Receipt Capture.

Employees give on their own, too. They photograph the receipt, we verify it, log the giving, and queue the match. Nothing for them to type, nothing for you to chase.

Each receipt finds the right path automatically.

High confidence ≥0.85

OCR is confident, nonprofit matches a verified record, amount and date are clear. Donation logged automatically; match fires; employee gets an Impact Story (Layer 1).

Medium confidence 0.60-0.84

OCR is mostly clear but uncertain on one field. Surfaces on the employee's dashboard as "Yes, that's right / Adjust / Not mine" with the existing extraction.

Low confidence <0.60

OCR is unsure; nonprofit doesn't match cleanly. Routes to an admin review queue. Admin picks the nonprofit + confirms amount before logging.

Snap a photo. That's the whole flow.

📸 Photo upload

From the employee dashboard or "Add Something I Did" page. JPG / PNG / WebP / PDF up to 10 MB. Built to work on a phone, in the moment.

👀 A human in the loop

Anything the scan isn't sure about goes to your GozAround team, who clear the review queue daily, visible in your activity digest, like everything else we do.

Duplicate detection

If a donation with the same employee + nonprofit + amount already exists within 7 days, marks as a duplicate so a single receipt isn't counted twice.

Pluggable OCR

AWS Textract by default; stub mode for offline dev; provider interface lets us swap OCR vendor without touching application code.

Off-the-clock giving, captured and credited, with no admin burden.