Microsoft's New Agent Framework
Microsoft has recently launched the Microsoft Agent Framework, an open-source software development kit (SDK) and runtime designed for creating AI agents and multi-agent workflows. Released on October 1 and available on GitHub, this framework supports both .NET and Python, allowing developers to build anything from basic chatbots to complex workflows featuring graph-based orchestration. By using this framework, developers can first experiment locally before deploying their applications to the Azure AI Foundry. This transition comes with built-in features that enhance observability, ensure durability, and maintain compliance.
This new framework combines and expands on concepts from previous projects such as the Semantic Kernel and AutoGen. For developers already familiar with these projects, moving to the Microsoft Agent Framework should be relatively easy. It supports various protocols, including the Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol, and is designed with an OpenAPI-first approach that enables cloud-agnostic operations.
The capabilities of the Agent Framework are noteworthy. Agents can discover and utilize external tools or data servers, leveraging the MCP for connectivity. They can communicate across different environments using structured messaging dictated by the A2A protocol. Additionally, any REST API that meets an OpenAPI specification can be effortlessly imported as a callable tool. This flexibility allows agents to operate in containers or across multiple cloud platforms, enhancing their adaptability in various business environments.
Moreover, the framework supports a diverse array of orchestration patterns. It includes an experimental features extension package and offers pluggable memory modules. Developers can choose from options like Redis, Pinecone, Qdrant, Weaviate, Elasticsearch, Postgres, or even a custom storage solution for conversational memory. Built-in connectors also facilitate integration with multiple platforms, including Azure AI Foundry, Microsoft Graph, Microsoft Fabric, SharePoint, Oracle, Amazon Bedrock, MongoDB, and various Software as a Service (SaaS) systems through Azure Logic Apps.
As businesses look for effective ways to implement AI-driven solutions, the Microsoft Agent Framework offers a robust foundation for innovative application development. By providing versatile tools and integration capabilities, it presents a valuable resource for developers and businesses aiming to enhance their workflow with AI technology.
“Content generated using AI”
We create intelligent software and AI-driven solutions to automate workflows, modernize legacy systems, and sharpen your competitive edge.