Agentic AI systems that schedule meetings, update CRMs, and perform sentiment analysis sound great in demos. Getting them into production inside an enterprise with existing APIs, security policies, and compliance requirements is a different problem entirely. This architecture-focused session covers integration patterns for connecting AI agents to enterprise systems using the Microsoft Agents Framework, Azure AI Foundry Agent Service, and Model Context Protocol (MCP). Topics include tool-calling design, authentication flows for agent-to-service communication, and the architectural decisions that determine whether an agentic system becomes a production asset or a stalled proof of concept.

You will learn:

• How to design integration architectures for AI agents that connect to existing enterprise APIs and services
• How to evaluate when to use the Microsoft Agents Framework vs. Azure AI Foundry Agent Service vs. custom orchestration
• How to implement authentication and authorization patterns for agent-to-service communication