Receipt & Invoice Extractor - OCR to Structured JSON

OCR reads every line of text from the photo, then a heuristic extractor looks for dates (multiple format patterns), money amounts (with currency symbols and decimals), tax keywords (VAT, TAX, GST, sales tax) and line items (lines ending in a money pattern). The largest amount near a "total" keyword is treated as the receipt total.

Vendor (first non-empty line, usually the store name), dates (every date pattern found), total (best guess), tax amount, currency (USD/EUR/GBP/JPY/INR detected from symbols), and a list of line items with descriptions and amounts. Output is JSON-ready for bookkeeping, expense reports or CRM imports.

Yes — the same field-extraction logic applies to both. Itemised invoices with line descriptions and amounts produce richer item lists; thermal-printed receipts work too, though stretched/faded thermal paper photographs can lose accuracy.

Common causes: the receipt was photographed at an angle (so OCR misreads numbers as letters), the total line uses a non-English word like "TOTAL" written in a different language, or the printed receipt has faded thermal text. Re-take the photo straight-on with good lighting and the OCR should pick it up.

The extractor recognises USD ($), EUR (€), GBP (£), JPY (¥), INR (₹) from currency symbols in the receipt text. The numeric values themselves are not converted — the user (or downstream system) should apply exchange rates if needed.

Yes — download the JSON and pipe it into QuickBooks, Xero, Concur or a custom expense API. The structure is stable: vendor, dates[], total, tax, currency, items[{description, amount}]. Build a small mapping script and you have automated receipt entry.

Yes — to our OCR service, processed in isolation, and deleted within 24 hours. We do not retain receipts or use them for training. For very sensitive expense reports (private investigations, journalism), consider running OCR locally.

Free during beta, no daily cap, no signup. The structured JSON has no watermark or branding.