Let's walk through how you can set up your AI agent to let users interrupt it, while making sure it doesn’t interrupt itself. Here’s a simple guide to help you: 1. Enable User Interruptions: Just set

allow_interruptions=True

in your

AgentSession

configuration. This way, users can jump in and interrupt the agent while it’s speaking. 2. Per-Turn Interruption Control: You can manage interruptions for each speech segment with the

SpeechHandle

object. Each piece of generated speech has an

allow_interruptions

flag, which is set to

True

by default. 3. Prevent Self-Interruption: The agent is smart enough to avoid interrupting itself. While it’s generating or playing back speech, it checks the

SpeechHandle

to make sure only interruptions from users are accepted. 4. Realtime LLM Models: If you’re working with a realtime LLM model that has built-in turn detection, just double-check your model settings to match the interruption behavior you want, since the

allow_interruptions

parameter might not apply there. Good news: by default, the agent runtime already takes care of blocking self-interruptions, so you don’t need to tweak anything extra for that. If you have any questions or need more help, feel free to ask—I’m here to help! Sources: Turn detection and interruptions | LiveKit Docs | tests/test_agent_session.py