Invoice Extract reads an invoice or a receipt — a photo, a scan or a PDF — and pulls out the fields you would otherwise type by hand: vendor, invoice number, date, due date, subtotal, tax and total, plus one row per line item with its amount.
The result is a two-column CSV (field, value) that opens in any spreadsheet, ready to copy into your bookkeeping. The document is read by an AI vision model in the cloud (Google Gemini by default, OpenAI as fallback). It is a professional tool: included in Abzar Plus, with one free try a day on a free account, and each run uses 1 credit.
How to use
Upload a photo, scan or PDF of one invoice or receipt.
Run it — the file goes to the AI model and 1 credit is used (returned if the run fails).
Check every amount, the date and the invoice number against the original.
Download the .csv, or copy the values straight into your accounting sheet.
FAQ
Which fields does it extract?
Vendor, invoice number, date, due date, subtotal, tax and total, and one row per line item with its name and amount. Fields missing from the document come back empty or are left out; addresses, VAT IDs and bank details are not part of the output.
Is it accurate enough for bookkeeping?
It saves typing, not checking. Printed invoices are usually read correctly; crumpled receipts, faded thermal paper and handwriting cause errors. A wrong digit in a total matters, so compare every amount before you book it.
Does it work with Persian and German invoices?
Yes. The field names in the CSV stay in English (vendor, total and so on), while the values are taken from the document — so a Shamsi date usually stays Shamsi. Check dates and amounts either way.
Can I process several invoices at once?
No — one file per run, and a PDF holding several invoices is read as one document. Run each invoice separately so the fields do not mix. The file is sent to the AI provider for that run only and is not kept by Abzar.