Stop approving fake payment screenshots by hand

Your app posts the receipt image and what it expected. Payment Pro Myanmar (PProMM) replies with approved, pending review or rejected — plus the reason.

Open the console Try the API

AI reads the receipt

A vision model extracts wallet, amount, transaction ID, receiver and timestamp, and flags edited or blurry screenshots.

Rules decide

Amount, receiver, receipt age and duplicate transaction IDs are checked against what your app expected.

Humans handle the grey zone

Anything uncertain lands in a review queue where your staff approve or reject in one click.

Your app gets told

Signed webhooks deliver the final verdict back to your callback URL, so your flow continues automatically.

One call from your app

POST /api/public/verify-receipt
x-api-key: rgk_live_xxx
Content-Type: application/json

{
  "image_base64": "data:image/jpeg;base64,...",
  "expected_amount": 25000,
  "expected_currency": "MMK",
  "expected_receiver": "Aung Aung",
  "external_ref": "order_9182"
}

→ 200 OK
{
  "id": "0f3c…",
  "verdict": "approved",
  "confidence": 0.94,
  "reason": "Amount, receiver and transaction ID all matched.",
  "reason_codes": [],
  "extracted": { "wallet": "KBZPay", "amount": 25000, "txn_id": "…" }
}

Poll GET /api/public/verify-receipt/{id} or receive a signed webhook when a reviewer makes the final call.