Version 26.5.0

Date: 10th May 2026

• Introduction to Data Products: ADOC introduces Data Products, a new capability that lets you group related data assets into named, business-meaningful products with aggregated health scores, governance checks, and a searchable Marketplace for discovery.
• Data Plane Maintenance Mode: ADOC introduces Data Plane Maintenance Mode for Data Plane V3, enabling data plane administrators to pause and resume data plane activity during upgrades and maintenance windows. When paused, scheduled executions are skipped and manual execution requests are blocked, while in-flight jobs continue to completion; on resume, scheduling restarts at the next configured cron schedule without backfilling skipped runs. For more information, see Manage Data Planes.
• Policy Execution Summary for Pushdown Engine: ADOC now supports policy execution summary downloads for the pushdown engine on Snowflake and Databricks, extending a capability previously available only for the Spark engine. Summaries include policy details, rule-level results, row counts, execution metrics, and good/bad record paths, are stored as JSON in global storage, and are available for download immediately after execution.
• Rules Page: ADOC now provides a centralized, policy-independent view of all configured rules on the Rules Page, allowing you to identify failing or unstable rules across policies without navigating individual policy pages. You can view execution history for up to the last 10 runs per rule and drill into execution results directly from each rule. For more information, see Rules.
• Redesigned Homepage: ADOC introduces a redesigned default homepage with a widget-based dashboard that consolidates key platform signals — Data Quality Score, Assets, Policies, Pipelines, Alerts, Watchlist, and Jobs — into a single view. You can also create custom dashboards using the same widget library for personalized monitoring views. For more information, see ADOC Home.
• Filter-Based Bulk Selection in Policy Export: When exporting policies, you can now select all policies matching the active filters in a single click, in addition to the existing options to select individual policies in the environment. This applies across filters including Data Source, Policy Type, Policy Status, and Tag. For more information, see Export and Import Manager.
• Asset Lineage — Pipeline Details on Lineage Nodes: The asset lineage view now displays pipeline job details directly on lineage nodes, replacing the previous hover-only behavior. Each node shows the job name, UID, execution status, last executed timestamp, duration, and owner and team where available, and selecting a node navigates to that specific job within the pipeline rather than the pipeline overview. Pipeline details can be toggled on or off using the Pipeline Details control in the lineage view. For more information, see Pipeline Run Details.

• Resolved an issue where the asset hierarchy on the Discover Assets page incorrectly displayed assets across all schemas instead of only under their assigned schemas. The hierarchy now correctly associates assets with their respective schemas.
• Resolved an issue where the "Runs" option under Look Back Window Type disappeared from Relative thresholds after switching away from and back to a supported rule type in data quality policies. The option now persists correctly across rule type changes.
• Resolved an issue where pushdown profiling on Trino data sources failed with a BIGINT overflow error when SUM aggregation exceeded the maximum BIGINT limit. Numeric columns are now cast to a higher-precision type (e.g., DECIMAL) before aggregation to safely handle large values.
• Resolved an issue where editing a Data Quality policy configured with Custom SQL (pushdown execution) allowed saving without SQL validation, causing the policy type to unexpectedly change from SQL-based to Column-based execution. The system now enforces SQL validation before allowing save on policy edits.
• Resolved an issue in Databricks where some table columns were not visible after crawling, even though they appeared during profiling. All columns are now correctly discovered and displayed after crawling.
• Resolved an issue where the pipeline search filter in the Monitoring Policy creation flow only searched within the currently loaded list instead of all available pipelines. Search now works across the entire pipeline list regardless of how many are loaded on screen.
• Resolved an issue where email notifications sent using the System Default Notification Template Group were intermittently delivered with a blank subject and body when multiple notification types were processed in quick succession.
• Resolved an issue where tag filters applied to a dashboard were not retained when the dashboard was reopened or edited, causing the filter selection to appear empty and data to load unfiltered.
• Resolved an issue where profiling failed for Snowflake columns of variant type when selected for column-level profiling using the Spark engine.
• Resolved an issue where onboarding multiple Iceberg data sources on the same data plane caused the second data source to display the catalog list from the first data source during test connection and crawl selection.
• Resolved an issue where editing or unarchiving a Reconciliation (Recon) policy failed when the underlying SQL View's join condition column was renamed, or when all columns referenced by the recon rule were renamed, by introducing an edit option that allows users to update the archived rules with the new column names instead of relying on the unarchive operation.
• Resolved an issue where SSO-authenticated users created via PingFederate were not transitioning to a SCIM-managed state in ADOC when later added to an Azure AD SCIM group, causing SCIM provisioning to fail with a "user create failed" error instead of detecting and updating the existing user.
• Resolved an issue where Cadence metrics were not generated for accessible Databricks (Unity Catalog) tables when the Torch Monitor service encountered permission errors on internal or unsupported tables within the same schema. The monitor now skips tables that fail due to permission errors or unsupported data sources and continues scanning the remaining assets in the schema.
• Resolved an issue where profiling and policy execution failed for AWS Glue data sources backed by S3 storage due to unsupported protocol handling and schema inference errors on Parquet files.