Skip to content

LiveKit Agents

Build multimodal voice assistants that can take calls

Visit official site

What LiveKit Agents is

LiveKit Agents is described as an end-to-end framework that lets developers build intelligent multimodal voice assistants interacting through voice, video and data channels. It combines the OpenAI Realtime API with LiveKit WebRTC infrastructure and documents a pipeline of STT, LLM and TTS models.

What you can do with it

  • Build AI voice assistants with STT, LLM and TTS models
  • Create speech-to-speech agents with the OpenAI Realtime API
  • Answer and respond to incoming calls through LiveKit agents and SIP libraries
  • Place calls on behalf of users

Application scenarios

  • A developer builds a voice assistant that answers customer service calls
  • An education institution builds a voice assistant that answers student questions live
  • A company builds an agent that records and summarises video meetings

Getting started

  • Read the LiveKit Agents documentation to learn how the framework works
  • Set up the OpenAI Realtime API and LiveKit WebRTC infrastructure from the guide
  • Select and integrate the STT, LLM and TTS models you need
  • Write code to create the AI voice assistant pipeline

Pros & cons

✓ What we like

  • Voice, video and data channels are all covered
  • Guidance for deploying and scaling agents is provided

! What to watch out for

  • The documented setup involves OpenAI Realtime API and LiveKit WebRTC
  • The assistant pipeline is assembled from chosen STT, LLM and TTS models

FAQ

What does the framework help build?

The material describes multimodal voice assistants that interact through voice, video and data channels.

Can agents handle phone calls?

The material lists answering and responding to incoming calls and placing calls on behalf of users.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review