LiveKit Documentation

The platform for voice, video, and physical AI agents.

Get started

Start building with LiveKit.

Agent Builder

Create your first agent right in your browser.

session = AgentSession(
   stt="deepgram/nova-3",
   llm="openai/gpt-5.3-chat-latest",
   tts="cartesia/sonic-3",
   turn_handling=TurnHandlingOptions(
      turn_detection=MultilingualModel(),
   ),
)

Voice AI quickstart

Build a voice agent in code with this step-by-step guide.

❯ create a friendly voice assistant
● Searched LiveKit docs
● Read voice-ai-quickstart guide
✎ Wrote agent.py

Coding agent support

Turn your favorite coding agent into a LiveKit expert.

Understanding LiveKit

Dive into concepts and overview of how LiveKit works.

LiveKit reference

Reference docs for LiveKit SDKs and APIs.

End-to-end platform

Tooling and guides no matter where you are in the development lifecycle.

Build

Realtime framework for production-grade multimodal and voice AI agents.

Integrate

Connect your agent to models, systems, and third-party services.

Deploy

Deploy and scale your agent in production.

Observe

Monitor your agent's behavior for quality and reliability.

Common use-cases

Explore common agent use-case examples and recipes.

See all recipes

Customer service agents

Try out typical agent workflows like call center and medical triage assistants.

Telephony implementations

Build agent-powered calling systems.

RAG integrations

Create chatbots with access to knowledge bases.

Vision agents

Create AI assistants that process video input.

SDKs and tools

Explore the full suite of LiveKit SDKs and tools.

See all SDKs