genai-agents-setup

To deploy GenAI agents on Databricks, use MLflow 3.0 ResponsesAgent for orchestration, Genie Spaces for data access, and Lakebase for memory. This ensures agents are scalable, continuously evaluated, and monitored for performance.

Building a customer support AI agent requires orchestrating Databricks services. Use Genie Spaces for user data access and Lakebase memory to maintain conversation history, ensuring the agent remains production-ready and continuously monitored.

MLflow 3.0 works directly with Databricks Genie Spaces. The ResponsesAgent API uses MLflow for end-to-end agent orchestration while Genie Spaces handles the underlying data access for comprehensive agent development.

Managing dependencies for GenAI agents using MLflow involves configuring mandatory core agent skills and cross-domain dependencies. This structured approach ensures comprehensive development, evaluation, deployment, and monitoring within the Databricks environment.

Lakebase memory is used to maintain conversation state and history in Databricks GenAI agents. It integrates directly with the MLflow 3.0 ResponsesAgent to ensure agents retain context during continuous production evaluation.