Building an Agentic Design System
summary
Describes building three Claude skills that became a retrieval-augmented pipeline for design systems, with auto-reporting, adoption tracking, and precise generative UI. Presents a four-layer compiler-like infrastructure (protocol/instructions, structure/index of components, components/code+machine-readable metadata, and tokens/design values) that lets AI consume a design system with full context, shows measured improvements (faster task completion, precision and consistency gains, better component discovery), real examples (duplicate detection, adoption gaps, auto-reports), and discusses maintenance challenges and an open-source, replicable implementation.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:11Design system governance issues
- 5:44Agentic design system layers
- 11:01Instructions and skills in code
- 15:19Token auditing and reports
- 20:51Testing precision and consistency
- 24:45Benefits and governance output
- 25:48Q&A
related talks
30mBuilding the systems that build the software
43mDesign Thinking: Generative AI Style
06mInterview
30mUsing AI to Create Better Requirements and Designs- 43mDesign Thinking: Generative AI Style
27mThe Shape of AI : How designers integrate AI into their workflow
12mThe future of design systems and AI
13mDesign systems and how to use it