LiveKit Agents
Python framework for realtime voice and video AI agents over WebRTC
What it is
A server-side framework for building conversational agents that join LiveKit rooms, listen, speak, and can see video. It lets you combine speech-to-text, LLM, text-to-speech, or realtime model providers through plugins, and adds job dispatch, semantic turn detection, multi-agent handoffs, tool calling, MCP tools, and phone calls via SIP. Developers building voice assistants, call-center bots, or interactive companions use it with the LiveKit media server, which can run on their own infrastructure.