Skip to main content

ProofPay AI helps small merchants avoid dispatching goods on fake or delayed “Payment Successful” screenshots. A screenshot is a claim; a matching successful record in the merchant’s own transaction CSV is evidence. A vision AI model reads the screenshot (amount, reference/UTR, time, recipient) in English or Hindi, with local OCR and manual correction as fallbacks. A deterministic rule engine, the ProofPay Evidence Match Engine, then compares the claim to the merchant’s CSV and returns one of four outcomes: verified credit found, needs manual review, no matching credit found, or unable to verify. Each outcome comes with a safe next action. The AI only reads and never decides. The CSV is parsed in memory and never sent to any AI, reports are masked, and the tool never gives fraud verdicts. It is decision support, not a bank or payment gateway.

Built with

pythonfastapihtmlcss

Prize Categories Entered

  • DigitalOcean

    Best Open-Source AI Project

    Submitted