LiveKit
Unicorn Β· $1BSan Jose, US Β· 5 known investors
LiveKit is a developer platform for building, deploying, and monitoring voice, video, and physical AI agents using open source SDKs, with cloud infrastructure that handles runtime, routing, autoscaling, and realtime observability. It targets developers and enterprise teams building low-latency, real-time conversational agents.
Also known as LiveKit, Inc.
Investors Β· 5
Funding
SEC filings, press & company announcements$100M disclosed across 1 of 4 rounds Β· 2025β2026
- $100MSeries CJan 2026 Β· 5 sources
Index Ventures (lead), Altimeter, Hanabi Capital, Redpoint Ventures, Salesforce Ventures
Source β - Undisclosed amountSeries BApr 2025 Β· 2 sourcesSource β
Source: company announcements and press reports β follow each round's link for the claim.
Valuation Β· disclosed
Disclosed eventsSource: SEC prospectus filings, and round valuations the company or its investors disclosed β follow each entry's link for the claim.
Company profile
researched Aug 2026LiveKit develops an open source realtime stack and a managed cloud platform for applications that connect humans and AI over voice, video and data. The core open source project is a scalable, distributed WebRTC SFU (Selective Forwarding Unit) written in Go on top of the Pion WebRTC implementation, deployable as a single binary, via Docker or Kubernetes, and licensed under Apache 2.0. It supports JWT authentication, UDP/TCP/TURN networking, simulcast, selective subscription, speaker detection, moderation APIs, end-to-end encryption, SVC codecs (VP9, AV1), webhooks and multi-region operation. Surrounding the media server are ecosystem components for Agents (programmable backend participants for multimodal AI), Egress (recording, multi-streaming and track export), Ingress (RTMP, WHIP, HLS, OBS) and SIP.
The commercial product, LiveKit Cloud, packages this stack into an end-to-end developer platform spanning the agent development lifecycle: an open source framework to build and test agents, LiveKit Inference as a gateway that routes requests to TTS, LLM and STT model providers (and increasingly to models LiveKit hosts in its own data centers), serverless deployment and scaling of agents, phone numbers and SIP integrations for telephony, and full-stack observability of agent sessions. Additional tooling includes Agent Builder for template- and prompt-based agent creation in the browser, unit testing of agent code, OpenTelemetry trace export, and partnerships with Bluejay, Hamming and Roark for conversation simulation testing.
Developer surface area is broad: agent SDKs in Python and Node.js; client SDKs for browser/JavaScript, Swift, Android/Kotlin, Flutter, React Native, Rust, Unity, Unity WebGL, ESP32 and C++; server SDKs for Go, Node.js, Python, Ruby, Java/Kotlin, plus community PHP and .NET; UI component libraries for React, Android Compose, SwiftUI and Flutter; a CLI and load tester; and starter apps. Documented use cases include customer service and call-center agents, medical triage, telephony calling systems, RAG chatbots, vision agents and robotics video streaming.
Founding story
According to TechCrunch, LiveKit was founded in 2021 by Russ d'Sa and David Zhao as an open source software project for building applications that transmit realtime audio and video without interruptions, at a time when the world was meeting over Zoom during the pandemic. The project was initially a free developer tool; the business took off when the founders recognized demand from large companies for a managed cloud version.
Business model
Dual model combining a permissively licensed (Apache 2.0) open source media server and agents framework with LiveKit Cloud, a hosted managed service. The open source project can be self-hosted; LiveKit Cloud provides globally distributed realtime infrastructure, serverless agent deployment, an inference gateway, telephony and observability as a paid service, with free monthly bandwidth and transcoding credits for each project. The company states the business accelerated after the founders found that large companies wanted a managed cloud version of the open source tool.
Consumption of LiveKit Cloud infrastructure (realtime bandwidth/transcoding, agent hosting, inference routing, telephony/SIP), sold to developers and enterprises; each Cloud project includes free monthly bandwidth and transcoding credits before paid usage.
Traction
The site cites more than 1,000 developers, over 2.5 billion calls annually, and enterprise compliance attestations (GDPR, SOC 2 Type 2, HIPAA). The livekit/livekit GitHub repository has about 20,500 stars and 2,300 forks; the livekit/agents repository is shown with 13.1K stars and the Agents framework is described as downloaded more than 1 million times a month. LiveKit Cloud lists 25 regions globally.
Latest developments
In January 2026 LiveKit announced a $100M Series C led by Index Ventures at a $1 billion valuation, with Salesforce Ventures, Hanabi Capital, Altimeter and Redpoint Ventures participating. Recent product work described in that announcement includes serverless agents for turnkey deployment, LiveKit Inference for cross-provider model routing plus LiveKit-hosted models, Agent Builder for no-code/low-code agent creation, unit testing of agent code with OpenTelemetry trace export, simulation testing partnerships with Bluejay, Hamming and Roark, and partnerships with telephony carriers connecting LiveKit's network directly to the PSTN. Community announcements in August 2026 include PII redaction for LiveKit Observability and an Expressive Mode for voice agents.
βΈFull profile β market position, technology, go-to-market, geography, history, risks & controversies
Market position
Positioned as realtime infrastructure for the voice AI stack, contrasted with HTTP-based web infrastructure that the company argues is unsuited to stateful, long-running realtime agent sessions. It powers ChatGPT's Advanced Voice for OpenAI and reports enterprise and startup adoption across financial services, healthcare, retail, customer support, education and robotics. Coverage describes it as an OpenAI partner that reached a $1 billion valuation in January 2026.
Combines an Apache 2.0 open source realtime media server and agents framework with a purpose-built global realtime network, rather than general-purpose web infrastructure. It offers an end-to-end lifecycle in one stack (build, test/evaluate, deploy, observe), model-agnostic access to hundreds of AI model integrations through an inference gateway that can route around provider latency and outages, direct carrier interconnects to the PSTN for low-latency phone calls, and broad multi-platform SDK coverage. Customers cite code-level control over agent behavior and the ability to use any model provider.
Technology
Go-based, horizontally scalable WebRTC SFU using Pion, with JWT-authenticated access tokens, UDP/TCP/TURN connectivity, simulcast, selective subscription, end-to-end encryption, SVC codecs (VP9, AV1) and webhooks. On top of this sits the LiveKit Agents framework, which orchestrates STT, turn/interruption detection, LLM and TTS models per conversational turn and handles conversational dynamics such as multilingual turn detection and interruption handling. LiveKit operates a global network of data centers configured as a unified fabric optimized for routing realtime voice and video, with direct interconnects to telephony carriers for PSTN traffic, and LiveKit Inference to route requests across model providers and LiveKit-hosted models to mitigate provider latency and outages.
Go-to-market
Developer-led adoption of the open source project and SDKs (GitHub, docs, CLI, quickstarts, starter apps, Slack and a public community forum), converting to LiveKit Cloud for managed deployment and scale; enterprise sales for large-scale production deployments; educational and ecosystem partnerships, including a course on building AI voice agents for production co-created with LiveKit and taught in part by CEO Russ d'Sa, and integrations with model, telephony and evaluation vendors.
Developers and engineering teams building voice, video and physical AI (robotics) applications; startups and large enterprises. Named customers and references include OpenAI (ChatGPT Advanced Voice), xAI, Salesforce (Agentforce), Tesla, 911 emergency service operators, mental health providers, Podium, Retell AI, Playback, Assort Health, Spara and Polymath.
Geography
Headquartered in San Jose, US, with LiveKit Cloud infrastructure across 25 regions including US West, US Central, US East, Brazil, UK, France, Germany, Switzerland, Israel, Saudi Arabia, UAE, South Africa, India, Singapore, Japan and Australia.
History
Founded in 2021 by Russ d'Sa and David Zhao as an open source project for transmitting realtime audio and video without interruption, during the pandemic-era shift to video meetings. The free developer tool later gained commercial traction as enterprises sought a managed cloud version amid the voice AI boom. The company worked on ChatGPT Voice Mode, which informed the design of the LiveKit Agents framework, raised a Series B in April 2025, and announced a $100M Series C at a $1 billion valuation in January 2026, alongside launches including serverless agents, LiveKit Inference and Agent Builder.
Risks & controversies
Public community threads in August 2026 report operational issues including LiveKit Inference timeouts over multiple weeks, TTS timeouts, API keys rejected with 401 errors, quota/runaway-session problems and 'zombie' sessions, and end-to-end latency complaints; at least one user requested account deletion after being unable to get a hardware SIP phone working. The business depends on third-party model and telephony providers for parts of the pipeline, and the permissive Apache 2.0 license permits self-hosting as an alternative to the paid cloud.
Compiled by commissioned research from 8 cited public sources β announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed β not independently audited.
Founder mafia
2 people who came through LiveKit went on to found or lead other companies.
Timeline Β· 9
launches, deals, and filingsAnnounced in the LiveKit community forum announcements section.
Community announcement titled 'Making Voice Agents sound human with Expressive Mode'.
Index Ventures led a $100M round at a $1 billion valuation, joined by Salesforce Ventures, Hanabi Capital, Altimeter and Redpoint Ventures. TechCrunch notes the round came 10 months after the previous fundraise.
$100M source β
LiveKit partnered with telephony carriers worldwide to link its network directly to the PSTN for lower-latency phone conversations with voice agents.
Partnerships to support simulation-based evaluation in which a voice agent calls another voice agent across thousands of permuted conversations.
Browser-based tool to start agents from a template, tune prompts, sketch workflows and share links, described as recently launched in the January 2026 Series C post.
Gateway that abstracts model orchestration by routing inference between TTS/LLM/STT providers and LiveKit-hosted models colocated with agents.
LiveKit announced its Series B in April 2025, per the company's Series C blog post.
Turnkey agent deployment released 'earlier this year' per the January 2026 Series C announcement.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
In the news
βΈResearch sources Β· 8
primary sources listed
- LiveKitlivekit.io Β· web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does LiveKit do?
- Open source realtime framework and cloud platform for building, deploying and scaling voice, video and physical AI agents.
- Who are LiveKit's investors?
- LiveKit's investors include Hanabi Capital, Index Ventures, Redpoint Ventures, Altimeter Capital, Salesforce Ventures.
- How much funding has LiveKit raised?
- LiveKit has disclosed $100M raised across 1 of its 4 known rounds.
- Where is LiveKit headquartered?
- LiveKit is headquartered in San Jose, US.