Once UI 2.0 - Builds consistent React apps for developers and AI agents

Once UI is an open-source design system for building polished React products with less repetitive code. Compose layouts with readable primitives, use a shared system of components and design tokens, and keep the result consistent as your product grows. Version 2.0 gives both you and your coding agent a clearer way to work. Its component catalog, compact rules, and task guides help agents use the system correctly; its foundations and predictable APIs make the output easier for you to understand.

Add a comment

Replies

Best
Hey Product Hunt, I’m Lorant, the creator of Once UI. I started Once UI because I wanted independent builders to have the kind of design infrastructure larger teams can afford to maintain. Over time, it became clear that this matters even more when an agent is helping write the code. Generating a screen is easy. Keeping the next twenty screens coherent is the real challenge. So for 2.0, we worked on the system underneath: clearer foundations, more consistent components and APIs, and a compact set of rules and examples that coding agents can actually use. You can build with an agent, write everything yourself, or move between the two while staying in control of the result. The core is open source. I’d love to see what you build with it, and I’d especially value honest feedback from anyone who has tried using a design system across a real product. Where does the system help, and where does it get in your way?
Which Vercel product did you use? What new capability did it give your team, and how did that shape the product you launched?
We used Vercel’s deployment platform, especially Preview Deployments. Every change to Once UI 2.0 could be tested in a working site before release, across our documentation, examples, and the products we build with the system. That let us judge components in context instead of only in isolation. It shaped 2.0 into a design system tested through real product work: more consistent APIs, clearer foundations, and patterns we could trust beyond a demo.