Build AI agents on Databricks

Databricks supports building, evaluating, and deploying AI agents, from simple LLM calls to tool-calling agents and multi-agent systems. These guides cover the concepts, development workflows, and tools you use to ship an agent.

Looking for classic ML or deep learning? See Machine learning on Databricks.

Get started

Try a quickstart or learn the foundational concepts.

• • AI Playground
  • Prototype and test agents and LLMs with no-code prompt engineering and parameter tuning.
• • Get started with AI agents
  • Build and deploy your first AI agent end-to-end.
• • Concepts: Generative AI on Databricks
  • Learn about models, agents, tools, and apps.
• • Agent development lifecycle
  • Understand the full lifecycle of building an AI agent.

Build and deploy

Develop and deploy agents.

• • Knowledge Assistant
  • Build and optimize domain-specific QA agent chatbots.
• • Supervisor Agent
  • Build a supervisor agent that orchestrates Genie Spaces, agent endpoints, Unity Catalog functions, MCP servers, and custom agents.
• • Custom Agents
  • Build and deploy agents, including RAG applications and multi-agent systems, with Python.
• • Databricks Apps
  • Build and deploy interactive UIs for your agents, such as chat apps and data entry forms.
• • MCP servers
  • Connect agents to tools, data, and workflows through standardized MCP servers.
• • Vector Search
  • Query a managed vector index to retrieve relevant text and unstructured data.

Evaluate and monitor

Trace, evaluate, and monitor agents in development and production.

• • Evaluation and monitoring
  • Evaluate agent quality and monitor production deployments.
• • MLflow Tracing
  • Record and analyze agent behavior to debug and improve performance.

Query and serve

Query LLMs and serve agents and models on scalable endpoints.

• • Query LLMs and agents on Databricks
  • Query LLMs and agents from notebooks, SQL, and applications.
• • Foundation Models
  • Serve LLMs through scalable APIs with built-in governance and monitoring.
• • Unity AI Gateway
  • Govern and monitor access to LLMs and agents with usage tracking, payload logging, and security controls.
• • AI Functions
  • Call LLMs directly from SQL to enrich data and build AI workflows.