Meaghan Choi, a Design Lead at Anthropic, now spends as much time in the AI coding environment Claude Code as in Figma, fundamentally changing her workflow.
The traditional design handoff is evolving from static Figma mocks to draft pull requests (PRs) created with AI, providing engineers a functional, higher-fidelity starting point.
Claude Code is used across the entire product lifecycle, from initial codebase analysis and 'zero-to-one' exploration to implementing final UI polish.
The tool's power is enhanced by customizable `claude.md` files, which allow users to define a persistent persona and instructions for the AI, tailoring it to specific roles like a designer.
12 quotes
Concerns Raised
AI models can produce generic, unoriginal UI designs ('shadowy and tailwind CSS look') without specific visual guidance.
There is a risk of non-specialist contributors introducing errors or incidents when pushing code to production, even with code reviews in place.
Opportunities Identified
Empowering non-engineers (designers, PMs) to contribute directly to the codebase, speeding up UI polish and minor fixes.
Transforming the design handoff process from static mocks to interactive, code-based prototypes or draft PRs.
Accelerating the entire product development lifecycle by using AI for codebase exploration, planning, and implementation.
Creating highly fluid and collaborative teams where roles are less defined by traditional titles and more by contribution.