Ctrlk
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Add Databricks Data to an Agentic Workflow

Add Read Data and Write Data actions to your Agentic Workflow in Kizen to query or update your external Databricks SQL Warehouse in real time.

Audience: Admins, Developers, Solution Architects

Purpose: Explains how to read and write Databricks data from an Agentic Workflow to query or update your SQL Warehouse

Overview

Caution: This setup reflects Kizen's default configuration. Your administrator may have customized your layout, so columns or navigation may appear differently. Trial accounts may have limited features.

Once you configure your connection to Databricks, you can add Read Data and Write Data action steps to your Agentic Workflows in Kizen to query or update your Databricks SQL Warehouse in real time.

Before You Begin

To add Databricks data to an Agentic Workflow, you need an active Databricks connection. For details, see Set Up Databricks in Kizen.

Add Databricks Data to an Agentic Workflow

To add Databricks data to an Agentic Workflow:

1

Go to the Agentic Workflow builder

In the global navigation, select Agentic Workflows. Then, to open the builder, do one of the following:

  • Select New Agentic Workflow.

  • Select the name of an existing Agentic Workflow.

The New Agentic Workflow button on the Agentic Workflows page
2

Add an action to the Agentic Workflow

In the builder, select +, and then select Add Action.

The Add Action context menu item in the Agentic Workflow builder
3

Select the Databricks Connection action

In the Add Action modal that appears, scroll to the Integration section, and select the Databricks Connection card.

The Databricks Connection card on the Add Action modal
4

Choose whether to read or write Databricks data

In the Actions Settings section of the Add Action modal, select the Action dropdown menu, and do one of the following:

  • To read Databricks data, select Read Data.

  • To write Databricks data, select Write Data.

The Read Data item of the Action menu under Action Settings in the Add Action modal

Caution: The Write Data action can insert, update, or delete data in your Databricks data warehouse. Validate your query in a non-production environment before running it against production data.

5

Configure the sources for the script inputs

For each of the fields in the Script Inputs section of the modal, select the source of the field's value from the Input Source dropdown menu, and then configure the source with the additional controls that appear.

The script inputs for the Read Data action in the Add Action modal

Note: The order of the fields on the screen depends on whether you selected the Read Data or Write Data action.

The fields are as follows:

Field
Type
Description
Required?

Connection Secret Tag

String

The name of the Databricks environment from your integration secret to execute the query on. Configure this source when you have more than one environment in your secret.

Not in a single-environment configuration

Database

String

The name of the Utility Catalog to query in Databricks.

Yes

Query

String

The SELECT statement to execute.

Yes

Return Single Value

Boolean

Whether you expect to return a single value (yes) or multiple values (no).

Yes

Schema

String

The name of the target schema within the specified Databricks Utility Catalog.

Yes

Note: Always use the LIMIT clause in SELECT queries against large tables. Requests for more than 25 MB of data fail outright, with no partial or truncated results returned.

6

Configure the destinations for the script outputs

For each of the fields in the Script Outputs section of the modal, select the destination of the field's value from the Output Destination dropdown menu, and then configure the destination with the additional controls that appear.

Note: The Read Data action uses only the Result field as a script output. The Write Data action uses both Result and Result Status.

The fields are as follows:

Field
Type
Action
Description
Required?

Result

String

Read Data

The result of the query

No

Result Status

String

Read Data, Write Data

The status of the Write Data operation

No

7

Configure error handling

In the Error Handling section of the modal, specify how to handle errors in this step of the Agentic Workflow.

8

Save the Databricks Connection action

Select Save.

What's Next

Continue to Troubleshoot Databricks to learn how to fix common technical issues with the Databricks integration.

Related Topics

Databricks and Unity Catalog are trademarks of Databricks, Inc. Delta Lake is a trademark of the Delta Lake project and its contributors. Kizen's Databricks integration is not affiliated with, sponsored by, or endorsed by Databricks, Inc. or the Delta Lake project.

PreviousSet Up Databricks in Kizen
NextTroubleshoot Databricks

Last updated

Was this helpful?