In an Era Where AI Does the Design, Why Will Design Systems Become Even More Important?

With the advancement of AI-based design tools, generating ideas (divergence) has become easy, but implementing them into a consistent product (convergence) requires an even more sophisticated design system.

Design systems in the AI era must be restructured around Machine Interpretability rather than human readability.

• Semantic Naming: Using clear, context-rich names such as 'Button' instead of 'Btn'
• JSON-based Hierarchical Structure: Systematic structuring in the order Primitive → Semantic → Components → Patterns
• md File-based Prompt Protocols: Strictly defining roles, workflows, style rules, and more

In actual workflows, a tech stack is used in which Figma Variables are converted into JSON, which is then converted into Tailwind v4 CSS variables via Claude Code, allowing AI to instantly access the design system.

As the tool ecosystem—including Google Stitch, Claude Design, and others—matures going forward, it is expected that an era will come where a flawless system can be automatically built from just design system files or screenshots.