Engineering brief
Your Design System Is Now an API Contract for AI
At a glance
- Relevance
- Practical value
- Warnings
- None
Gus Iwanaga's team built a product where AI generates the entire UI from a query. The breakthrough isn't the generation—it's the declarative protocol that enforces design system compliance.
Dynamic, AI-generated UIs will reshape how teams design, ship, and maintain product experiences.
Summary
Static UIs force users to adapt to software. Gus Iwanaga argues this is backward—AI should compose interfaces around user intent, not the other way around. His team built a product where an orchestrator extracts intent, invokes tools, and delegates UI rendering to a "UX agent" that assembles components from a catalog. The result is
a dynamically generated, personalized interface, not a fixed screen. Three UI protocols exist: controlled (opinionated components), open-ended (full LLM autonomy), and declarative (a middle ground). His team chose declarative because it enforces design system compliance while still allowing AI-driven composition. The tradeoff: less autonomy for the LLM means more upfront curation of the component
catalog and layout rules. Key challenges include information architecture—how components are arranged—and the shift in team workflow. Teams must codify UX principles into the agent, maintain a precise component catalog as the contract between agent and UI, and teach designers and PMs to work with schemas and synthetic data instead of pixel-perfect mockups.
Watch the video
This video is blocked due to your privacy settings. To watch this video, please accept YouTube marketing cookies.
Related breakdowns
Why AI Can’t One-Shot Design and What That Means for Your Team
AI can’t one-shot good design. Paul Bakaus explains why adjectives and verbs are the missing control layer for steering AI design output—and what that means…
Why design taste is the new competitive moat for AI apps
The secret to making AI apps stand out isn't a better model — it's giving your agent high-quality design references and avoiding the telltale signs of AI slop.
Desktop OS Is Stuck in 2005 — Here’s What Needs to Change
Desktop UX hasn't evolved in 20 years. Scott Jenson proposes local-first, context-aware OS features that could reshape how teams build and use productivity…
Get TL;DW
Too Long; Didn't Watch.
A concise breakdowns of the AI and devtools videos that actually matter for engineering leaders.
Free. Weekly. No hype.
Video and thumbnails remain the property of their respective creators. tldw.news provides editorial analysis, commentary, and discovery links to original content.