[ Idea ]Found 13 Nov · 1 GitHub · 1 Devfolio · 1 DEV

Munim

Shared bank-statement parsing for private finance apps, run on-device.

Should you build this?Checking

This idea is still going through the six checks.

0 products · 11 sourcesPending

The problem

Indian bank statement PDFs come in dozens of layouts that change without warning, so every private finance app has to build and fix its own parser.

Privacy-minded users won't upload statements to cloud AI, and bank data APIs are closed to individuals and small apps. When a bank changes its PDF, each app breaks silently, and no app learns from the others' failures.

Seen in 1 GitHub · 1 Devfolio · 1 DEV

The idea

Munim is an open-source parser that reads statements on the user's device and shares layout fixes without sharing any user data.

1Drop in a password-protected statement PDF; it is read in memory.
2Each row is marked verified or mismatch by checking the running balance.
3Pro users get signed fixes within 72 hours of a detected change.

Price: Free open-source core covering the 5 biggest banks. Pro is ₹1,499/month (about $18) for 50+ banks and cards, signed rule-pack updates within 72 hours of a detected layout change, and a regression-test fixture feed. Team is ₹4,999/month for private custom-bank packs and priority fixes. A hackathon weekend pass is ₹499. Billing through Razorpay/UPI.

[ The six checks ]0 of 6 done

1Not built in alreadyNot checked

2An AI agent can't do itNot checked

3Nothing similar existsNot checked

4People pay for thisNot checked

5Strong scoresNot checked

6Real sourcesNot checked

1 · Not built in alreadyNot checked

2 · An AI agent can't do itNot checked

Who paysThe developer or small team shipping the finance app, paying from their own pocket or a small product budget. Not the end user.

Who uses itIndian indie developers, hackathon teams (especially Local LLM and fintech tracks) and open-source maintainers building private personal-finance, tax or net-worth tools on top of statement PDFs.

First 10 users1) Post a short X video of an HDFC or Kotak statement parsed in the browser with the DevTools Network tab showing zero requests, then link a waitlist page with prices. 2) DM the maintainers of the public Indian local-finance repos (VyayaMitra, LocalFinance, Finance Copilot, HumDono) and offer free Pro in return for sharing redacted layout fixtures. 3) Post in Syntax Error and other hackathon Discords and WhatsApp groups, where Local LLM finance projects already exist. 4) Offer GDG on Campus and Takshila members a weekend pass for their hackathon teams. 5) Ask the Superteam India builder chat who is building money tools. The target is 10 conversations, 3 paying, and 5 apps that opt in to drift reports.

Why not just AIThe user's own agent can write a parser for one statement you hand it, but it has to read that statement in the cloud, which is exactly what this audience refuses. It also cannot know that Kotak changed its layout this week for 200 other people. That knowledge needs opt-in structure-only fingerprints (token types, column buckets, header strings, no amounts or names) pooled across many users, which one agent or one developer can't see. Munim ships a deterministic, balance-verified parser plus signed rule updates that run offline. An LLM is only a fallback for unparseable rows, and it runs locally.

[ Inspired by ]What it borrows, and what's different

VyayaMitra
On-device Indian statement parser. It verifies running balances and remembers formats that reconcile, with no network use. VyayaMitra learns formats only on one device. Munim is infrastructure, not an app. It pools structure-only layout fingerprints across many apps, so a layout fix found once reaches everyone as a signed rule pack.
FinAnalyser / AarogyamFin (cloud statement analysers)
Paid per-statement or monthly parsing for 50+ Indian banks, sold to CAs, DSAs and NBFCs. They need the PDF uploaded and target lenders. Munim runs in the end user's browser or phone, is sold to developers as an SDK, and is priced for indie budgets.
Finance Copilot
Local-first statement parsing with balance-checked rows and a local LLM for categorisation. It is a single app with one dedicated bank parser. Munim is a shared, versioned parser layer with drift detection that any of these apps can embed.

[ Sources ]11 cited

Indian bank statements come in 30+ PDF formats, and Kotak's format reportedly changes every 6 months. Regex parsers break constantly, so maintenance is the real cost.dev.to ↗
A pure-play fintech without an RBI, SEBI, IRDAI or PFRDA registration cannot become an Account Aggregator FIU. Indie apps therefore can't pull bank data by API, and statement PDFs remain the only route.casparser.in ↗
Many separate on-device Indian finance projects exist, each parsing statements on its own. VyayaMitra is a PWA with no uploads, runs offline, and remembers formats only on the user's device.github.com ↗
LocalFinance is another 100% offline finance app for the Indian banking and credit-card ecosystem, which shows the duplicated-parser problem.github.com ↗
HumDono markets in-browser parsing of HDFC, SBI and ICICI PDFs with no statement reaching its servers, so privacy is now a selling point for Indian finance apps.humdonoapp.com ↗
Finance Copilot ships a dedicated Kotak parser plus a generic parser verified against running balances, with local LLM categorisation. This confirms that balance-chain verification is the working approach and that only a few banks are covered per project.

[ Idea ]Found 13 Nov · 1 GitHub · 1 Devfolio · 1 DEV

Munim

Shared bank-statement parsing for private finance apps, run on-device.

Should you build this?Checking

This idea is still going through the six checks.

0 products · 11 sourcesPending

The problem

Indian bank statement PDFs come in dozens of layouts that change without warning, so every private finance app has to build and fix its own parser.

Seen in 1 GitHub · 1 Devfolio · 1 DEV

The idea

Munim is an open-source parser that reads statements on the user's device and shares layout fixes without sharing any user data.

1Drop in a password-protected statement PDF; it is read in memory.
2Each row is marked verified or mismatch by checking the running balance.
3Pro users get signed fixes within 72 hours of a detected change.

[ The six checks ]0 of 6 done

1Not built in alreadyNot checked

2An AI agent can't do itNot checked

3Nothing similar existsNot checked

4People pay for thisNot checked

5Strong scoresNot checked

6Real sourcesNot checked

1 · Not built in alreadyNot checked

2 · An AI agent can't do itNot checked

Who paysThe developer or small team shipping the finance app, paying from their own pocket or a small product budget. Not the end user.

[ Inspired by ]What it borrows, and what's different

VyayaMitra
On-device Indian statement parser. It verifies running balances and remembers formats that reconcile, with no network use. VyayaMitra learns formats only on one device. Munim is infrastructure, not an app. It pools structure-only layout fingerprints across many apps, so a layout fix found once reaches everyone as a signed rule pack.
FinAnalyser / AarogyamFin (cloud statement analysers)
Paid per-statement or monthly parsing for 50+ Indian banks, sold to CAs, DSAs and NBFCs. They need the PDF uploaded and target lenders. Munim runs in the end user's browser or phone, is sold to developers as an SDK, and is priced for indie budgets.
Finance Copilot
Local-first statement parsing with balance-checked rows and a local LLM for categorisation. It is a single app with one dedicated bank parser. Munim is a shared, versioned parser layer with drift detection that any of these apps can embed.

[ Sources ]11 cited

Indian bank statements come in 30+ PDF formats, and Kotak's format reportedly changes every 6 months. Regex parsers break constantly, so maintenance is the real cost.dev.to ↗
A pure-play fintech without an RBI, SEBI, IRDAI or PFRDA registration cannot become an Account Aggregator FIU. Indie apps therefore can't pull bank data by API, and statement PDFs remain the only route.casparser.in ↗
Many separate on-device Indian finance projects exist, each parsing statements on its own. VyayaMitra is a PWA with no uploads, runs offline, and remembers formats only on the user's device.github.com ↗
LocalFinance is another 100% offline finance app for the Indian banking and credit-card ecosystem, which shows the duplicated-parser problem.github.com ↗
HumDono markets in-browser parsing of HDFC, SBI and ICICI PDFs with no statement reaching its servers, so privacy is now a selling point for Indian finance apps.humdonoapp.com ↗
Finance Copilot ships a dedicated Kotak parser plus a generic parser verified against running balances, with local LLM categorisation. This confirms that balance-chain verification is the working approach and that only a few banks are covered per project.