All products

Organise receipts with AI

Extract receipt fields into reviewable records, keep the source of each amount and flag possible duplicates before anything is posted.

Organise receipts with AI when the same purchase might appear in both an inbox and a folder of scanned expenses. The receipt-processing skill keeps the source beside the extracted amount so a reviewer can decide whether two records refer to one transaction.

Start with a mix of clear and difficult examples. A faint total, an ambiguous currency symbol or missing tax information should remain visible as a question. The goal is a structured draft someone can verify against the document, not a complete ledger assembled by guesswork.

What it does

  • Extract the merchant, date, amount and currency.
  • Preserve the reference back to the original document or message.
  • Flag an unreadable field instead of inventing a value.
  • Compare potential duplicates using the evidence from both sources.
  • Summarise records that have already been extracted.

Install and use

  1. Provide authorised files, existing OCR text or an explicitly permitted inbox source. The optional Python 3 helper summarises existing structured JSON; it does not perform the OCR.
  2. Ask, “Compare this receipt photo with its email confirmation. Show which fields match and which still need a person to read the original.”
  3. Review uncertain amounts and duplicates before authorising any entry in an accounting system.

FAQ

Source

Made by Receiptor AI. Original skill: receipt-processing, published on GitHub under the MIT licence.

View on GitHub