Guide

Convert PDF bank statements to structured JSON

Convert any bank statement PDF to a clean JSON array — one object per transaction with date, description, amount and balance. Free tier, no signup, OCR included.

Bank statements are the highest-value document class for financial pipelines — every AR/AP reconciliation, spend-analytics dashboard and LLM-powered accounting agent needs them as structured JSON. Generic PDF-to-JSON tools miss the point: they emit box-per-word coordinates, not accounting rows. Convert Bank Statements emits one JSON object per transaction, with dates normalised to ISO, amounts as signed numbers, and a reconciled running balance.

Step-by-step

Step 1
Upload the PDF

Any bank, any period. Password-protected files stay local — the password never leaves your browser.

Step 2
We parse the statement

200+ bank templates auto-detected; scanned pages OCR-ed; multi-line descriptions collapsed to one row per transaction.

Step 3
Download JSON

Get a JSON array — one object per transaction — plus a summary object with opening/closing balance and period totals.

FAQs

What does the JSON look like?

An array of objects: { date: 'YYYY-MM-DD', description: string, amount: number (signed, credits positive), balance: number|null, fitid: string }. A top-level summary object carries opening/closing balance and totals.

Does this handle scanned PDFs?

Yes — image-only pages are OCR-ed before parsing.

Can I feed this to an LLM for categorisation?

That's the intended flow. The JSON is compact and LLM-friendly; pair it with a chart-of-accounts prompt to auto-classify.

How is this different from generic PDF-to-JSON tools?

Generic tools produce coordinate-based JSON (one object per text box). We produce transaction-level JSON with balance reconciliation.

Related

More guides