SnowPro Specialty: Gen AI Question 182
Single answerSnowflake CortexA retail company stores customer support tickets in a Snowflake table and wants to build a lightweight solution that automatically produces a short summary and sentiment label for each new ticket. The data must remain in Snowflake, and the team wants to minimize operational overhead by avoiding external model hosting or data movement. Which approach best meets these requirements using Snowflake Cortex?
- A
Create a SQL pipeline that calls Snowflake Cortex functions such as SUMMARIZE and SENTIMENT directly against the ticket text stored in Snowflake tables.
- B
Export ticket text to an external Python service, host an open-source LLM on a VM, and write the predictions back to Snowflake.
- C
Create a Snowpark Container Services application because Cortex functions require custom container deployment before they can be called from SQL.
- D
Train a custom classification model inside a virtual warehouse because Cortex only supports user-trained models, not managed inference functions.
Show answer and explanation
Correct answer: A
Explanation
The best answer is to use Snowflake Cortex functions directly in SQL because the scenario prioritizes keeping data inside Snowflake and minimizing operational overhead. Cortex is designed for in-platform AI inference, allowing teams to apply managed functions to Snowflake data without standing up separate model-serving infrastructure or exporting data to external systems. For this scenario, summarization and sentiment analysis are standard inference tasks that map well to Cortex capabilities. This aligns with Snowflake best practices of reducing data movement, simplifying architecture, and leveraging managed services where possible. Candidates should recognize that Cortex supports direct AI function invocation from Snowflake workloads, while options involving external hosting, container deployment, or custom model training add complexity that is not required for the stated business need.
- A. Correct.
Correct. Snowflake Cortex provides managed AI functions that can be invoked directly from SQL on data stored in Snowflake, which is well aligned with the requirement to keep data in Snowflake and reduce operational complexity. For a use case like summarizing support tickets and deriving sentiment, using Cortex functions such as SUMMARIZE and SENTIMENT is the most direct and operationally simple approach.
- B. Incorrect.
Incorrect. While this could technically work, it violates the stated goal of minimizing operational overhead and introduces unnecessary data movement and external infrastructure management. One of the main benefits of Snowflake Cortex is using managed AI capabilities within Snowflake instead of exporting data to externally hosted models.
- C. Incorrect.
Incorrect. Snowpark Container Services is used for deploying custom containerized workloads, but it is not a prerequisite for using Snowflake Cortex functions. This option reflects a common misconception that all AI inference in Snowflake requires custom application hosting. Cortex managed functions are designed to be called directly without building and operating containers for standard tasks.
- D. Incorrect.
Incorrect. Cortex is not limited to user-trained models. Snowflake Cortex includes managed LLM-powered functions for inference tasks such as summarization and sentiment analysis. This option confuses custom model training with managed AI inference capabilities available through Cortex.