Open beta. Self-host is MIT. Report a bug on GitHub. GitHub
FluxyChat
Start free
All rooms

ModuleVoiceAttach when you need it

Voice AI

joinVoiceStage is signaling. Live audio is LiveKit when configured; recorded clips land as room messages.

Open the consoleSee the kernel

What runs

  • Stage roster and signaling on the room
  • Live media through LiveKit
  • Voice clips back on the room timeline

Not forNot an SFU we sell. Huddles media stays labs; you bring LiveKit and its meter.

Where it sits

It shares the room socket, history, and agent stream with chat core. Turning it on does not add a second realtime stack or a second place to look for a message.

The SDK contract is the room. Start with chat core, attach the surface when the flow needs it, and keep the transcript in one place.