Skip to main content
When modifying a concept, it’s important to understand the distinct roles of definitions versus SQL components:

​
Contextual Definitions:

  • Impact how the agentic flow understands and retrieves the concept
  • Help translate natural language queries to the right entities
  • Provide business context and meaning
  • Should include relevant terminology, synonyms, and use cases
  • Affect which questions will surface this concept as relevant

​
SQL Components:

  • Determine the actual data retrieval logic
  • Control which records and fields are included
  • Define calculations, transformations, and aggregations
  • Impact query performance and accuracy
  • Determine how data is structured and joined
Real-World Example: Imagine users asking “Show me winning deals by product line” but getting no results. There could be two different issues:
  1. Definition Issue: The Opportunity concept definition doesn’t mention that opportunities can be filtered by product line, so the agentic flow doesn’t recognize this as a relevant entity for the question. Solution: Update the contextual definition to include terms like “product line,” “product category,” and “winning deals.”
  2. SQL Issue: The concept correctly matches the query, but the base query doesn’t include product line information or doesn’t properly identify winning deals. Solution: Modify the SQL to include product line fields and ensure the logic for identifying winning deals is correct.
Making effective changes to your semantic model often requires addressing both the semantic understanding (definitions) and the technical implementation (SQL) to ensure questions are both correctly interpreted and accurately answered.