AI automation
In short
Photos or PDFs of invoices become structured data: ChatGPT or Gemini reads out date, vendor, net, VAT and gross – you enter the result into Google Sheets. It goes fully automatic with Make.com/Zapier plus OCR: every new attachment in Gmail or Drive lands extracted directly in your sheet or accounting tool.
Read this invoice and return a table with date, vendor, net amount, VAT and gross amount.
Limits: Manual per receipt, no automatic filing.
Read this invoice and return the result as JSON with the fields date, vendor, net, vat, gross.
Added value: Fully automatic capture straight from your inbox.
Very reliably on clearly legible invoices – even from photos. Crumpled receipts or exotic layouts are harder; there the JSON output plus a quick totals check helps. For tax-relevant data: spot-check.
As pre-capture, yes: the structured table saves the typing. You still need to keep the receipts themselves, and retention and formal rules vary by country – when in doubt, check briefly with your tax advisor.
Make/Zapier have free quotas that cover a few receipts per month; OCR services like Mindee charge per document (cents). At low volume, full automation stays very cheap.