KITT by LiveKit is a demo web app that enables real-time audio and video conversations with ChatGPT over WebRTC. It shows how a text-based model can be turned into a live voice and video assistant with low-latency interaction.
Real-time AI conversations
KITT combines LiveKit Agents with ChatGPT and voice components to support natural dialogue in a call-like experience.
- Speech recognition for capturing user input
- AI response generation with ChatGPT
- Text-to-speech voice output
- Audio/video delivery via WebRTC
Built for developers and prototyping
The project is available as an open example: you can run the demo, then review and reuse the source code in your own apps. It’s a practical starting point for products that need live AI dialogue.
- Voice assistants
- Interactive podcasts and live shows
- Training and tutoring experiences
- Any app that needs real-time spoken interaction
Foundation for custom solutions
KITT demonstrates how to connect LiveKit’s WebRTC infrastructure with common AI services, helping teams prototype faster without building real-time plumbing from scratch.
