What happens when an AI agent is asked to support an important enterprise decision but the information it receives is incomplete or inconsistent?

In this practical, demo-first session, we'll use Microsoft Copilot Studio to build and progressively improve an enterprise decision-support agent. Through a realistic production-change scenario, we'll explore how agent instructions, document evidence, and organizational knowledge influence the quality of its analysis.

You'll see how an agent can identify missing information, surface conflicting evidence, and prepare useful insights while keeping final decisions with people.