Question 20
At fictional Redfern Regional Insurance, a Snowflake data and AI engineering team is conducting a multi-tenant knowledge assistant improvement before any automated output can change customer or financial records. The separately composed technical incident is: A regional retailer routes every natural-language analytics question through document similarity search, producing narrative extracts rather than correctly aggregating recognized sales measures across structured tables. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?
Explanation
Writer-proposed answer: D (UNVERIFIED)
Preliminary individual option explanations — NOT expert validated
Correct Answer
Record supported intent types, test routing decisions against signed business metrics and flag unanswered questions for semantic-model or retrieval-coverage improvements. Structured analytical intent requires governed business semantics and SQL evaluation; vector retrieval alone cannot reliably calculate authoritative aggregates.
Incorrect Answers
This response concerns Custom model artifact governance, which is different from the stated failure. A regional retailer routes every natural-language analytics question through document similarity search, producing narrative extracts rather than correctly aggregating recognized sales measures across structured tables.
Gen AI evaluation set representativeness is insufficient to demonstrate this result: A curated test set of aggregate and narrative questions returns correct reconciled warehouse metrics from structured queries and source-linked passages from documents.
Grounded employee policy assistant architecture is not the cause of the reported failure: A regional retailer routes every natural-language analytics question through document similarity search, producing narrative extracts rather than correctly aggregating recognized sales measures across structured tables.