I had the pleasure of presenting “Scalable Multi-Agent AI with AutoGen” at Boston Code Camp 38, and wow—what a crowd! The room was full, the energy was high, and the questions and conversations just kept flowing. In fact, the interactions were so engaging that I couldn’t even complete my demo! That’s always a great problem to have.
What We Covered
The session was focused on demystifying modern AI agent architectures, particularly using AutoGen, an open-source framework from Microsoft designed for orchestrating multi-agent conversations with LLMs, tools, and human-in-the-loop workflows. Here’s a quick recap of the key topics:
- What Are AI Agents?
We explored the core traits that distinguish agents from simple chatbots—autonomy, perception, goal orientation, and tool usage. - Building Blocks of Modern AI Systems
From completion and chat to function calling, RAG, and agentic reasoning, we broke down how LLM systems are architected. - Agent Frameworks in the Microsoft Ecosystem
A deep dive into AutoGen alongside tools like Semantic Kernel, Azure AI Foundry, and Copilot Studio. - AutoGen Capabilities
Code generation, tool orchestration, customizable agent roles, multi-agent collaboration, and human supervision. - Model Context Protocol (MCP)
We walked through a real-world agent interaction scenario using MCP to access web services via structured agent communication.
Interactive Energy
The highlight of the session was definitely the audience participation. From use case questions and architecture design discussions to practical deployment considerations, the room was buzzing. It’s clear that multi-agent AI is a topic that’s resonating with developers and solution architects alike.
Missed the Demo?
No worries—you can find presentation slides and sample code from the session below:
📂 Download the Presentation (PPTX)
💻 Get the Sample Code on GitHub
Final Thoughts
Thanks again to everyone who attended and brought the energy to Boston Code Camp 38. If you want to keep the conversation going, feel free to connect with me on LinkedIn, check out my site at udai.io, or drop in at the New Hampshire Cloud .NET User Group for more AI sessions.
Until next time—keep building, keep learning!