Ayushman Bharat Pradhan Mantri Jan Arogya Yojana pays an empanelled hospital nothing at the counter and everything on the claim. The patient is treated cashless, the hospital raises a pre-authorisation and later a claim, and a State Health Agency and its auditors decide, from the record you submitted, what gets paid and what gets deducted. That makes PM-JAY unusually document-driven for the hospital: the clinical work earns the payment, but the written record is what proves it. This piece is about what that record has to carry, where money leaks out of it, and the narrow place where an AI scribe helps.
Key takeaways
- A PM-JAY payment is assembled from the clinical record: diagnosis, package indication, treatment, and discharge summary must be present and consistent.
- Most avoidable deductions at medical audit are documentation gaps, not clinical ones.
- An AI scribe writes the note; it does not select packages, file pre-auths, or connect to the scheme’s systems.
From end of consult to drafted note
Note always returns in clean clinical English
Pre-auths or claims the scribe files
Source: Patient Square product specifications, 2026.
How PM-JAY actually pays a hospital
The scheme covers eligible families for up to ₹5 lakh per family per year for secondary and tertiary hospitalisation, delivered cashless at empanelled public and private hospitals (National Health Authority, PM-JAY portal). Treatment is billed against a defined package rather than as loose line items. The Health Benefit Package master, HBP 2022, lists the procedures, their specialties, and their rates, and the hospital selects the package that matches what the patient needs (National Health Authority, HBP 2022).
The flow that follows is where documentation starts to matter. For most packages the hospital raises a pre-authorisation before or shortly after admission, stating the diagnosis and the clinical indication for the package it wants approved. Treatment proceeds, and at discharge the hospital files a claim with the supporting record. A State Health Agency processes and audits that claim. At each of those steps someone who was not in the room reads your record and decides whether it supports what you billed. The clinical decision was yours; the payment turns on whether the paper behind it holds up.
Where the money actually leaks
Ask any PM-JAY billing desk where claims get reduced and the answer is rarely “the treatment was wrong.” It is almost always the record. A pre-authorisation that names a package but does not spell out the clinical indication invites a query. A claim whose diagnosis does not line up with the procedure billed gets flagged. A discharge summary that omits a procedure that was actually done, or that contradicts the pre-auth, gives an auditor a clean reason to deduct. These are documentation failures wearing the costume of clinical ones.
The reason this happens is structural, not careless. A busy IPD generates the clinical facts in real time and the paperwork after the fact, often reconstructed from memory at discharge under time pressure. The gap between what the team did and what the record says is where deductions live. Closing that gap does not require doing anything different clinically. It requires the note to carry, at the moment of care, the same detail the auditor will later look for: the diagnosis, the indication for the package, what was actually done, and a discharge summary that agrees with all of it. Our reads on how notes affect cashless approvals and whether better notes cut claim denials draw the same line for private cashless, and it holds just as firmly for a government scheme.
PM-JAY is not the private cashless you already run
It is worth separating PM-JAY from the private insurance and TPA cashless most hospitals also handle, because the documentation discipline overlaps but the machinery does not. Private cashless runs through insurers and third-party administrators, increasingly over the National Health Claims Exchange, against IRDAI timelines. PM-JAY runs on its own rail: State Health Agencies, the scheme’s own transaction systems, its own package master, and its own medical-audit and de-empanelment consequences for records that do not hold up. If you want the private-side picture, our plain-English TPA and cashless glossary and the NHCX explainer cover it. The common thread across both worlds is the one this post keeps returning to: the claim is only as strong as the note it is built from.
What a claim-ready PM-JAY note has to contain
The specifics vary by package and specialty, but the spine is consistent. A record that survives audit tends to carry all of the following, stated plainly and consistently across the pre-auth and the discharge summary.
| Record element | Why the auditor looks for it |
|---|---|
| Working and final diagnosis | Must support the package billed; a mismatch is a standard deduction |
| Clinical indication for the package | The reason this procedure, for this patient, now |
| Procedure and treatment actually done | Must match what was pre-authorised and claimed |
| Investigations and findings that justify it | The evidence trail behind the indication |
| Discharge summary consistent with the above | The single document auditors read most closely |
None of this is new to a clinician. What is new is how expensive an omission has become under a scheme that pays entirely on the strength of the written record. A finding you noted mentally but never wrote, or a summary assembled at speed that drops a detail, is not a clinical lapse. It is a payment risk.
Where an AI scribe fits, and where it stops
At the note, and nowhere near the scheme. We will be exact about the boundary, because a government-scheme claim is precisely the kind of thing a health vendor is tempted to overstate.
The AI Medical Scribe by Patient Square, one module of Practice Copilot, captures the visit as it happens and hands back a structured SOAP note, ICD-10 suggestions, and a prescription draft, ready to review and sign about two minutes after the visit. Read what is absent from that sentence for PM-JAY. It does not select an HBP package. It does not raise a pre-authorisation. It does not connect to the Transaction Management System, to a State Health Agency, or to any TPA. The ICD-10 outputs are suggestions for your coder or billing team to confirm, not a coding engine and not a claim submission. Package selection, pre-auth, and claim filing are your billing team’s work on the scheme’s own portal, and they stay there.
The benefit is upstream and deliberately modest. A complete, consistent note is better raw material for whoever prepares the pre-auth and the claim, so the diagnosis, the indication, and the discharge summary that reach the auditor are already coherent. For an Indian ward that carries a practical detail: the scribe captures the code-mixed Hindi and English of a real ward round and returns the note in clean clinical English, which is the language the record has to be in. Visit audio is processed in memory and discarded once the note drafts, so no recording sits on a server, which is the cleaner position under the DPDP Act 2023. Data is encrypted in transit and at rest, notes belong to your hospital to export or delete, and a SOC 2 Type II audit is underway. The full posture is on our security page.
The short version
PM-JAY pays on the record, not at the counter, so a hospital’s exposure under the scheme is largely a documentation exposure. The claim is built from the diagnosis, the package indication, the treatment done, and a discharge summary that agrees with all three, and the deductions that hurt most are the ones an auditor can trace to a gap in that chain rather than to the medicine. Nothing about that requires new clinical work; it requires the note to carry, at the point of care, what the audit will later ask for.
If your PM-JAY deductions keep clustering on missing indications and summaries that do not line up, that is a note-quality problem you can watch fixed on a real case. Book a short demo and bring a recently queried scheme claim, then look at what a complete structured note for that admission would have carried. Or run the 7-day trial across a ward day and check whether the notes hold the indication and findings your claims desk keeps chasing. How the data is handled is on the security page.