<ul data-eligibleForWebStory="true">Databricks introduced Agent Bricks at the Data + AI Summit for building and deploying AI agents using enterprise data.Agent Bricks automates AI agent development by transforming task descriptions and data into deployable agents.The system uses synthetic data and benchmarks to optimize agent performance based on accuracy and cost tradeoffs.Use cases include Information Extraction Agent, Knowledge Assistant Agent, Multi-Agent Supervisor, and Custom LLM Agent.Agent Bricks addresses challenges in AI development by providing domain-specific evaluations and a streamlined workflow.AstraZeneca and Flo Health are early adopters seeing positive results with Agent Bricks in data extraction and accuracy improvements.Databricks also launched serverless GPU support and MLflow 3.0 to enhance AI system management.MLflow 3.0 includes features for generative AI like prompt management and integration with data lakehouses.The combination of Agent Bricks, serverless GPUs, and MLflow 3.0 positions Databricks as a comprehensive platform for enterprise AI systems.