Platform Overview

How OnFinance AI reads circulars and builds evidence packs

From a new circular appearing on the RBI website to a structured, inspection-ready evidence pack, automated end to end.

RBI SEBI IRDAI PFRDA
Step by step

The processing pipeline

Each circular goes through four discrete stages before the evidence pack is available to your compliance team.

STEP 01

Ingestion

We monitor the publication feeds of RBI, SEBI, IRDAI, and PFRDA. Every new circular or amendment notification is downloaded and parsed the day it is issued, with source URL and publication timestamp preserved.

STEP 02

Structuring

The circular is parsed into its component structure: preamble, numbered paragraphs, schedules, and annexures. Legal cross-references (amended circular codes, superceded directions) are resolved.

STEP 03

Change Detection

Each provision is compared against the prior version of the same circular area. The system identifies whether a provision changed, was clarified, was removed, or is a restatement of existing rules.

STEP 04

Evidence Assembly

The evidence pack is assembled with source citations, change descriptions, prior-version references, and policy register linkages. It is exportable as structured PDF or JSON.

Platform concept

Designed around how Indian regulators actually issue circulars

RBI Master Directions are amended by subsequent circulars that do not always make the change relationship explicit. SEBI circulars reference prior circular codes without reprinting the base text. OnFinance AI resolves these relationships automatically.

  • Paragraph-level attribution preserved across the document history
  • Cross-circular amendment chains resolved automatically
  • Numeric limits, frequencies, and penalty structures extracted as structured fields
  • Human review workflow for low-confidence extractions
Platform concept diagram showing circular ingestion and evidence assembly
Technical capabilities

What the engine is built to do

The document intelligence layer is designed specifically for the structural patterns of Indian financial regulation, not general-purpose document parsing.

Cross-Document Reasoning

Amendments that reference prior circulars by code are resolved by tracing the citation chain. The engine understands that "in partial modification of circular X" means the prior provision is partially replaced, not wholly superceded.

Provision-Level Extraction

Numeric limits, reporting frequencies, entity types, and penalty structures are extracted as typed fields, not just narrative summaries. A changed limit is flagged with its old and new value explicitly.

Point-in-Time Snapshots

The platform maintains a versioned history for every circular area. You can see what the requirement was on any given date, not just the current state, which is essential for retrospective audit documentation.

Policy Register Linkage

Provisions are mapped to the sections of your internal policy register they govern. When a provision changes, the platform identifies which policies need review, based on the specific paragraph that changed.

Integrations

Connects to your existing compliance workflow

Evidence packs can be exported into the systems your team already uses. No rip-and-replace required.

PDF Export

Structured evidence packs in PDF format, formatted for RBI inspection team review and internal audit committee presentations.

JSON API

Structured JSON output for teams that want to ingest evidence packs into their own compliance management systems via REST API.

Spreadsheet Export

Excel-compatible export of change logs, mapping tables, and policy register update checklists for teams that manage compliance in spreadsheets.

Early access

See it on a circular your team actually handles

We are running early access with compliance teams at Indian banks and NBFCs. If you handle RBI, SEBI, IRDAI, or PFRDA circulars, we can show you a live run on a recent circular relevant to your institution type.