As enterprise AI adoption accelerates, data agents that can plan, retrieve, reason, and act across structured and unstructured sources are becoming foundational. But building agents that work is no longer enough, you need to build agents you can trust. This 60-minute workshop walks through how to design, deploy, and evaluate multi-agent systems using Snowflake Cortex. You’ll build agents that connect to enterprise data sources (structured and unstructured) and perform intelligent, multi-step operations with Cortex Analyst and Cortex Search. Then we’ll go beyond functionality and focus on reliability. You’ll learn how to instrument your agent with inline, reference-free evaluation to measure goal progress, detect failure modes, and adapt plans dynamically. Using trace-based observability tools like TruLens and Cortex eval APIs, we’ll show how to identify inefficiencies and refine agent behavior iteratively. By the end of this workshop, you’ll: - Build a data agent capable of answering complex queries across multiple data sources - Integrate inline evaluation to guide and assess agent behavior in real time - Debug and optimize execution flows using trace-level observability - Leave with a repeatable framework for deploying trustworthy agentic systems in production