What I do

  1. FrameUnderstand the problem with the people who have it, then decide what to test first.
  2. Human-led

    01Understand

    Talk to users and stakeholders, on site where possible. AI clusters the notes, I check every pattern.

  3. Human-led

    02Define

    Turn findings into insights and testable hypotheses, and pick the riskiest assumption to test first.

  4. MakeExplore many directions fast, then build working prototypes on the real design system.
  5. AI-accelerated

    03Explore

    Generate many directions fast in Figma Make and Claude Design. Exploring is cheap, choosing is the skill.

  6. AI-accelerated

    04Prototype

    Build working prototypes with Claude Code on the real design system, tokens synced via Figma MCP.

  7. ProveTest with users, hand over code-ready components and feed what we learn back into the loop.
  8. Human-led

    05Validate

    Users test the working prototype. AI agents check accessibility and design system consistency first.

  9. Human-led

    06Ship & Learn

    Hand over code-ready components, measure the signal from step 02 and feed it back into the loop.

Why AI sits in the middle

The tools speed up the making. Framing the right problem, talking to users and deciding what ships stay with me. That is why the first and the last steps are human-led, and the middle runs as fast as the tools allow.

Ways to work together