Release v3.3

v3.3

This release adds better DQ-AI Assistant context, ECLASS workflows, and SQL/Python issue exceptions. It also improves ruleset actions and navigation.

Added support for context files and instructions in our DQ-AI Assistant
The files and instructions are stored on the ruleset level and the Assistant will consider them for all users who can access this ruleset
Allows users to give the DQ-AI Assistant more relevant input when generating rules, explanations, or suggestions
Helps improve answer quality by allowing users to provide additional business or technical context

Added a new ECLASS node for product classification workflows
This supports product data quality and enrichment use cases where product data needs to be checked or mapped against the ECLASS standard (e.g., for DDPs, AAS)
Helps automate classification-related checks and reduce manual validation effort

Added support for issue exceptions in SQL and Python rules that was only available for other rules before
Users can now exclude accepted or known issues from the issue count without removing them from the total checked records

Added the "Improve issues" button to the ruleset header and above the issue table in the rule detail view
Opening it from a rule detail creates a filtered empty workflow for that rule’s issues when no workflow exists yet
Moved the "Add rule" button into the Rules tab

Made the Custom SQL condition available when two tables from the same connector are selected
Fixed an issue where column names with double quotes could break ruleset checks
Fixed issue where row filters were applied after the LLM max row limit instead of before it
Improved reading of static Excel files and prevent uploading of .xls uploads going forward
Added automatic restart handling for stale pending profilings
Fixed issue where users already belonging to a tenant could be invited again
Fixed a bug where no notification was sent when only table-based rules resulted in an error