A design system that isn’t reliable for people, won’t be reliable for AI.

The AI gold rush is exposing who actually invested in their design system and who treated it as a “one-and-done project” or dismissed it as “hype.” Spoiler! A system that’s riddled with gaps and rot cannot be a reliable source of truth… for humans or AI.

💀 The mismatched tokens between design tools and code you never reconciled? Yeah... those semantics sure do matter a lot more now if you expect AI agents to use them.

💀 The “documentation” that’s just tribal knowledge rattling around in a few design system veterans’ heads? That’s a big blocker when training an LLM.

These are daily pain points for teams depending on systems, and they slow down design system maintainers already stretched thin. The tech/design debt from years of under-prioritization has always been a problem.

And now, with leadership screaming “Go do AI!” 🚀 people are frantically duct-taping gaps while stacking AI on top of a wobbly foundation. It’s slop. Like grabbing muddy, undried bricks and immediately trying to build a house. You’re not innovating. You’re making a mess.

If you actually want LLM-powered design agents that help, treat the design system as infrastructure.

Audit it. Properly prioritize it. Resource it. Close the gaps.

Otherwise, let’s be real. You’re not shipping. You’re shitting! 💩 All in a rush to say that you "delivered an AI." 🥴 Just compounding debt, wasting effort, and handing teams more low quality solutions.

Does it need to be perfect? Nope! Got to start some where. But ignoring glaring holes tho? They need to be identified, prioritized, and fixed before you start pretending to have an AI strategy.