detect weird charges before they become an "i definitely didn't buy that" situation.
upload a bank statement and get anomaly detection, spending insights, merchant cleanup, and a month-end balance prediction.
everything runs in your browser. no backend, no database, no data leaving your device.
- upload csv, xls, xlsx, txt, or pdf statements
- flag suspicious transactions using statistical anomaly detection
- automatically categorize spending across 40+ categories
- normalize messy merchant names
SWGY*O123 UPI 9988@okaxis
becomes
Swiggy
- spending breakdown with charts
- transaction search, sorting, and filtering
- end-of-month balance prediction
- Next.js 16
- TypeScript
- Chart.js
- SheetJS
- PDF.js
- PapaParse
all processing happens client-side.
git clone https://github.com/yourusername/finlens.git
cd finlens
npm install
npm run devopen http://localhost:3000
transactions are compared against your usual spending patterns using z-score based anomaly detection.
a ₹12,000 food delivery order might be perfectly normal for someone.
for most people, it probably deserves a second look.
merchant names are cleaned up using a pattern-matching system built around common Indian banking descriptions because banks seem committed to making transaction names unreadable.
tested with exports from:
- HDFC
- ICICI
- SBI
- Axis
- Kotak
and hopefully whatever spreadsheet format your bank's intern invented.
- scanned PDF statements are not supported yet
- forecasting uses simple spending trends, not financial wizardry
- some local merchants may not be recognized
- OCR support for scanned PDFs
- recurring subscription detection
- budgeting tools
- PDF report exports
- smarter forecasting
built out of equal parts curiosity and annoyance with Indian banking UX.