Skip to Main Content

Scene

Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Scene — Demo video

Scene is a group chat with a new layout. You make yourself once: a name, your own voice (read three sentences; Muse Voice Transcribe confirms it is you), and a clay figurine that Muse Image sculpts from three words. Your figure stands on a turntable on your home screen, and you can walk the camera all the way round it. Below it sit your chats and friends. Add a friend by code, they accept, and a room for the two of you opens behind a pair of clay double doors. Inside, there is no list of bubbles. You sit at a round table in a miniature stop-motion dining room, and your friends sit around it as clay figures. When a message arrives, its sender's figure says it in that friend's own voice and turns to whoever it is for, while everyone else reacts. Missed the chat? Press play and Muse Spark directs the whole table: who looks at whom, who laughs, who wakes up when their name comes up. Then Spark sums up what you missed, quoting only what was really sent. Every voiced line is first heard by Muse Voice Transcribe, and it plays only if it says exactly the message. It says what you said, nothing else. Rooms are yours to dress: pick a set, or say one ("a Seoul street food tent on a rainy night") and Spark plans it while Muse Image paints the walls, floor and tabletop for everyone at the table. The demo chat that every new user gets is a clearly marked script. Real chats are real: friends, groups and live messages, no login. Live: https://scene-lovat.vercel.app

Team