livekit

Why we included this project

LiveKit is the transport layer many teams build real-time voice and video products on, and it has become a common backbone for conversational AI. The Go server is a production-grade WebRTC Selective Forwarding Unit that handles the low-latency audio and video plumbing, so developers can focus on application logic instead of networking, TURN relays, and multi-region scaling. It deploys as a single binary, in Docker, or on Kubernetes, and ships client SDKs for web, mobile, and desktop. For AI work, it pairs with the LiveKit Agents framework to connect LLM-driven voice agents to end users with minimal round-trip delay, which is why it keeps appearing in voice assistant and live-interaction stacks. If you are building anything that needs real-time human-to-AI or human-to-human media, this is the infrastructure worth evaluating.