Microsoft has rolled out a new open-source tool called the Microsoft Agent Framework. This software development kit (SDK) and runtime makes it easier for developers to create, manage, and deploy AI agents and multi-agent workflows. It supports both .NET and Python, two popular programming languages. The framework was announced on October 1 and is now available on GitHub.

This new framework aims to help developers build everything from simple chatbots to complex systems with multiple AI agents working together. It offers graph-based orchestration, which means you can design workflows that connect different agents seamlessly. Developers can test their projects locally and then deploy them to the Azure AI Foundry for production. The system comes with features like observability, durability, and compliance built in, making it suitable for enterprise use.

The Microsoft Agent Framework can connect to various platforms including Azure AI Foundry, Microsoft 365 Copilot, and other agent systems. Microsoft explained that this new framework combines and extends ideas from earlier projects like Semantic Kernel and AutoGen. Developers familiar with those will find the transition to the new framework smooth.

What Makes the Microsoft Agent Framework Stand Out

The framework supports key protocols like the Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol. It is designed with an open API approach and a cloud-agnostic runtime, which means it can run anywhere—on-premises, in the cloud, or across multiple clouds. This flexibility makes it highly portable.

Agents built with this framework can discover and use external tools or data sources dynamically. They can also communicate with each other using structured messaging based on the A2A protocol. Any REST API that follows the OpenAPI specification can be imported and used as a callable tool instantly. This makes integrating new services straightforward.

The framework supports various orchestration patterns, allowing developers to design workflows that match their needs. It also offers an extension package for experimental features and pluggable memory modules. Developers can choose from Redis, Pinecone, Qdrant, Weaviate, Elasticsearch, Postgres, or even their own custom memory stores to manage conversational context.

Connectivity and Compatibility

One of the key strengths of the Microsoft Agent Framework is its extensive support for connectors. It includes built-in integrations with services like Azure AI Foundry, Microsoft Graph, Microsoft Fabric, SharePoint, Oracle, Amazon Bedrock, MongoDB, and SaaS platforms via Azure Logic Apps. This wide range of connectors makes it easier to build multi-platform, multi-cloud AI solutions.

Developers can also leverage the framework’s support for containers, enabling deployment across various environments. Whether deploying on local servers, private data centers, or multiple cloud providers, the framework adapts easily. Its open API-first design ensures that new tools and integrations can be added with minimal effort.

The introduction of this open-source framework signals Microsoft’s commitment to making AI development more accessible and flexible. It builds on existing projects and standards, aiming to accelerate the creation of intelligent multi-agent systems that can operate across different platforms and environments. This move could significantly lower the barriers for developers wanting to build advanced AI workflows and integrate them into larger enterprise systems.

In summary, the Microsoft Agent Framework is a powerful new tool for AI developers. Its support for multiple languages, protocols, and deployment options makes it a versatile choice for building next-generation AI solutions. As it gains traction, it could reshape how businesses and developers approach multi-agent AI development.

Inspired by

• https://www.infoworld.com/article/4067500/microsoft-unveils-framework-for-building-agentic-ai-apps.html

Sources

• gmpg.org
• microsoft.com
• github.com
• computerworld.com
• github.com