The more I do this work, the more firmly I believe that a well-built design system is critical infrastructure for AI. I’m presenting a talk at Into Design Systems next week about that topic, and it’s chock-full of insights from over 8 months of building out a design system and making it understandable for LLMs.
Here’s just a few things I’ll be covering:
- Reducing randomness, making AI output more reliable/repeatable
- Figma MCP vs. custom component library MCP
- Parity between code and design
- Primitive and applied tokens/variables
- A place for design in a world of prompts
- AI as a personalized guide for learning to understand code
- Don’t use AI to recreate silos
- AI helping to supercharge “old” processes
- Agents vs. skills vs. MCP
- Fixing context bloat
- Agents teams as SDLC
Just hope I’m able to fit it all in. Hope I see you there… If you can’t make it, stay tuned, I plan to keep working transparently and sharing publicly.