
Highest conviction — a date forces the purchase.
Consent-manager integration, one-year-plus audit logging, role-based access and operationalised patient rights — access, correction, withdrawal, grievance. Obligations start November 2026, full compliance May 2027, no small-entity exemption, and a hospital cannot transfer liability to its software vendor. We already hold the primitives: consent records, record-access log, e-signatures, retention batches.
IRDAI mandates cashless claims be processed within three hours of discharge authorisation. Our seven-gate discharge clearance and NHCX-shaped insurance stages are already the right shape — add a live countdown per case and a “what is blocking the clock” view.
Entry Level covers five chapters and certified hospitals earn 10% extra on PMJAY claims. Ship the objective-element checklist, bilingual patient-rights display, standardised consent forms and the three-month evidence trail on top of the QMS loop we already have. That is a payback calculation, not a feature.
The ABHA creation flow is small work; the real value is that FHIR R4 conversion, sandbox certification and consent management sit with the vendor so the hospital never sees them. Sequenced after persistence — NHA milestone certification is calendar time, not engineering time.
Proposed · 2 of 3
Why deals in this segment actually die.
Poor connectivity is the most-cited cause of HMIS failure in non-metro India. Local-first with sync is hard, and therefore defensible. Pulling the network cable mid-demo and continuing to admit a patient wins the room.
Doctors reject software that needs too many clicks. Commit publicly: admit a patient in three clicks or fewer, record a dose in two, and print the counts on the features page.
“Most vendors provide training for one week and then vanish.” Extend the getting-started checklist into guided setup plus Excel and legacy import, so a hospital can run without us.
Small hospitals have no dedicated data-entry operators; the receptionist is also billing and also pharmacy. One combined shift-worker view instead of three role-shaped silos.
A standards-based five-level engine — auditable, not a model. Absent on both sides of the competitive comparison, and the first question any clinical buyer asks.
Proposed · 3 of 3
Only the AI our data moat lets us defend.
The one AI feature no competitor can copy, because none collects the data. “Which patients will still be boarding in four hours, and which bed frees first” trains on boarding duration, escalation and suitability data that exists nowhere else.
Highest value, lowest clinical risk — the structured inputs already exist in admission, diagnosis, MAR, orders and results. Draft it, clinician edits and signs, and it feeds the three-hour cashless clock.
For India this means TPA-specific document completeness and pre-auth gap detection before submission. Denial-prevention programmes cut denial rates 30–40%.
Suggesting the correct PMJAY package is a direct revenue lever, and it compounds with the NABH 10% uplift.
The constraint engines already exist — rest hours, consecutive nights, credential validity, exact cost. “Suggest the cheapest compliant roster” is an optimisation problem, not a research problem.
Word error rates run 15–25% on Indian accents and code-switching, always-on cloud fails on hospital connectivity, and patient audio is DPDP-scoped. Build it on-premise or in-country, ship it as an assistive draft requiring sign-off, and publish the 15–25% figure ourselves.
Explicitly not building
Anything diagnostic, or triage acuity assigned by a model. Regulatory exposure — and it repels precisely the clinically literate partner we are recruiting.
What isn’t built
Four empty boxes. For a partner with hospital relationships, the last one is the most valuable thing on this page.
The entire hospital domain runs in memory; the first Emergency migration exists and nothing consumes it. The DPDP position is the one deadline we can uniquely serve, roughly fourteen months out. Persistence is the only thing standing between us and it.
No practising clinician has reviewed the triage flows, MLC handling or drug rules. The rule packs carry an explicit unverified_demo flag and read “verify before use”, deliberately, so nothing poses as a legal citation it hasn’t earned.
NABH mapping, Drugs & Cosmetics Rules schedule assignments and retention rules are modelled but unverified by anyone qualified to sign them off.
No pilot site, no paying customer, no go-to-market. For a partner with hospital relationships, this is the most valuable empty box on the page.
We are looking for
A clinical partner who has run a small hospital, and a distribution partner who knows fifty of them.
If you have argued with a management report about boarding patients, or watched a vendor vanish after training week, we would like to talk.
Start the conversationFounding hospitals shape the order after that. Tell us which deadline matters most to you — DPDP, the three-hour clock or NABH Entry Level.
