The RBI requires NBFCs to submit a range of periodic returns covering asset quality, provisioning, and NPA movements. Each return has a prescribed format, a filing deadline, and a consequence for late or inaccurate submission. The Provisioning AI generates every IRACP-related regulatory filing automatically from the classification and provisioning data it maintains daily — no manual data extraction, no spreadsheet assembly, no last-minute data reconciliation before the deadline.
What regulatory filing requires — and where manual processes create risk
A regulatory filing that is filed late is a compliance failure. A regulatory filing that is filed on time but contains data inconsistent with the institution's own books is a worse failure — it creates a discrepancy between the regulatory submission and the financial statements that an RBI inspection will identify. The most common source of this discrepancy in manual filing processes is the time-gap between the data extraction, the spreadsheet preparation, and the final review — during which the underlying data may have changed (an account reclassified, a provision updated) and the filing data no longer matches the books.
The Provisioning AI eliminates this gap entirely. Every report it generates is sourced directly from the classification and provisioning database that it maintains in real time — the same database that feeds the financial statements. There is no extraction, no transfer, no opportunity for the data to diverge between source and submission.
The IRACP reporting outputs: what gets filed, when, and in what format
(Form A)
Asset quality position as at quarter-end — account counts, outstanding balances, and provisions by category
Reports the number of accounts and outstanding balance in each IRACP category (Standard, Sub-Standard, D1, D2, D3, Loss), the provision held against each category, and the movement from the prior quarter (downgrades, upgrades, write-offs, fresh NPAs). Generated from the classification database as at the quarter-end date; no manual data extraction required. XBRL taxonomy applied automatically for RBI portal submission.
Within 21 days
of quarter-end
Schedule
Reconciliation of NPA balance movements: opening, additions, upgrades, write-offs, and closing position
A reconciliation schedule that bridges the prior quarter's NPA balance to the current quarter's: opening NPA → fresh NPA additions → upgrades back to Standard → write-offs and settlements → closing NPA. Each movement category is sourced from the classification audit trail, ensuring the schedule ties precisely to the account-level data. Provided to the statutory auditor as a supporting schedule.
With financial
statements
Return
Total provisions held, provisions required, and coverage ratio — by IRACP category and product segment
Reports total provisions held against each NPA category versus total provisions required under IRACP norms, producing the provision coverage ratio for each category. Also computes the aggregate provision coverage ratio — a key metric in RBI supervision. Any under-provisioning (where actual provision is below required minimum) is flagged automatically, and the gap and remediation requirement are stated in the report.
Within 21 days
of quarter-end
Disclosure
Gross NPA ratio, Net NPA ratio, and Provision Coverage Ratio — required public disclosure for large NBFCs
Computes and prepares the three NPA ratios required for public disclosure: Gross NPA ratio (total NPA outstanding / total advances), Net NPA ratio (NPA outstanding net of provisions / total advances), and Provision Coverage Ratio (provisions held / gross NPA). Generated from the same source as the regulatory return, ensuring consistency between the public disclosure and the regulatory filing.
Simultaneous
with RBI filing
Report
Monthly asset quality update for the Board: NPA position, movements, provisioning adequacy, and management commentary
A structured monthly report for the Board that tracks NPA trends, identifies the highest-exposure NPA accounts with their individual provision positions, and flags any accounts that have crossed a classification threshold in the month. The management commentary section notes trends (e.g., "MSME segment NPA increased 40bps this month — primarily from 3 large accounts in the textile sector") sourced from the classification data. Presented at the Board's monthly risk review meeting.
Before Board
meeting
Confirmation
Pack
Complete NPA and provisioning workings for statutory audit — account-level detail with full audit trail
At each statutory audit, the Provisioning AI generates an account-level pack containing: every NPA account's classification, outstanding balance, security value, DPD count, provision required and held, and the full classification audit trail showing every movement. The auditor can verify the aggregate provision from the account-level workings without needing to sample and reconstruct — the entire population is documented and traceable.
(+ quarterly
interim)
The filing status dashboard: what the compliance team monitors
The discrepancy that the Provisioning AI eliminates
In a manual filing process, the sequence is: extract data from CBS → transfer to spreadsheet → apply IRACP calculations → reconcile to financial statements → review and correct discrepancies → submit. Each step in this sequence is an opportunity for the data to diverge from the source. An account reclassified between the extraction and the submission creates a discrepancy. A provision updated in the books after the spreadsheet is locked creates a discrepancy. A formula error in the IRACP calculation creates a discrepancy.
The Provisioning AI's filing process has no extraction step, no transfer step, and no spreadsheet. It reads directly from the classification database, applies the IRACP calculations using the same rules that it uses for the daily classification run, and writes the output directly to the filing format. The filing is not a representation of the books — it is the books, formatted for regulatory submission. The discrepancy between the two cannot exist because they are the same data.
The filing deadline is not the target — it is the backstop
An institution that submits its NPA return on the last day before the deadline has allowed a risk to accumulate: any data quality issue discovered on filing day cannot be corrected before submission. An institution that has its filing data-ready a week before the deadline has created a buffer for review, correction, and internal approval without any regulatory risk. The Provisioning AI generates every IRACP filing from the live classification database — not from a manually extracted snapshot taken the night before the deadline. The data is always current, the filing is always preparable in hours, and the institution's compliance team manages filings proactively rather than reactively. The deadline is a backstop they never approach.