Agents are impressive in demos. In production? That’s where things get… interesting.
Because an agent that usually works is not the same as one you can trust.
When you start putting agents in front of real users, things change. Prompts behave differently, grounding gets creative, and that one action you were sure about suddenly isn’t so reliable anymore.
In this session, we’ll look at what it means to test agents in Microsoft Copilot Studio. Beyond just running through a few happy paths and hoping for the best.
We’ll explore how agents behave when users don’t follow the script (because they won’t), and how to validate prompts, grounding, actions, and orchestration in a way that reflects reality.

Join me to learn how to move from “it seemed fine when I tested it” to something you can confidently put in front of users… without holding your breath 😉