Yui is a mobile hub for your AI agents.
You talk to your agent. When a button, a timer or a table would work better than text, the agent takes over the screen and generates that UI on the spot, inside one shared design system.
What it does
Chat first
The default screen is a conversation with your agent. Type it or say it.
The agent takes the screen
When words are not enough, the agent builds the UI: a timer, a form, a multi-select, a table, a dashboard.
Generative, not templated
Two people should get wildly different screens. A shared component kit and UX rules keep it clean.
Bring your own agent
Hermes and OpenClaw users plug in first. Starter agents come later for people with nothing set up.
Any channel
Talk on Telegram, SMS or voice. Say "pull this up on Yui" and a push notification deep-links to the screen.
A face per agent
Each agent has its own colors, avatar and voice. Arnold looks and sounds like Arnold.
The example that started it
Mid-workout, Chris tells Arnold: "I want intervals of 40 on, 20 off, eight rounds." Arnold does not reply with a paragraph. The screen becomes an interval timer in Arnold's colors, with big start and pause buttons. Asked a yes or no question, the app shows two buttons. Asked to pick, it shows a multi-select with a free-text escape hatch.
In Chris's words
From pitch recording 366, Sep 23 2026. Working name in the recording: Project Nexus.