Session Abstract

AI agents look magical from the outside—but behind every “intelligent” action is a stack of Azure services that must be secured, monitored, and governed. This session opens the hood and shows you exactly what’s happening when an agent reasons, retrieves, calls tools, and interacts with enterprise systems. We’ll use live demos to trace real agent activity across Azure services like Azure OpenAI, Search, Foundry, and others. You’ll learn how to instrument, enforce identity boundaries, apply safety controls, monitor performance, troubleshoot failures, and manage cost. If you want a practical, scenario‑driven guide to running AI agents in production—this is your session.

Vinicius Apolinario

Vinicius has been working with IT for 20+ years. He has a passion for helping ITPros/Ops and whenever possible, you'll see him on conferences engaging with customers and community. Vinicius holds many Microsoft certifications, has 2 books published (Learning Hyper-V, and Windows Containers for ITPros), is an MCT since 2009, has worked as a Program Manager in the Windows Containers team, and blogs about anything he thinks will help customers be successful.