Industry challenges
Claim volume
Thousands of refund requests reviewed by hand, against a response-time target.
The photo is the only evidence
A returnless refund is decided on one image that nobody has the means to check.
Edited and recycled images
A photo can be retouched, reused from an earlier order, or generated by AI.
The cost of a false positive
Refusing an honest customer usually costs more than the refund itself.
Documents processed
- โDamaged product photos
- โDelivery notes and proof of delivery
- โInvoices and purchase receipts
- โUPS, FedEx and USPS receipts
- โReturn labels and slips
- โCertificates of authenticity
- โBank details for refunds
- โScreenshots attached to the claim
Automatic validations
- โConsistency between the photo, the order and the delivery date
- โImage metadata: device, timestamp, traces of editing
- โPhoto already submitted on another claim
- โAmounts matching between invoice and request
- โSignature and geolocation on the proof of delivery
- โAlignment with card-network chargeback evidence requirements
- โSigns of AI generation, as a complement to your existing controls
What is covered
3,200+
document types supported
24
reading languages
48 h
to run a pilot on your files