Jeff Zych: Claude Code is reshaping my design process

Published on his personal site in February 2026, Jeff Zych’s post is a practitioner’s account of a workflow shift rather than a general guide. Zych is a product designer who noticed a concrete change in his own behavior: his git commit frequency went up, and the time he spent in Figma went down. The post explains why.

What changed

Zych began using Claude Code to move from a described user experience directly to a functional prototype in code, bypassing the static wireframing stage he had used previously. The workflow he describes breaks down by feature scope.

For minor features, he skips Figma entirely. He describes the problem to Claude Code, it builds an interactive artifact, and he iterates from there. For major features, he still does upfront discovery work — understanding the problem, mapping user needs — but then uses Claude Code for brainstorming, generates interactive artifacts early, and builds in code rather than static mockups. Figma remains in the workflow specifically for visual design exploration: color, typography, and component testing where the visual rendering still matters more than interactivity.

Why functional prototypes revealed more

The core observation in the post is that working in real code surfaces UX problems faster than working in static mockups. Zych gives specific examples: issues with sorting behavior, filtering edge cases, and how content behaves at different lengths are all visible in a functional prototype and invisible in a wireframe. The feedback loop from functional code is tighter because the prototype behaves the way the real product will behave.

What stayed the same

Zych is careful to note that the approach to design work itself did not change — only the medium. He still focuses on understanding users and their needs, explores multiple options, iterates based on what he observes, and tests and refines. The shift is in where that iteration happens: in working code rather than in design files.

Who it is useful for

Product designers and UX designers who work closely with codebases or on engineering teams where direct code access is practical. The post is also useful as a concrete data point for teams debating whether to adopt AI coding tools in design workflows — Zych is specific about where the gains are and where he still uses traditional tools.