databricks-agent-bricks

To build a RAG knowledge assistant on Databricks, you can use this Skill to create a Knowledge Assistant that generates document-based Q&A systems directly from files stored in Unity Catalog Volumes. It streamlines parsing company policies, manuals, or reports into a conversational interface.

The best way to orchestrate multiple AI agents in Databricks is by deploying a Supervisor Agent. It routes user queries to the most appropriate specialized agent, integrating Knowledge Assistants, Genie Spaces, custom model serving endpoints, and Unity Catalog Functions into a single unified chat interface.

Yes, you can use Unity Catalog Volumes and Genie Spaces together for multi-agent orchestration. This Skill supports configuring a Supervisor Agent to route queries between Knowledge Assistants built on Volume files and Genie Spaces for usage analytics, providing a unified conversational AI experience.

You connect external MCP servers to a Databricks conversational AI application by adding them as specialized components within a Supervisor Agent. This multi-agent orchestration layer routes user queries to the external MCP server alongside internal Unity Catalog Functions and model serving endpoints.

No, pre-existing model serving endpoints are not strictly required to create a multi-agent system. While the Supervisor Agent supports integrating custom endpoints, you can orchestrate a sophisticated conversational AI application using only Knowledge Assistants, Genie Spaces, and Unity Catalog Functions.

You should use a Supervisor Agent instead of a standalone Knowledge Assistant when you need to route user queries across multiple specialized AI agents. If your conversational AI application requires unifying document Q&A with usage analytics or external MCP servers, multi-agent orchestration is necessary.