VHOL: Agent Evaluations with Cortex Code

Building AI agents is just the beginning — understanding how well they perform and systematically improving them is what separates prototypes from production systems. Join this instructor-led lab to learn how to build, evaluate and optimize AI agents using Cortex AI, Cortex Code and Agent GPA. You’ll deploy a multi-tool analytics agent, stress-test it with real-world queries, analyse failures using observability logs — and iteratively improve orchestration instructions using evaluation-driven workflows. You’ll walk away with a repeatable workflow for continuously improving AI agent performance using real production interaction data.

In this lab, you’ll learn how to:

• Build a Cortex Agent with multiple tools including Cortex Analyst, Cortex Search, stored procedures, web search and data-to-chart generation

• Use Snowflake Intelligence and Cortex Code to capture observability traces, mine failures and curate evaluation datasets

• Run Agent GPA evaluations, identify orchestration weaknesses and optimize agent instructions for measurable performance improvements

• Validate improvements by comparing evaluation results across agent iterations

Prerequisites

• A Snowflake account with ACCOUNTADMIN access or a Snowflake free trial account with AI functionalities enabled. We will send instructions on how to set it up.

• Basic familiarity with Snowflake SQL and Cortex Agents