How It Works
From PDF to structured Excel in seconds
Upload Your PDF
Drop your bank statement PDF into the upload zone. Encrypted or password-protected statements are fully supported.
Text Extraction
The PDF is parsed entirely in-memory using pdfjs. Text is extracted page-by-page, sorted by position, and prepared for analysis. For image-heavy or scanned PDFs, each page is rendered to an image and sent to a vision AI model for OCR — so even scanned statements work seamlessly.
Bank Detection
A parser registry scores the extracted text against known bank formats — Union Bank, RBC, Scotiabank, and any previously learned formats. The best match is used automatically.
Transaction Parsing
If a known parser matches, transactions are extracted locally with regex — no API calls needed. For new bank formats, an LLM analyzes the structure and you preview results before confirming.
Quality & Reconciliation
Every extraction is validated — balance reconciliation, date checks, contamination detection. Issues are flagged clearly so you always know the data quality.
Excel Download
Transactions are formatted into a structured Excel workbook with mapping and raw data sheets — amounts, dates, payees, balances — ready for your workflow.
Your Privacy, By Design
We built this tool to process — not to collect
No Data Stored
Your PDF is processed entirely in-memory. Once the Excel is generated and delivered, the file buffer is discarded. Nothing is saved to disk, no database, no logs.
No User Accounts
There is no authentication, no sessions, no cookies tracking you. Every upload is a stateless, isolated transaction.
Local-First Parsing
For known bank formats, everything runs locally on the server with regex — your statement text never leaves the machine. LLM and OCR extraction are only triggered when needed (unrecognized formats or scanned PDFs), and you approve results before anything is finalized.
Only Structure is Saved
When an LLM learns a new bank format, only the parsing rules are saved (regex patterns, column layout) — never your transactions, balances, or personal data.
Zero storage. Zero tracking. Your data stays yours.
The tool processes your statement in real-time and forgets it immediately after.