Learning pathDesign Systems

Preparing a Design System for AI Tools

For design system leads and designers who want components, guidelines and code to work reliably with AI agents and Figma Make. You will learn to make the system readable to AI and check its output.

8 lessons · 3 hr 8 min, plus 1 optional (6 min more with the optional lessons)

  1. Beginner (not covered)
  2. Intermediate
  3. Advanced

What you’ll learn

  • Write design system documentation as AI-readable Markdown guidelines
  • Connect components to code with Code Connect, Storybook MCP and zeroheight MCP
  • Brief an AI agent to build on-system UI, then audit and test its output
  • Set up Figma Make with guidelines and templates so output follows the system
  • Judge which agentic design system uses are realistic now and where humans must review

Lessons

  1. AI Tooltip 7 min Intermediate

    Start here: see why design systems must be readable by AI, and how Markdown files prepare the system.

    Explains why design systems need to change now that AI tools do much of the manual UI work, and shows how to build an AI-friendly design system from Markdown files. Includes an extraction prompt for documenting a product's identity, visual style, layout, components and UX rules.

  2. Figma 5 min Intermediate

    Adds how Code Connect maps Figma components to code, so the Figma MCP server gives agents real component names and props.

    A short Figma developer-advocate video explaining how Code Connect maps Figma design components to production code, so that AI agents using the MCP server see component names and props rather than raw style data. It is aimed at teams who already maintain design system components.

  3. zeroheight 6 min Intermediate

    Moves from setup to practice: briefing an agent through MCP with system docs, building on the code, then auditing the result.

    Luke Murphy of zeroheight demonstrates how an AI agent can be briefed with design system documentation through MCP, build a prototype on the system's code, and be audited against the design system. It suits teams already working with design systems who want to try AI agents on that work.

  4. Brad Frost 15 min Intermediate

    Shows the same idea from the code side: Storybook MCP lets AI assemble a page from existing components and test it.

    A live demo of Storybook MCP, in which AI generates a new search page from existing design system components, runs it through Storybook's testing tools, and catches an accessibility issue. It is aimed at people already working with component libraries and AI-assisted interface building.

  5. Nick Babich 6 min Intermediate

    Optional detour: auditing a Figma system with Claude Code to check readiness and see what the audit costs to run.

    Shows how to audit a Figma design system and fix its issues using Claude Code's design plugin. The video also covers the cost of running such an audit.

  6. zeroheight 66 min Intermediate

    Hour-long deep dive putting the demos into practice, covering how to brief, audit and test agent code and avoid drift.

    The first session of a zeroheight webinar series shows how to brief an AI coding agent with a design system's package and documentation, then audit and test the code it produces to keep it on-system. It uses the zeroheight MCP and Claude Code and covers common pitfalls.

  7. Figma 30 min Advanced

    Moves into Figma Make: wiring components, guidelines and templates so generated output follows the system.

    Shows how to set up Figma Make so its output follows a design system, covering guidelines, file structure, component wiring, prompting, templates and presets. Uses a large open-source enterprise design system as the worked example and covers how to recover when gaps appear.

  8. Figma 16 min Advanced

    Meta's case study shows the earlier techniques applied at scale to a legacy system adapted for AI generation.

    A Config 2026 talk in which Meta's business design system team describes rebuilding its legacy design system for AI-driven design in Figma Make. It covers the problems they hit, how they resolved them with Figma, and a three-step playbook teams can use without rebuilding their whole system.

  9. zeroheight 44 min Advanced

    Closes the path by separating hype from what is realistic for agentic design systems and where the field is heading.

    A conversation between Luke and Michelle on what agentic AI means for design systems, separating hype from what is currently possible and where they see the field heading.

When you've finished

Next path: Scaling Design Systems Across Many Brands and Products (Advanced)

Want an easier start? Building Design Systems in Figma

All paths in Design Systems

All paths
  • Learning pathIntermediate → AdvancedDesign Systems

    Running a Design System Day to Day

    For designers and design system leads who already build components and want to keep a shared system working. You will be able to manage tokens, flexible components, governance, adoption and handoff to code.

    7 lessons · 3 hr 9 min

  • Learning pathIntermediate → AdvancedBranding

    Building and Governing Brand Identity Systems

    For designers and content specialists ready to move beyond single assets, this path shows how to build, scale and govern brand identity systems using real company cases and guidelines.

    5 lessons · 2 hr 59 min

  • Learning pathIntermediateDesign-Dev Collaboration

    Design to developer handoff in Figma

    For intermediate designers who want developers to build their designs accurately. You will learn to structure Figma files, prepare designs for handoff, and work alongside engineers.

    6 lessons · 2 hr 41 min

  • Learning pathIntermediateIllustration

    Intermediate Illustration: Vector, Texture and Systems

    For designers who already make basic vectors or digital drawings and want polished, distinctive illustrations. You will be able to produce detailed Figma and Photoshop work and set up an illustration system that a whole team can use.

    5 lessons · 3 hr 21 min, plus 1 optional

Search and sort within Preparing a Design System for AI Tools

Sign in to mark lessons complete and see how far through you are.