Microsoft has just released the Microsoft Agent Framework (MAF). An open-source SDK and runtime designed for building, orchestrating, and deploying AI agents and multi-agent systems.
If you’ve worked with Semantic Kernel or AutoGen, you can think of MAF as their next evolution. Combining their best features into one unified, enterprise-ready framework.
Microsoft describes MAF as:
“A unified evolution of Semantic Kernel and AutoGen, providing an open-source, production-grade agent runtime.”
What is MAF?
It is an open-source SDK for .NET and Python that enables developers to:
-
Build intelligent AI agents
-
Create multi-agent workflows
-
Integrate with Azure AI Foundry, MCP servers, and other enterprise systems
What is the difference between MAF and Azure AI Foundry?
Both of them serve different roles in the AI ecosystem. The differences are:
-
Where it runs: MAF runs locally in your own runtime, while AI Foundry run in the cloud.
-
Purpose: MAF is an SDK for building and orchestrating AI agents, while AI Foundry is a cloud platform for managing, deploying, and monitoring agents.
So, in short, MAF helps you build the agent. Azure AI Foundry helps you manage it.
*MAF vs Semantic Kernel: *
Before MAF, many developers used Semantic Kernel (SK) to add LLM capabilities to their apps. SK remains great for single-agent copilots, but MAF takes things further. Enabling multi-agent collaboration and enterprise orchestration. MAF is pretty new and evolving, while SK is mature.
Source: DEV Community.
Leave a Reply