About this Session
Design systems codify decisions and contracts, but they've always depended on human governance to stay honest. Documentation drifts from implementation. Tokens diverge from code. The "source of truth" depends on who you ask.
The fix isn't more process. It's choosing the source of truth that can't lie: code.
This session introduces a layered architecture for agentic design systems, where AI agents don't just consume components the way developers do, but actively enforce the contracts your system already defines. The architecture separates concerns across three primitives: skills (reusable executable capabilities), rules (project-specific constraints and decisions), and instructions (the methodology that orchestrates both).
The result: governance becomes an automated output, not an expensive special project. Weekly audits at cents per report. Drift detection without dedicated headcount. Design decisions that are explicit in the system rules and enforced by the agents that read them.
This isn't a better-indexed component library. It's a design system that governs itself.
