[Next-Generation Apps May Have No Fixed Interfaces, Jev Enables Real-Time UI Reorganization] Beating AI Newsflash: Vercel Labs engineer Chris Tate has integrated Jev into json-render, experimenting with AI directly assembling UI during runtime. json-render prepares buttons, cards, input fields, data, and actions in advance. Jev doesn’t need to write complete JSON token by token like large models; it simply decides which components to use, how to arrange them, and where to place them, then hands it off to the standard UI renderer. The new interface defaults to selecting primary components in one pass and uses a second pass for layout arrangement when needed. In a demo of generating a train ticket interface, the default JSONL mode took 3.21 seconds from request to completion, while Jev only took 0.88 seconds. In the future, app interfaces may not need to be fully hardcoded. UI could become runtime-driven, dynamically assembling different pages for different users, tasks, and states. Jev’s demonstration has already reduced the total generation time to under 1 second, making this type of interaction feel at least somewhat real-time. Runway’s Solaris, released late last month, is betting on a similar direction but with a more radical approach. Solaris uses a world model to generate entire interfaces frame by frame, bypassing traditional code and components entirely. Jev, on the other hand, still relies on developer-prepared components and actions, dynamically deciding how to combine them. One aims to turn software into a real-time generated video world, while the other is more like giving software a face that can be reorganized at any moment. [Original Link]