Thoughts and Feelings on Claude Design

It argues that as Figma grows complex design systems centered on components, styles, variables, props, the distance from the actual implementation medium has widened.

• Figma has long claimed to be the source of truth, but it argues that its locked format and unwieldy structure actually put it at a disadvantage as a learning foundation in the agent era.
• By contrast, it argues that Claude Design doesn't presuppose the canonicity of design files, and instead positions itself as a tool that doesn't hide the fact that it's built on HTML and JS.
• Its sibling relationship with Claude Code, support for repository imports, and a flow that connects design and implementation within a single conversation are presented as strengths.

The piece also cites examples of Figma file complexity. For instance:

• 946 color variables linked across multiple modes
• A modal footer's 12 variants and various props
• Complex structures like slider effect styles and a combo input's 16 variants

It says this complexity makes debugging and back-porting increasingly difficult, ultimately turning the act of porting design changes made directly in code back into Figma into a burden.

It suggests future design tools may split into two branches.

• One direction, like Figma Make, maintains the design file's canonical status
• The other, like Claude Design, takes a truth to materials path that stays close to code

Along another axis, it also mentions the possibility of a purely exploratory canvas with little expectation of code, or even a high-fidelity compositing tool closer to Photoshop.

Comment reactions are split. Some see Claude Design as potentially lowering the initial barrier to entry, while others believe it can't replace Figma unless it eliminates the complexity itself. Still others argue that Figma remains the collaboration standard at large organizations, and that the core problem isn't the tool but the handoff gap and the maintenance cost of design systems.