LiveKit Agents
Conversational AI FrameworksOpen-source framework for building real-time voice AI agents with WebRTC, STT, LLM, and TTS pipelines.
Run an HVAC, plumbing or other home-services company?
We'll set up LiveKit Agents, or the right alternative, to answer your calls and book jobs. Start with a free audit.
👍 Pros
- ✓Fully open-source and self-hostable
- ✓Strong real-time media layer
- ✓No vendor lock-in
- ✓Active and growing community
👎 Cons
- ✗Requires infrastructure expertise to self-host
- ✗Less opinionated — more setup work
- ✗Fewer managed deployment options
Features
- ✓ Open-source (Apache 2.0)
- ✓ WebRTC native
- ✓ Composable STT/LLM/TTS pipeline
- ✓ Python & JS SDKs
- ✓ Deepgram & ElevenLabs support
- ✓ Self-hostable
- ✓ Active community
From the blog
View all →15 May 2026 · 19 reads
ElevenLabs, OpenAI TTS, and Azure: Which Voice Is Right for Your Agent?
The TTS provider you choose shapes how users perceive your voice agent. Here's a practical comparison of the three leading options for production deployments.
Read more →12 May 2026 · 11 reads
How to Evaluate a Voice AI Agent: 5 Metrics That Actually Matter
Not all voice agents are created equal. These five metrics will tell you whether an agent is ready for production — or still needs work.
Read more →9 May 2026 · 6 reads
The Voice Stack Explained: STT, LLM, and TTS
Every voice AI agent is built on a three-layer stack. Understanding each layer — and the tradeoffs between providers — is essential for building reliable, natural-sounding agents.
Read more →