Follow the money through every ledger the state already holds.

Ledger ingestion

Upload one or many departmental exports in .xlsx, .xls, .csv or .json. Payment registers, bill books, sanction orders, beneficiary lists and vendor masters are merged into a single dataset before analysis. Departments most often send the exports they already produce for their own record keeping - expenditure and release statements, procurement and award registers, subsidy and beneficiary rolls, work orders and utilisation certificates. None of it requires NIRNAE to be given access to the system that produced it.

Nine-agent detection pass

Every dataset is examined by all nine agents in a single run. Threshold splitting, ghost billing, year-end dumping, jurisdiction mismatch, infant vendors, head-of-account misuse and pass-through chains are checked together, not one at a time.

Entity resolution

Shared addresses, phone numbers, tax-ID prefixes, bank accounts and directors are collapsed into clusters, so a ring of nominally separate firms surfaces as one beneficiary. This runs entirely on the vendor and company particulars already present in what a department exports. No external registry is queried and nothing beyond the submitted dataset is consulted to form a cluster.

Scored, defensible findings

Each finding carries a severity score, the transactions it rests on and the arithmetic used to reach it — written so it survives review by a superior officer or an audit committee.

Constituency development funds

Utilisation review of MPLADS and MLALADS style allocations — works sanctioned against the same estimate twice, implementing agencies retaining unspent balances, and recurring contractors across unrelated constituencies.

Centrally sponsored scheme leakage

Trace state-level disbursal of central funds down to the paying unit: muster rolls with impossible attendance, material bills without measurement books, and diversion between scheme heads. A release statement and the corresponding payment register are usually enough to begin.

Subsidy and DBT beneficiary integrity

Duplicate and near-duplicate beneficiaries across districts, one bank account receiving many identities, and payments continuing after eligibility lapsed.

Loan and priority-sector fraud screening

For public sector banks and refinance bodies: circular routing between borrower group entities, collateral pledged more than once, and disbursals against newly created firms with no operating history.

Defence and infrastructure procurement

Review of high-value contracting for single-vendor steering, repeat rate escalation, and subcontracting chains that pass most of the value to an unqualified party. Award registers and vendor masters exported from the department's own procurement records are the only input required.

Licence and allocation audits

Where a scarce public resource is allocated administratively — spectrum, mineral blocks, land parcels — reconstruct the sequence of applications, eligibility changes and downstream transfers of the awarded entity.

Grant utilisation for R&D and rural credit

For bodies such as DRDO and NABARD: verify that sanctioned grants reached the stated institution, that equipment purchases match the approved head, and that recipients are not recycling funds through associated vendors.

Post-event forensic reconstruction

When an irregularity is already suspected, reconstruct the full money trail across departments and years and produce the entity map and evidence pack an inquiry requires.

Every finding shows its arithmetic

A flag without a calculation is an accusation. Each NIRNAE finding lists the transactions, the totals and the comparison that produced it, so an officer can re-derive it by hand.

Written for the file, not the headline

Output is phrased in the language of audit observation and recommended verification — suitable for an inspection note, an audit para or a reference to a vigilance cell.

Your data stays your data

Datasets are processed for the analysis you request. Nothing is sold, published or used to train a public model, and on-premise deployment is available for restricted records.

Data handling

• NIRNAE holds no live access to treasury, PFMS, banking or departmental systems, and asks for no credentials or network route into them.
• All data originates with the department as a deliberate export. NIRNAE collects nothing on its own and connects to no external register without written instruction.
• Datasets are transmitted over encrypted channels and processed only for the analysis the department has requested.
• Access during an engagement is limited to the personnel named for it.

Confidentiality boundaries

• Departmental data is never sold, shared with another client, or published in any form.
• It is not used to train a public model, and findings from one engagement never inform another.
• Results are returned to the requesting authority only; NIRNAE does not disclose them onward.

Retention and disposal

• Datasets are retained for the engagement window agreed in writing, then deleted.
• Deletion can be requested at any point during or after the engagement and is confirmed back to the department.
• For restricted or classified records, on-premise deployment is available so data never leaves departmental infrastructure.

Audit-grade evidence standard

• Every finding states the transactions, the totals and the comparison behind it, so a reviewing officer can re-derive it by hand.
• Output is phrased as audit observation and recommended verification — never as an accusation or a legal conclusion.
• Findings are indicators for scrutiny. Verification, adjudication and any action remain entirely with the department.