lakebase-setup

To configure persistent memory for Databricks LangGraph agents, provision Lakebase and attach it as a resource in databricks.yml to enable short-term chat history and long-term user facts across sessions and deployments.

Setting up short-term conversation history and long-term user facts involves provisioning Lakebase, initializing the memory tables, and configuring environment variables for runtime access within your Databricks LangGraph app.

Yes, you need to grant service principal permissions for Lakebase memory storage. The setup process includes performing required permission grants via databricks.yml and related scripts to secure access to the initialized Lakebase tables.

Yes, you can use Lakebase with an autoscaling Databricks setup. The configuration provides specific guidance for both provisioned and autoscaling Lakebase setups, including memory templates and deployment steps.

To enable agent persistence in Databricks, the required dependencies are databricks-langchain with memory features and databricks-ai-bridge, which ensure proper environment configuration and runtime access to Lakebase storage.