Here's a new open source UI tool I came across that launched yesterday that's worth keeping an eye on. It's called OpenUI and it’s designed to help product teams build more visual experiences inside AI tools.

OpenUI is a framework that lets the AI compose actual interactive UI (charts, forms, tables, dashboards) on the fly, in response to whatever the user asks. Instead of your AI Assistant saying "here's a summary of your Q3 metrics," it builds and renders a real bar chart. The interface adapts to the conversation, not the other way around.

Rather than having the LLM output JSON (which can be slow to parse), they designed a new compact language called OpenUI Lang specifically for this. The benchmarks show it's 67% more token-efficient than JSON - which at scale means meaningfully lower API costs and noticeably faster responses.

The company behind it, Thesys, has been running this as a commercial product for a while with 400+ teams, so it's not vaporware.

They open-sourced it this week as what they're calling an open standard for generative UI.

Here’s the link for more info with a playground to try it out: openui.com/playground