What if your next project manager was… an AI? And what if it actually did stuff instead of just sending calendar invites?

Welcome to the rapidly evolving world of autonomous agents, powered by Microsoft Copilot and extended through Model Context Protocol (MCP) and Agent-to-Agent (A2A) collaboration.

In this entertaining and hands-on session, you’ll build your own intelligent agent—one that can reason, plan, act, and coordinate with other agents like a tiny digital middle manager (but useful).

Using tools from the Microsoft ecosystem—including Microsoft Copilot Studio, Prompt Flow, and MCP-enabled integrations—you’ll go from “what’s an agent?” to “I just built one that handles workflows, talks to other agents, and maybe escalates issues before I even notice them.”

We will cover:

• What makes an agent truly autonomous - (Hint: it’s not just a glorified macro with commitment issues)
• How MCP (Model Context Protocol) enables agents to access tools, data, and context so they can actually do things instead of hallucinating confidence
• How A2A (Agent-to-Agent) communication lets multiple agents collaborate because one overworked AI apparently wasn’t enough
• Designing, testing, and deploying agents with Microsoft Copilot without accidentally creating a productivity-themed horror story
• Common pitfalls like agents that loop forever, over-automate nonsense, or schedule meetings at 3 a.m. because “technically available”
• Real-world scenarios - where agents coordinate across systems, workflows, and even other agents (great for impressing your boss and mildly alarming your coworkers)

Whether you're a cloud architect, an AI tinkerer, or just someone tired of doing repetitive work, this session will show you how to build a digital sidekick—or a small army of them—that actually delivers value.

Coffee fetching still not guaranteed. Delegating that to an A2A network of caffeine-optimized agents is left as an exercise for the brave.