Using OpenAI’s Assistants API with LiveKit in Real-Time — Any Best Practices?

Hey Devs 👋

We’re building a voice-based real-time app using LiveKit for audio/video and want to connect it to the OpenAI Assistants API (with Threads) to keep the conversation context without manually storing messages in Redis or a database.

Here’s what we’re aiming for:

Real-time voice via LiveKit

Whisper-based transcription of user speech

Send transcribed messages to an Assistant Thread

Receive replies (text or TTS) and stream them back

Maintain full session memory using Threads

This used to be manual work with chat completion + Redis. Threads seem perfect now — just not sure how to glue them together in a real-time app.

❓ Has anyone tried this? Are there any repos, patterns, or official docs that show LiveKit + Assistants API integration?

Any help is appreciated! 🚀