Skip to main content
When you interact with the AI — correcting a query, explaining what a metric means, or clarifying how a field should be filtered — that context is valuable. Learn from conversation lets the AI capture those insights and apply them to your semantic model, so future queries benefit from what was discussed. For example, if you tell the AI that “revenue should only include orders with status Complete, Shipped, or Processing,” the AI can learn that rule and encode it into the model as field descriptions, ai_context, synonyms, or new field definitions.

​
Requirements

This feature requires Querier permissions or higher, as these users can create workbook-level model changes and branches. Users with lesser permissions will not see the icon in the chat interface.

​
How it works

After a conversation where you’ve shared business context, click the that appears alongside the AI’s response. This triggers a review flow:
  1. The AI analyzes the conversation and generates model updates — such as field descriptions, labels, synonyms, ai_context annotations, or new dimensions and measures
  2. A review modal displays the proposed changes as model YAML
  3. You can accept, edit, or reject the changes before they’re applied
Changes are written to the model as YAML, making them available to all users querying through that model. Check out the following video to see how it works in Omni!

​
Auto-learn

Auto-learn is enabled by default for all organizations. With auto-learn enabled, the AI proactively identifies moments in a conversation where you’re sharing useful business context — like defining a metric, specifying filter criteria, explaining what a field means, or providing alternate names for a concept. When this happens, the AI surfaces a learning summary in the chat. You can then click to apply it through the same review flow described above.

​
What the AI can learn

The learn flow can generate updates including:
  • Field descriptions and labels — document what fields mean in your business context
  • ai_context annotations — add hints that help the AI select the right fields in future queries
  • Synonyms — register alternate names so the AI recognizes different ways users refer to the same concept
  • New dimensions and measures — define calculated fields that came up in the conversation
  • View-level context — describe what a view represents and how it should be used