Building got cheap.Knowing what to builddidn't.

Everyone loves the plan.

Nobody has spoken to a user.

Six interviews later, it's wrong.

It usually is. The build started three weeks ago.

Two directions. No evidence either way.

So it goes to whoever argues hardest.

We ship the wrong one.

Then spend two quarters taking it back out.

Most teams never leave this part.

Nothing pruned, nothing owned. It just grew.

Then someone builds a bridge.

AI carries the weight now. It still doesn't know where you're going.

This is what I carry over.

Six things I'm good at.

• Design to productionSame person from first frame to deploy.FigmaReactNext.jsTailwind
• Design systemsAnd the rules that stop the drift.TokensStorybookFigma libraryGovernance
• AI in the pipelineAgents take the repetitive half.AgentsAuditsCodegenQA sweeps
• Prototypes you can clickIt runs in a browser, not on a slide.three.jsR3FGSAPWebGL
• Proof before polishI show people while it's still ugly.AuditsHeuristicsA/BInterviews
• Holds its shapeLong names, empty states, old phones.Edge casesResponsivea11yReal content

I design it, then I build it.

Figma to production. No handover, so nothing falls through one.

I'd rather build the system.

One library, a hundred screens.

AI does my chores.

The thinking stays mine.

This side is real.

You get the time back.