Learn to build voice AI
that actually works.
Project-based courses from architecture to production. Built on the only open-source platform where AI, humans, and phones share one room.
The only platform where AI, humans, and phones share one room.
WebRTC-native
Sub-100ms latency for voice AI. Not WebSocket buffering — real UDP transport with adaptive bitrate.
Room architecture
Humans, AI agents, and SIP callers in one session. Same primitives, same SDKs, infinite combinations.
Fully open source
Self-host everything. Apache 2.0 licensed. Same APIs whether you use Cloud or your own infrastructure.
12+ SDKs
Web, iOS, Android, Flutter, React Native, Unity, C++, Rust, Python, Node.js, Go, ESP32.
From first principles to production deployment
Voice AI Foundations
Build a dental clinic AI receptionist from scratch
Build a dental clinic receptionist that greets callers, answers FAQs, checks appointment availability via tool calls, books appointments, and handles edge cases. Deployed to LiveKit Cloud and testable from Playground.
Start this course→Chapters
- 01What is voice AI? Why WebRTC matters
- 02Project setup & your first agent
- 03Understanding AgentSession
- 04Writing great agent instructions
- 05Adding your first tool: check availability
- 06RunContext, state & the booking tool
- 07Noise cancellation & room configuration
- 08Testing your agent
- 09Turn detection deep dive
- 10Conversation context & multi-turn state
- 11Beyond audio: the data plane
- 12Deploy to LiveKit Cloud
- 13Polish & production readiness