I'd love to help you add memory for personalization and collect user preferences in LiveKit Agents! Here are some friendly approaches you can try: • In-Session Memory: You can use

session.userdata

(Python) or

ctx.userData

(JavaScript/TypeScript) to keep track of user preferences during a session. Just a heads-up, this memory is only for the session and won’t stick around once it ends. • Participant Metadata: Another great option is to use

participant.attributes

to tailor interactions based on things like the user’s language or location. It’s a nice way to make conversations feel more personal! • External Memory Services: If you’re looking for longer-term memory or more advanced options, you might want to connect with external services like Letta, LlamaIndex, or Mem0. These tools help your agents remember user preferences across sessions and provide even more personalized responses using past data. If you’d like, I’m here to walk you through any of these methods with detailed examples or help you decide what fits best for your needs—whether that’s memory just for the session or something that lasts across sessions. Just let me know what you’re aiming for! Sources: flows/simple_flow.py | README.md | README.md | Inside a session | LiveKit Docs | Inside a session | LiveKit Docs