Built in Bengaluru, 2023, to stop the same report being rebuilt by hand
OnFinance AI was founded by people who had worked inside Indian banking compliance and watched the same circular impact analysis get rebuilt from zero, every time a new RBI or SEBI notification arrived.
The problem we are fixing
Indian financial regulation is dense, frequently amended, and spread across four major regulators. RBI alone issues hundreds of circulars a year, many of which partially amend prior ones without reprinting the base text. For a compliance officer, understanding what actually changed requires reading the new circular, finding the prior circular it references, and comparing the specific provisions side by side.
This work is done manually at most institutions. The same analysis is rebuilt from scratch each time a circular drops. The institutional knowledge of what-changed-and-why lives in one person, and when they leave, the institution loses that context.
We built OnFinance AI to make that analysis automatic: detect the change, attribute it to the specific provision, link it to the prior version, and assemble the evidence pack that an inspection team would expect to see. The compliance officer's judgment still matters for deciding what to do about the change. We automate the work that precedes that judgment.
The people building this
A small team based in Bengaluru, with backgrounds in banking regulation, financial technology, and machine learning applied to structured documents.
Anuj Srivastava
Anuj founded OnFinance AI in Bengaluru in 2023 after years building compliance and risk systems for Indian banking institutions. He saw compliance teams reconstruct the same circular impact analysis from scratch, each time a new RBI notification arrived, with no institutional memory carrying over. That repeated manual rebuild is the problem OnFinance AI is designed to eliminate.
Priya Nair
Priya leads the platform engineering: the circular ingestion pipeline, NLP-based change detection, and the structured output layer that produces evidence packs. Her background spans distributed document-processing systems and NLP applied to structured regulatory text, with prior work building data extraction infrastructure for financial services use cases.
Siddharth Rao
Siddharth spent the better part of his career interpreting RBI and SEBI circulars from inside the compliance departments of regional and private-sector banks. He understands how inspection teams read evidence documentation, what they look for first, and where manually assembled packs typically fall short.
What we care about
A few principles that shape how we build the product and engage with the compliance teams we work with.
Auditability above summaries
A summary of what changed is useful. An evidence pack with source citations and prior-version references is what a compliance team actually needs when an inspection team arrives. We build for the latter.
Honest about what AI cannot do
Language models are good at identifying structural changes in regulatory text. They are less reliable when provisions reference legal-defined terms with narrow meanings, or when two circulars contradict each other. We flag low-confidence extractions for human review rather than presenting them as certain.
Built with the people who use it
We are in early access with compliance teams at Indian banks and NBFCs. The specific circulars we prioritise, the evidence pack structure, and the policy register mapping approach are shaped by the people who handle this work every day, not by our assumptions about what they need.
Join our early-access program
We are actively seeking compliance officers and regulatory affairs teams at Indian banks, NBFCs, insurance companies, and mutual funds who want to shape what this platform becomes. If you handle RBI, SEBI, IRDAI, or PFRDA circulars, we want to hear about your workflow.