Snowflake Cortex Analyst enables natural language queries on enterprise data - but it requires accurate semantic metadata to work reliably. DE Copilot generates the business definitions, field context, and governance metadata that powers a trustworthy semantic layer.
Cortex Analyst is only as accurate as the semantic metadata it reads. DE Copilot generates that metadata from your STTM and business rules - ensuring business context is captured, governed, and kept in sync.
Field-level business definitions, domain context, and plain-language descriptions - generated from STTM metadata and kept in sync with the actual data model.
Table relationships, join paths, measure definitions, and dimension hierarchies - derived from the Canonical Metadata Model and structured for Cortex Analyst consumption.
Auto-generated data dictionaries provide the field-level context that Cortex Analyst needs to interpret natural language queries accurately.
When semantic metadata is accurate and complete, Cortex Analyst produces reliable SQL. DE Copilot ensures the metadata layer is governed, traceable, and up to date.
The AI layer suggests business synonyms, common query patterns, and semantic relationships - enriching the metadata model beyond what is explicitly documented in the STTM.
All semantic metadata passes through human review before it is published to the Cortex Analyst semantic model. Engineers validate business definitions before they power natural language queries.