SnowPro Specialty: Gen AI Question 68
Single answerStored natively in semantic views (Public Preview)A retail analytics team is building a semantic view in Snowflake so business users can ask natural-language questions about sales performance. The team wants the semantic definitions to live with the semantic view itself, rather than storing YAML files in an external stage and managing separate versioned files. Which approach best meets this requirement for semantic views stored natively in Snowflake (Public Preview)?
- A
Define the semantic model directly in the semantic view so the metadata is stored natively in Snowflake rather than in external YAML files.
- B
Store the semantic model only in a Cortex Search Service, because semantic views automatically read their definitions from the search index.
- C
Create a standard SQL view and attach masking policies, because semantic definitions are inferred automatically from the view text.
- D
Upload the semantic model YAML to an internal or external stage, because semantic views in Snowflake require staged files and cannot store metadata natively.
Show answer and explanation
Correct answer: A
Explanation
The best answer is to define the semantic model directly in the semantic view so it is stored natively in Snowflake. This Public Preview capability is designed for teams that want semantic definitions governed and managed as Snowflake objects instead of as separate YAML artifacts in stages. In practice, this reduces operational overhead, simplifies lifecycle management, and aligns better with Snowflake-native governance. The distractors reflect common confusion between semantic views, Cortex Search, ordinary SQL views, and the older staged-YAML workflow. Candidates should recognize when a requirement is specifically about where semantic metadata is stored and managed. Refer to Snowflake documentation for semantic views and the Public Preview support for storing semantic definitions natively in Snowflake.
- A. Correct.
Correct. In the Public Preview capability for semantic views stored natively, the semantic metadata is defined and managed within Snowflake as part of the semantic view object itself. This addresses the team's requirement to avoid maintaining separate YAML files in stages and keeps the semantic definition colocated with the governed Snowflake object.
- B. Incorrect.
Incorrect. Cortex Search Service is a separate capability for retrieval over indexed text and is not the storage mechanism for semantic view definitions. A common misconception is that all GenAI-related metadata in Snowflake is backed by search indexes, but semantic views and Cortex Search serve different purposes.
- C. Incorrect.
Incorrect. A standard SQL view can expose relational logic, but it does not by itself become a semantic view with native semantic metadata. Masking policies help with governance and data protection, not with defining business semantics for natural-language querying.
- D. Incorrect.
Incorrect. This describes the earlier file-based pattern in which a semantic model could be managed as YAML in a stage. The scenario explicitly asks for the native-storage approach, which is intended to avoid dependence on separately staged model files.