This invoice and receipt extractor turns a stack of bills into a spreadsheet without uploading them. Drop PDFs, phone photos or scans; digital PDFs are read instantly, and photos and scans use on-device OCR after a one-time language download. You get vendor, invoice number, dates, subtotal, tax, total, currency, tax ID and contact details in an editable table with a totals check, ready to export as CSV or JSON.
| Field | Found from |
|---|---|
| Invoice no. | “Invoice No”, “Receipt #” |
| Dates | Date and due-date lines |
| Total | Last amount on the total line |
| Tax ID | NTN, VAT, GSTIN, EIN… |
Digital PDFs instantly; photos and scanned PDFs with the OCR option, in English, Urdu, Arabic, French, German or Spanish.
Clear, typed invoices extract well; check the totals flag and edit any field before exporting.
It recognises common symbols and codes (Rs, $, €, £, ₹, PKR, AED…) and tax IDs such as NTN, VAT, GSTIN and EIN.
No. Files and text are processed in your browser. Only the model itself is downloaded, once, when you ask for it.
Yes, the Downloaded models panel removes OCR language data after a warning.
Digital PDFs are read through their text layer with PDF.js. Photos and scanned PDFs (pages rendered at 2× scale, up to five pages) are read with Tesseract OCR after the visitor downloads a language pack, which is stored in IndexedDB until cleared. Fields are then found by rules: the first substantial line as vendor; labelled invoice or receipt numbers; dates in numeric and month-name formats with due dates kept separate; the last money amount on total, subtotal and tax lines (ignoring tax-ID lines); currency from symbols or ISO codes; tax IDs such as NTN, STRN, VAT, GSTIN, TRN, EIN and ABN; email and phone. Subtotal + tax is checked against the total. All values are editable and export as CSV or JSON.