Onboarding cut from days to minutes, without cutting compliance
A digital onboarding and KYCflow built around source-verified identity,AI identity checks, tuned AMLscreening, and a decision trailthat holds up in anaudit.
The Challenge
Every applicant was cleared by hand: a team checked IDs by eye, screened names against sanctions lists in a spreadsheet, and re-keyed the same details into three systems. It took up to three days, and roughly a third of applicants dropped off before finishing. Each manual step was also a control gap with no record of why anyone was approved, which is a hard thing to defend to a regulator or an auditor.
Days to open an account
A team checked government IDs by eye, screened names against sanctions lists in a spreadsheet, and re-keyed the same details into three systems. The loop took up to three days, and a third of applicants gave up.
Every manual step a control gap
Two reviewers reading the same ID reached different answers, a tired analyst could miss a near-match, and there was no record of why anyone had been approved.
Compliance had to be provable
The regulator expects customer due diligence to be consistent, evidenced, and reproducible, and PDPL expects personal data held under defined controls. Inboxes met none of that.
Our approach
Every build follows the same software development life cycle, from requirements and design through build, testing, and support. Each phase is planned, demoed, and signed off before the next begins, so quality is engineered in rather than checked at the end.
Discovery & requirements
We wrote a requirements baseline and a compliance map that ties every KYC, AML, and data-protection obligation to a specific step, control, and owner in the flow.
(Outcome):
Architecture & design
We designed a verification pipeline around defensibility: identity confirmed against the source, computer vision for the document, biometrics for the person, and a reviewer in the loop for anything uncertain.
(Outcome):
Build
We built the onboarding flow, the screening, and the reviewer tooling in milestones, demoing each control so the compliance team could see exactly how a decision was reached.
(Outcome):
Testing & UAT
We calibrated the match and liveness thresholds on a labelled set of genuine and tampered documents to balance false accepts against false rejects, then ran user acceptance testing with the compliance team.
(Outcome):
Deployment & support
We released through a feature flag to a small cohort before opening the flow to everyone, with a complete audit trail behind every decision.
(Outcome):
Outcomes
Onboarding cut from up to 3 days to under 10 minutes
Roughly 35% fewer applicants dropping off mid-flow
Around 70% fewer applications needing a manual review
Every decision recorded with its evidence in one exportable, PDPL-aligned audit trail, replacing data scattered across inboxes