LiveKit Agents is an open source framework designed for building realtime AI agents that can participate as programmable entities within communication sessions. It enables developers to create conversational and multimodal agents capable of processing voice, audio, and other inputs in realtime environments. These agents can join LiveKit rooms as participants and interact with users or systems through speech, text, and other modalities. LiveKit Agents provides libraries and tooling that allow developers to combine speech-to-text, large language models, and text-to-speech services to build interactive AI experiences. It is designed to run server-side and can integrate with various AI model providers and realtime APIs to support different application requirements. LiveKit Agents also includes tools for scheduling and managing agent tasks, making it easier to connect users to automated assistants in live communication scenarios.
Features
- Framework for building realtime conversational and multimodal AI agents
- Integration with speech-to-text, language models, and text-to-speech systems
- Programmable agents that can join realtime communication rooms
- Built-in job scheduling and task distribution for agent workloads
- Extensive plugin ecosystem for integrating multiple AI providers
- Support for WebRTC-based client applications across major platforms