Deep Trace is an AI-powered anti-counterfeit product that allows users to verify physical products using their smartphone camera.

The technology was sophisticated. The experience didn't need to feel that way.

When I began working on the product, only 46% of users who started verification were completing it. I led the end-to-end product redesign—from research and UX through UI, testing, design systems and iteration—to understand where users were struggling and make verification easier to complete.

Users had to interpret instructions while positioning and scanning a physical product correctly. Screens carried too much information, important actions competed for attention, navigation wasn't always obvious, and error states didn't provide enough guidance.

For a product designed to reduce uncertainty about whether something is genuine, the experience itself was introducing friction.

I owned the experience from understanding the problem through design, testing, release and iteration, working directly with Cypheme's co-founder alongside engineering and marketing.

Product analytics showed where users dropped out. Research helped uncover why.

I combined behavioral data with user feedback, stakeholder interviews, surveys and prototype usability testing. I looked for hesitation, misunderstood instructions, failed attempts, retries and moments where users weren't sure what to do next. Research → Prototype → Test → Release → Measure → Refine

Designing for clarity, not complexity.

Research pointed toward a simple principle: users didn't need to understand everything happening behind the AI. They needed to understand what the system expected from them at each moment.

I broke verification into smaller guided steps, introduced information only when it became relevant, strengthened camera guidance and processing feedback, and redesigned failure states so an unsuccessful scan became a recoverable moment rather than a dead end.

The redesign wasn't only about individual screens. I created a reusable visual and interaction system around recurring product states—actions, loading, analysis, success, warning and failure—so the experience could remain consistent as Deep Trace expanded.

Verification completion increased by 26 percentage points following the redesign and subsequent iterations.