The problem

An accountant handling several SMEs in Tangier types hundreds of supplier invoices by hand each month: supplier, date, amounts, TVA. It is slow, boring and error-prone.

The solution, step by step

  1. Clients send invoices to a dedicated Gmail address or drop photos in a shared folder.
  2. A Make scenario picks up each new file and sends it to ChatGPT (vision) with a strict extraction prompt.
  3. The AI returns structured fields: supplier, ICE, date, amount excl. tax, TVA rate, total.
  4. Make writes a row per invoice in Google Sheets, with a link to the original file and a confidence flag.
  5. The accountant reviews flagged rows only, then imports the sheet into the accounting software.

    6. Result and estimated gains

    • Data entry time cut by 70 to 80% — from 3 days to half a day per month per client.
    • Fewer typos on amounts and TVA, so fewer reconciliation headaches.
    • Every entry keeps a link to its source document for audits.