The problem
Account-opening checks were split across teams, inboxes, and manual reviews. KYC, legal searches, and missing-document follow-ups slowed onboarding and made the customer experience harder to control.
What I did
I mapped the current journey, designed the automated workflow, translated policy into decision logic, defined the exception paths, and worked with engineering on the requirements, API integrations, AI behavior, and UAT.
The system
- AI agents and OCR extract and validate submitted information.
- APIs cross-check BVN, NIN, IDs, CAC, TIN, watchlists, PEP, and other risk signals.
- A decision engine recommends approve, restrict, or flag.
- High-risk and incomplete cases move to a human review queue.
- An audit trail records every check, decision, and exception.
What changed
Routine verification moved toward near-real-time processing, with fewer manual touchpoints, clearer exception handling, stronger traceability, and a more scalable onboarding foundation.