LiveKit Agents

LiveKit Agents is a framework for developers building voice assistants, phone agents and applications that combine speech, video and text. Python and Node.js agents interact with users through LiveKit rooms connected to web, mobile or telephone clients.

• Voice conversations: The framework handles streaming speech recognition, language model responses and speech synthesis, with turn detection and interruption handling.
• Model choice: You can combine speech and language model providers or use realtime model APIs through provider integrations.
• Tools and handoffs: Agents can call tools, use MCP servers, pass conversations to other agents and exchange data with a frontend.
• Phone calls: LiveKit's SIP integration lets agents receive incoming calls and place outgoing calls.
• Deployment and testing: Built-in scheduling, load balancing and Kubernetes compatibility support server deployments, while automated tests and judges help check agent behavior.

The framework and LiveKit server are open source under Apache 2.0, and you can host the entire stack on your own servers. Where model processing happens depends on the providers you connect.

LiveKit Cloud separately offers managed agent deployment, transcripts and traces, and model access through LiveKit Inference. Its browser-based Agent Builder provides a hosted way to prototype and deploy agents without writing code.