Databricks

Connect to Databricks to unify data, run analytics, and trigger workflows.

Overview

Databricks is a Data Intelligence Platform enables organizations to unify data, analytics, and AI workflows, allowing teams to build generative AI applications, democratize data insights, and govern enterprise data across multiple cloud environments while integrating with existing tools.

Connecting Fours to your Databricks account enables you to automate tasks within Databricks and integrate it with other applications, allowing for actions such as executing SQL queries, uploading data files, and triggering to run jobs defined in Databricks.

Create Integration

  1. Set up a Databricks account.
  2. Fours supports two kinds of Databricks authentication, following each documents to generate the corresponding credentials:
  3. Click the Databricks “Connect” button on the “Integrations” page.
  4. Enter Connection Information:
  • Host: The Databricks host URL, e.g. https://<databricks-instance>.cloud.databricks.com
  • Choose the authentication method you generated in step 2.
    • For Personal access token authentication, enter:
      • Token: The personal access token you generated in step 2.
    • For OAuth service principal authentication, enter:
      • Client ID: The client ID you generated in step 2.
      • Client Secret: The client secret you generated in step 2.

Example 1: Create integration using Personal access token authentication:

Databricks integration setup with personal access token authentication

Example 2: Create integration using OAuth service principal authentication:

Databricks integration setup with OAuth service principal authentication

Fours AI Tools

When Databricks is connected, the Fours AI agent can explore your Unity Catalog and run SQL on your warehouses through Fours’ built-in tools (10 actions).

Org-level only: Databricks tools use the organization’s Databricks credentials.

ToolDescription
list-catalogsList Unity Catalog catalogs
list-schemasList the schemas in a catalog
list-tablesList the tables in a catalog and schema
get-tableGet one table’s columns and schema, by its full name (catalog.schema.table)
list-warehousesList SQL warehouses — run-sql needs a warehouse ID
run-sqlRun a SQL statement on a warehouse and return the results. If they aren’t ready in about 50 seconds, it returns a statement ID to check with get-statement
get-statementGet the status and result of a statement by its ID
list-clustersList compute clusters
list-jobsList Databricks Jobs
current-userGet the Databricks identity the connection uses — a quick connection check

run-sql runs whatever statement it is given, including ones that change or drop data, so an agent gets it only with Full access to Databricks. With Read-only or Edit access it keeps the other nine tools, which only read. See Agent access levels.

Edit Integration

Edit is not supported for Databricks integration. For security, if you need to update credentials, delete the integration and recreate it.

Delete Integration

To delete the integration, simply click the 🗑️ button in the “Integrations” list. We will remove any credentials in our system.

Spotted something wrong or out of date on this page? Tell us and we'll correct it.