Skip to content

Native Data Quality Advisor

The Native Data Quality Advisor is a conversational agent that helps you decide which data quality checks to apply to your data, and helps you build reusable data quality skills. It works in two modes.

  • Recommendation mode — given a table and its columns, it proposes appropriate data quality checks for each column.
  • Skill authoring mode — it helps you create, refine, and test reusable data quality skills through conversation.
  • In recommendation mode, the agent reads the table and column information you supply and outputs structured checks keyed by column.
  • In skill authoring mode, the agent asks clarifying questions before generating or refining skill content, drawing on a catalog of data quality rules, domain skills, and standards.

Required:

  • message (string): What you want help with (e.g., “recommend quality checks for this table” or “help me build a freshness skill”).

Optional:

  • table_id (string): A table to use as initial context, e.g. "table.25".
  • skill_ids (list): Restrict the agent to specific skills, useful for testing one skill in isolation.
  • skill_content (string): Existing skill content you want the agent to improve.

The agent draws on Alation’s data quality knowledge through internal tools that load the rules catalog, domain-specific skills, and data quality standards. These are configured on the agent by default.

  • In recommendation mode the agent outputs checks immediately rather than holding a conversation.
  • In skill authoring mode it asks clarifying questions first.
  • It loads the rules, skills, and standards catalogs once per session rather than re-fetching them repeatedly.