Not all agents should be fully autonomous. Many enterprise scenarios require human-in-the-loop control, transparency, and override mechanisms. This session presents advanced design patterns for building AI agents in Azure AI Foundry that collaborate effectively with humans. We’ll cover interaction protocols, confidence-based escalation, agent explanations, and feedback capture. Using practical examples, you’ll see how agents can pause, request clarification, or defer decisions to users while maintaining context and task continuity. We’ll also explore how to integrate these agents into applications, dashboards, and workflows without breaking user trust or usability.