Assess data quality using AgentFlow™ Quality

Learn how to assess the quality of a Databricks table and publish validated data to Reltio using AgentFlow™ Quality.

Attention: This feature is available to limited users through the Reltio Early Access (EA) program. Interested in finding out more about this feature or participating in our EA program? Get details in topic Early Access (EA) features.

AgentFlow™ Quality assesses a Databricks table against data quality rules and returns pass/fail results for each rule. You can then publish the validated data to Reltio using Data Loader, all in the same conversation without re-entering connection details.

Prerequisites

Before you begin, ensure that you have the required system roles and Databricks connection details:

The following system roles to access AgentFlow Quality:
- ROLE_AGENT_FLOW_QUALITY
- ROLE_EXECUTE_AGENTS
- ROLE_EXECUTE_MCP.
- ROLE_DATALOADER
Access to a Databricks account
Connection details for your Databricks workspace:
- Server hostname
- HTTP path
- Access token
- Catalog name
- Schema name

To connect to a Databricks source

Open AgentFlow Quality.
Select the required tenant.
Select Start Data Quality.
The Data Quality orchestrator appears.
Select Databricks to connect to your source data.

In the Connect to Databricks page, enter the following connection details.

Connection details	Description
Connection name	A label to identify and save the connection for later reuse.
Server hostname	The hostname of your Databricks workspace.
HTTP path	The HTTP path of the Databricks SQL warehouse you want to connect to.
Access token	A Databricks personal access token to authenticate the connection.
Catalog name	The Unity Catalog catalog that contains the tables you want to load.
Schema name	The schema within that catalog that contains the tables you want to load.

Select Connect.
AgentFlow Quality validates your credentials, saves the connection under the name you provided, and displays the Select tables screen for the catalog and schema you provided.

To assess data quality using AgentFlow Quality

Select the checkbox for each table you want to assess, then choose Select.
AgentFlow Quality lists each table's description, number of columns, and number of rows to help you choose. You can select more than one table in the same conversation.
In the Suggested rules section, review the rules AgentFlow Quality suggests for each table.
(Optional) To add a custom rule, describe it in natural language, and AgentFlow Quality adds it to the suggested rules.
Select the checkboxes for each rule you want to run, and choose Select.
The Project Review summary appears.
Review the Project Review summary, then select Approve.
The summary lists the project name, source, number of datasets, and number of rules linked to the project.
AgentFlow Quality runs the selected rules against the table and displays the AgentFlow Quality Report and DQ rule outcomes.

To publish validated data using AgentFlow Quality

Review the following reports.

Report	Description
AgentFlow Quality Report	A summary of the quality score, rules passed, records scanned, violations, and a breakdown of quality by dimension.
DQ rule outcomes	Displays the result for each rule you selected.

Respond to the AgentFlow Quality prompt asking whether to map the table and load it into Reltio.

Response	Description
Yes	AgentFlow Quality maps the table to your Reltio tenant schema using the same connection and creates a Reltio Data Loader job.
No	AgentFlow Quality does not publish the table. The report stays in the conversation so you can refer to it at any time.

Result

AgentFlow Quality generates the quality report and displays it in the conversation. If you continue to publish the validated data, AgentFlow Quality creates a Reltio Data Loader job from the current conversation.

To review the mapping and run the data loader job, see Publish a Databricks table to Reltio using AgentFlow™ Quality.

Unify and manage your data

Assess data quality using AgentFlow™ Quality

Footer