Microsoft Developer presents Agent Framework, a new open source SDK for building and orchestrating multi-agent AI systems. The session, featuring Jorge Arteiro, Evan Mattson, Elijah Straight, Eric Zhu, and Victor Dibia, covers core features, migration from Semantic Kernel and AutoGen, and ways to contribute.

Agent Framework: Building Blocks for the Next Generation of AI Agents

This video session from Microsoft Developer introduces the Agent Framework, an open source project that brings together Semantic Kernel and AutoGen, offering developers a unified pro-code SDK for designing and deploying multi-agent AI systems.

Overview

Open Source and Interoperability: The framework is designed for openness, scalability, and community contribution, converging AI agent development resources into a single platform.
Pro-Code SDK: Agent Framework enables direct programming and orchestration of agents across clouds, tools, and models, aligning with open standards like the Model Context Protocol (MCP).
Migration Pathways: Guidance is provided for migrating from previous Microsoft AI initiatives—Semantic Kernel and AutoGen—to the new framework.
Enterprise Ready: Features include robust memory and conversation state management, with demonstrations of enterprise-grade agentic AI workflows.
OSS Community Focus: The session encourages developers to get involved in the OSS project, emphasizing contribution and support for community-driven standards.

Key Chapters

Introduction to Agent Framework
Big announcement of open source SDK and convergence
Walkthrough of agent systems and key architecture
Migrating from Semantic Kernel and AutoGen
Memory management and enterprise deployment
Demo: Team Conversations
Community contribution and getting started

Resources

Contributors

Jorge Arteiro
Evan Mattson
Elijah Straight
Eric Zhu
Victor Dibia

Stay connected via LinkedIn and contribute to Microsoft’s OSS initiatives. New episodes released every Tuesday.