# LiveKit

> Open source framework and cloud platform for building, deploying, and scaling voice, video, and physical AI agents in production.

LiveKit provides an open source framework and managed cloud platform for developers building voice, video, and physical AI agents. The project began as an open source WebRTC media server and has since evolved into a full-stack developer platform covering agent authoring, inference, telephony, observability, and global media transport. According to the company's About page, over 300,000 developers and teams use LiveKit as their infrastructure layer for real-time AI.

## What It Is

LiveKit is a developer platform in the real-time AI agent infrastructure category. Its core job is to give developers everything needed to build conversational voice agents, video AI applications, and robotics integrations — from the open source `livekit/agents` framework (which the GitHub repository shows at 10.9K stars) to a fully managed cloud that handles deployment, scaling, and observability. The platform sits between the AI models (LLMs, STT, TTS) and the end user's device, routing audio and video in real time with sub-250ms global latency.

## How the Stack Fits Together

LiveKit describes its offering as "the complete stack for Voice AI," composed of several distinct layers:

- **Open source agents framework** — Python and Node.js SDK for building and testing agents locally, with automatic turn detection and interruption handling
- **Inference gateway** — A single API key to access LLM, STT, and TTS models from providers including OpenAI, Google, Deepgram, Cartesia, ElevenLabs, xAI, AssemblyAI, and others
- **LiveKit Cloud** — Managed deployment, auto-scaling, cold-start prevention, instant rollback, and a global edge network spanning 18+ regions
- **Telephony** — US local and toll-free phone numbers, inbound SIP, and third-party SIP trunk support
- **Observability** — Session recordings, turn-by-turn transcripts, trace spans, and metrics export APIs

## Open Source Lineage

LiveKit started as an open source project for livestreaming and video conferencing using WebRTC. The core media server (`livekit/livekit`, 19.2K GitHub stars) and the agents framework (`livekit/agents`, 10.9K stars) are both published under open source licenses. The company's FAQ confirms that both can be self-hosted on custom infrastructure. LiveKit Cloud is the commercial managed layer built on top of this open source foundation, and the company also offers on-premise and private cloud deployments for enterprise customers.

## Audience and Use Cases

The platform targets developers building production-grade AI agents across three primary modalities:

- **Voice AI** — Conversational phone and web agents, customer service bots, healthcare intake, sales outreach
- **Video AI** — Real-time video processing and multimodal agent interactions
- **Robotics / Physical AI** — Low-latency video streaming for autonomous systems and drones

The About page notes customers range from AI and robotics labs to Fortune 500 companies. The company has published testimonials from engineering leaders at Podium, Retell AI, Playback, Assort Health, Spara, and Polymath, among others. The homepage states the platform powers "billions of calls" annually, and OpenAI is cited as having built ChatGPT's Advanced Voice on LiveKit Cloud.

## Enterprise and Compliance Posture

LiveKit Cloud carries GDPR, SOC 2 Type 2, and HIPAA compliance. The platform advertises 99.99% uptime across all plan tiers. Enterprise features include role-based access control, region pinning, SSO, shared Slack channels, designated solutions engineers, and support SLAs. Private cloud and on-premise deployments are available via the sales team. The company is backed by Index Ventures, Altimeter Capital, Redpoint Ventures, and Salesforce Ventures, among others, and has received a Series C round.

## Features
- Open source agents framework (Python and Node.js)
- Managed cloud deployment and auto-scaling
- Inference gateway for LLM, STT, and TTS models
- Automatic turn detection and interruption handling
- Telephony with US local and toll-free numbers
- SIP trunk integration
- Session recordings and observability
- Turn-by-turn transcripts and trace spans
- Cold-start prevention
- Instant rollback to previous agent deployments
- Background noise suppression
- Voice isolation
- Global edge network (18+ regions)
- WebRTC media transport
- Role-based access control
- Region pinning
- HIPAA compliance and SOC 2 Type 2
- SSO support
- Metrics export APIs
- CLI and dashboard
- Custom voices
- Self-hosting support

## Integrations
OpenAI, Google Gemini, xAI Grok, DeepSeek, Moonshot AI Kimi, Deepgram, AssemblyAI, Cartesia, ElevenLabs, Speechmatics, Inworld, Rime, Krisp, ai-coustics, Twilio, Railway, AWS S3

## Platforms
WEB, API, CLI, DEVELOPER_SDK

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://livekit.com
- Documentation: https://docs.livekit.io/
- Repository: https://github.com/livekit/agents
- EveryDev.ai: https://www.everydev.ai/tools/livekit