A Human-Check Queue for UPI Payment Screenshots in WhatsApp Order Flows

Description and use

In small Indian businesses, payment confirmation often relies on WhatsApp screenshots, but automatic approval can lead to fraud. A human‑review queue is proposed where an operator compares the bank reference with the OCR‑extracted reference before marking the payment as confirmed.

Problem

Small Indian businesses rely on customer‑sent screenshots on WhatsApp to confirm payments, but screenshots can be edited, reused, or mis‑matched to orders, making automated approval risky.

Outcome

Implementing a human‑review queue for UPI payment screenshots reduces the risk of fraudulent or mis‑matched payments while maintaining customer communication.

Workflow

• 1. Customer sends payment screenshot on WhatsApp.
• 2. System creates a PaymentClaim with order_ref, wa_phone, image_id and extracts reference and amount via an OCR model (suggestion only).
• 3. Claim is added to a review list; status remains pending_review.
• 4. Human reviewer checks the bank or payment app, enters the actual reference seen.
• 5. System compares entered reference with extracted_ref; if they match, status becomes matched, else rejected.
• 6. Once matched, system notifies customer with “payment confirmed.”
• 7. Additional details: acknowledge receipt instantly, detect duplicates, store image ID, log reviewer name.

Evidence

Screenshots can be edited, reused or sent for the wrong order.

Source

The reviewer checks the bank or payment app, types the reference they actually see, and the system compares.

Source

Acknowledge instantly, confirm later.

Source