You cannot fix an org chart with an API
On 30 April 2026, at the 10th National Summit in Chandigarh, the Health Ministry launched the Swasth Bharat Portal: multiple national health programmes behind one login, on an API-based federated architecture. It is the right thing to build and it is overdue. But India never had a digitisation problem. India digitised at enormous scale. What it had was an org-chart problem, and you cannot fix an org chart with an API.
- Swasth Bharat is an aggregator of multiple national health programmes, not a scheme of its own. It does not fund care or cover patients; it orchestrates programmes that do.
- Convergence is real at the interface. Underneath, each programme keeps its own budget, targets, reporting line and definition of a beneficiary.
- The user-mapping step is the tell: existing programme credentials are bridged into the new environment, not retired.
- There are five stages of convergence. Stages 1 and 2 are announced; 3 and 4 are execution-dependent; 5 is not claimed. The likely failure mode here is a stall at Stage 2, with everything still working.
- The efficiency projections, 20-30% on infrastructure and 20-40% on data entry and HR duplication, were issued within days of launch. They are design targets, not results.
My position: this is the correct diagnosis and the correct architecture. Every serious analysis of Indian public health data has asked for exactly this for a decade. My argument is not that it fails. It is that the part being celebrated is the easy part, and the numbers that would prove the hard part worked are not currently published.
02 — The object
What it is
Swasth Bharat is a horizontal layer, not a vertical programme. It draws identity from ABDM above and orchestrates existing programme systems below. Its users are ASHAs, ANMs, community health officers, medical officers and programme staff. It is a workplace for health staff.
It is not a health insurance or entitlement scheme of its own, not a replacement for ABDM, and not primarily citizen-facing. The most common error is treating it as an entitlement. It is infrastructure.
The programmes on the portal as of August 2026, with their full names:
That list will grow. The Ministry has said it plans to bring in more programme systems and the national registries over time, so the count today says little about where this ends up. A wider MoHFW ecosystem graphic circulating online shows a much larger set including AAM, PM-JAY, RBSK and e-RaktKosh; that is the ministry's programme landscape, not the current Swasth Bharat integration list. Worth separating the two, because conflating them overstates what has actually been integrated.
View the data behind this
| Programme | Domain | Counter |
|---|---|---|
| U-WIN | Universal Immunization Programme | 15.2 crore beneficiaries immunised |
| NCD portal | Hypertension and diabetes | 9.13 crore under treatment |
| Sickle cell | Sickle Cell Anaemia Elimination | 7.1 crore screened |
| Ni-kshay | National TB Elimination Programme | 1.75 crore cases notified |
| JANANI | Maternal and newborn care | 1.33 crore under maternal health management |
| PMNDP | PM National Dialysis Programme | 31.13 lakh availed dialysis |
Source: swasthbharat.mohfw.gov.in portal counters (official). These are counters, not unique people. One citizen can appear in several.
Two caveats. First, they are cumulative counters: one citizen can appear in four of those rows at once, which is precisely the double counting ABHA linkage exists to resolve. Second, the counters lag. Government reported over 7.24 crore sickle cell screenings as of 16 July 2026 while the portal displayed 7.1 crore. An aggregator whose own numbers refresh differently from its sources has not finished aggregating.
Owns no data. Funds no care. Its authority over the layers below is delegated, which is the open question of this edition.
View the data behind this
| Level | Layer | Components |
|---|---|---|
| 1 | Citizen identity | ABHA |
| 2 | Ecosystem registries | HFR, HPR |
| 3 | Health information interoperability | ABDM ecosystem, consent, exchange |
| 4 | Programme convergence | SWASTH BHARAT PORTAL |
| 5 | Programme systems | NCD, JANANI, U-WIN, Ni-kshay, PMNDP, sickle cell |
| 6 | Care delivery | ASHA, ANM, CHO, Ayushman Arogya Mandir, PHC, CHC, District Hospital |
Analysis: six-level stack synthesis by the author, built on ABDM and MoHFW documentation.
The problem it was built for is real and well documented: a separate portal per programme, multiple logins per worker, the same patient registered repeatedly, fragmented datasets, duplicated hosting, separate maintenance teams, and administration crowding out service delivery. The government's diagnosis is right: the problem was never digitisation, it was that every programme digitised separately.
03 — The hard part
An API can join the screens. It cannot join the budgets.
Each integrated programme has its own budget line, its own targets, its own reporting chain, its own review meetings and, in practice, its own definition of who counts as a beneficiary. A programme officer is appraised on their programme's numbers. None of that changes when a portal puts them behind one login.
The user-mapping step is the tell. A worker's existing programme credentials are bridged into the Swasth Bharat identity and kept alive. That is what federation means in practice, and it is the honest engineering choice. It also means the old systems, and the old accountabilities, keep running underneath.
The constraint nobody can engineer around: health is a state subject. State health departments face an integration question they did not choose. A federated API invites integration. It cannot require it. We already know from PM-JAY that household coverage runs from 21% in Bihar to 90% in Chhattisgarh, and that spread follows state execution, not income levels. Convergence will land the same way.
04 — The ladder
Five stages, and the one where things quietly stop
This maturity model is mine, not the government's. The Ministry does not publish a staged framework. I am using it to separate what has been announced from what depends on execution.
Convergence is not binary. It has stages, and each one is harder than the last.
View the data behind this
| Stage | What it means | Status |
|---|---|---|
| 1. Access convergence | One login, one interface across programmes | Announced |
| 2. Data convergence | Demographic and clinical detail entered once | Announced |
| 3. Patient convergence | One person linked across programmes via ABHA | Announced, execution-dependent |
| 4. Clinical convergence | Longitudinal disease and treatment view at point of care | Announced, execution-dependent |
| 5. Intelligence convergence | Risk stratification, predictive intervention, resource allocation | Future opportunity, not a live capability |
Analysis: five-stage maturity model by the author. Stages 1-4 map to announced capabilities; stage 5 is not claimed by government.
Stage 1 is a login. Stage 2 is entering data once. Stage 3 is knowing that the pregnant woman in JANANI and the hypertensive in the NCD portal are the same person. Stage 4 is a clinician seeing that at the point of care. Stage 5 is the system telling you who to visit next week.
The likely failure mode here is not collapse. It is a perfectly functioning portal that never gets past Stage 2, which is the most common outcome for aggregation layers everywhere. Nobody writes a press release about stalling. The platform keeps working, the logins keep counting, and the clinical convergence that justified it never arrives.
And notice where the difficulty changes character. Stages 1 and 2 are engineering. Stages 3 and 4 are execution-dependent, meaning they turn on ABHA seeding quality and state adoption, neither of which the platform controls. The portal can succeed completely at everything it can control and still stop at Stage 2.
05 — The chain
Seven links, and where the value leaks at each one
Illustrative model. Counters are structural (five programmes means five registrations), not published statistics.
Traced end to end, the chain runs from identity to outcome. Each link creates something real and each has a specific way of failing. The failures compound.
There is a pattern in those leaks. The early links fail on data quality; the late links fail on institutional will. Link 6 is where most health data platforms in every country come to rest: the analytics exist, the clinical workflow does not change, and the dashboard becomes the deliverable.
06 — Failure modes
Four ways this quietly fails
None of these is dramatic, which is why they are easy to miss. Each one produces a platform that still works, still reports, and quietly stops delivering the thing it was built for.
07 — The tension
The equity problem nobody is naming
Three of the programmes currently integrated sit squarely in equity territory. Sickle cell screening targets tribal districts. Dialysis support was built for households that cannot afford it. Maternal care is where India's outcome gaps are widest. On its programme mix, this is an equity platform.
And yet. Identity-anchored systems concentrate their benefits on people with clean identity records. ABHA seeding is likely to be weakest exactly where health need is highest: remote blocks, migrant households, women without independent documentation, tribal populations. The convergence dividend accrues first to the people already easiest to find.
What to demand: ABHA seeding rates disaggregated by district, tribal block and gender. If that number is never published, the equity question stays unanswerable. If nobody publishes it, nobody can check it.
08 — The number
About those efficiency projections
MoHFW published unusually specific expectations, which most launches do not. But they were issued within days of launch, which makes them design targets.
View the data behind this
| Area | Projected reduction | Status |
|---|---|---|
| Infrastructure load | 20 to 30% | Ex-ante projection, unaudited |
| Data entry effort | 20 to 40% | Ex-ante projection, unaudited |
| HR duplication | 20 to 40% | Ex-ante projection, unaudited |
| Decision-making speed | Stated as expected to increase | No numeric range given |
Source: MoHFW efficiency estimates, PIB release 2258283, 6 May 2026. Issued within days of launch.
The mechanisms are plausible and worth stating: aggregated hosting and compute, demographics entered once, consolidated development and maintenance teams, higher interoperability from the federated design. Each is a real saving. None has been measured.
Treat the bands as design targets to be audited, not results to be quoted. The Ministry could publish the methodology it will use to measure them before the first measurement exists. That is a governance decision and it costs nothing.
09 — So what
Ten questions, answerable within a year
These are open questions, not reported failures. The portal is months old. Each one becomes answerable inside twelve months.
ABDM built much of India's digital health foundation. Swasth Bharat is the layer that could finally make that foundation usable across national programmes. The architecture is right. The projections are unaudited. The gap between the two is a year of measurement nobody has committed to publishing.
India has repeatedly built excellent rails and then declined to publish whether anyone rode them. The portal converges multiple programmes on a screen. Whether it converges them in a budget, a review meeting and an ASHA's evening is a governance question, and governance questions are settled by what gets measured, not by what gets launched.
The question I would put to an oversight committee: if the only numbers ever published about this platform are logins, integrations and cumulative counters, how would anybody, inside or outside government, ever know whether it got past Stage 2?
10 — Sources
Sources and evidence status
What is officially stated, what is a government projection, and what is my analysis. Nothing above is presented as fact without one of these three tags.
Before citing externally: the portal counters move. Check them again at source. Everything here reflects the position as of August 2026.