What is the best way to give your agent long term memory?

To give your agent long-term memory, you can store conversation history and context using events like

user_speech_committed

and

agent_speech_committed

. These events allow you to capture and save the conversation data incrementally, which can then be stored in a database for future retrieval. This approach ensures that when a user returns, the agent can access past interactions to provide contextually relevant responses. For implementation, you can refer to the LiveKit documentation which provides examples of how to handle these events. Additionally, you can use a database like Firebase to store the transcripts as they are committed, as suggested in this Slack discussion. Would you like more details on setting up the database or handling these events? Sources: - this Slack discussion 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.