How it works
The converter runs pdfplumber, a well-known PDF extraction library, compiled to run inside your browser.
Reading the statement
It finds the transaction table from its header (Date, Description, Money in, Money out, Balance and their usual variants), then reads every row by the position of each word on the page. Descriptions over several lines, dates printed once per day, and balances shown only at the end of the day are handled.
Checking it
Every statement prints balances. The converter adds up the transactions and compares: opening plus transactions must give the closing balance, and every printed running balance must match. If a single amount was misread, the check fails and tells you where.
Your statements stay on your computer
The PDF is read by code running in this browser tab. Nothing is uploaded: there's no server that could receive it. You can check: load the page, turn off Wi‑Fi, and convert a statement.
We count anonymous events (a file was read, a check passed, a CSV was downloaded) to know what to improve. No file names, amounts or text from your statements are ever sent.
The reading engine is open source: pdfplumber.js.