Databricks Vector Search Tool

The Databricks Vector Search tool leverages an existing Databricks vector search index, and provides semantic document retrieval capabilities.

It empowers other agents to query data stored in a Databricks vector search index.

Prerequisites

The Databricks Vector Search Tool is provided by the “Databrick AI” plugin, which you need to install.

You need:

  • A Databricks Workspace with Model Serving enabled.

  • A Dataiku Connection to Databricks configured with either a Personal Access Token or OAuth2 credentials.

  • A configured Databricks Vector Search index

Usage

You need to configure:

  • The Databricks Connection

  • The Vector Search endpoint and index (possible values will be automatically fetched)

  • The list of columns to return