Develop agents on Databricks

Agent Bricks is the Databricks agent developer platform. Use it to build and deploy the agents that power your agentic products and workflows.

Agent Bricks supports the core workflows for building a production agent:

• Deploy your agent: Agent Runtime hosts agents built with any framework or harness, stateful or stateless, and AgentKit adds durable execution so that runs survive restarts and crashes.
• Run code safely: Agents run the code they write in Databricks Sandbox, an isolated environment whose home directory persists across sessions.
• Enrich your agent's context: Agent memory and sessions let your agent recall what it learned in earlier conversations, so it can personalize responses and reduce token costs.
• Debug and test your agent: MLflow Tracing records each step that your agent takes, and you can store and govern traces at scale in Unity Catalog.

The Agent Bricks CLI scaffolds agent code that connects to these services. You can iterate locally, including with coding agents, and then deploy your agent.

Get started

Learn what Agent Bricks provides and build your first agent.

• • Agent Bricks quickstart
  • Build and deploy your first agent with the Agent Bricks CLI.
• • What is Agent Bricks?
  • Learn the components of an agent and how Agent Bricks supports each one.
• • Agent Bricks CLI
  • Create, run, and deploy agents from the command line, or bring an existing agent.

Connect to models

• • Connect to models
  • Call frontier and open models through Unity Gateway with one API, and switch models without changing agent code.

Add agent context

Enrich your agent's context with memory, tools, and your organization's data.

• • Agent memory and sessions
  • Store conversation state and long-term memory in managed stores backed by Lakebase.
• • MCP servers and agent tools
  • Connect agents to Databricks-managed, custom, and external MCP servers and tools.
• • Genie One MCP server
  • Enrich agents with your organization's combined business context through natural-language questions over governed data.
• • AI Search
  • Retrieve relevant text and unstructured data from a managed AI Search index.

Deploy your agent

Run your agent with an agent server on managed compute.

• • Deploy agents on Databricks
  • Learn the agent compute stack: framework or harness, agent server, and agent runtime.
• • Agent Server
  • Serve your agent with DurableAgentServer: synchronous, streaming, and background runs, with crash recovery.
• • Agent Runtime
  • Deploy and manage agents on Databricks-hosted compute.
• • Databricks Sandbox
  • Give agents an isolated environment to run the code they write, with scoped access to governed data.

Query your agent

• • Query agents deployed on Databricks
  • Send requests to deployed agents through the invocation API, the Responses API, or SQL.

Agent observability and quality

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

• • Agent observability and quality
  • Learn how MLflow traces, evaluates, and monitors agents.
• • MLflow Tracing
  • Record each step that an agent takes to debug and improve it.
• • Evaluate and improve
  • Measure agent quality with LLM judges, custom scorers, and expert feedback.

Govern your agents

• • Unity Gateway
  • Govern access to models, MCP servers, and skills, with guardrails, rate limits, and usage tracking.

Transform unstructured data

• • AI Functions
  • Apply AI models to your data from SQL to classify, extract, summarize, and parse documents.