Slip verification is a full-time machine-learning problem: formats change, photos are bad, and mistakes cost money. We own that problem end to end and hand your platform a simple API.
Every slip format
Paper deposit slips, including partly handwritten ones, plus ATM and CDM receipts, online-banking confirmations, and mobile-app screenshots, across all major Sri Lankan banks and payment apps.
20+ fields per document
Transfer amounts separated from bank fees, dates and times, both banks, account numbers, depositor details, remarks, and every reference number printed on the slip.
Confidence on every field
Each extracted field carries its own 0–100 confidence score, so you decide what auto-approves and what gets a human look.
Fraud signals built in
Duplicate detection across references and re-uploaded images, tampering suspicion, and photo-of-a-screen flags, all returned before you release the payment.
API-first and fail-safe
One REST call in, structured JSON out. If extraction fails, you get a clean, typed error to fall back on, so payments don’t get stuck waiting on the AI.
We own the AI lifecycle
Fine-tuning, evaluation, and retraining when banks change their formats. Already running in production on The Search Bar, our EdTech platform serving 80,000+ users.