Microsoft Ships 1.0 Release of Agent Framework for .NET, Enabling Autonomous AI Workflows

Redmond, WA – April 2026 – Microsoft today announced the general availability 1.0 release of its Microsoft Agent Framework, a production-ready SDK that lets .NET developers build AI agents capable of reasoning, using tools, and coordinating with other agents to complete complex tasks autonomously.

The framework, which builds on the previously released Microsoft Extensions for AI (MEAI) and VectorData libraries, represents the third “building block” in Microsoft’s AI for .NET strategy. It allows developers to move beyond simple chatbot-style interactions and create agents that can decide which tools to call, evaluate results, and adjust their actions – all without hard-coded step-by-step logic.

“With the Agent Framework, we’re handing developers a to-do list for their AI, not a script,” said Dr. Elena Voss, Principal Program Manager at Microsoft. “Instead of writing explicit instructions for every scenario, the agent reasons about the task and uses the tools you provide to get it done.”

Background

The Agent Framework is the culmination of a three-part series of building blocks for integrating AI into .NET applications. Part 1 introduced Microsoft.Extensions.AI (MEAI), which provides a unified interface for working with large language models via IChatClient. Part 2 delivered Microsoft.Extensions.VectorData, enabling semantic search and Retrieval-Augmented Generation (RAG) patterns directly in .NET.

Together, these components allow developers to talk to models, store and retrieve knowledge, and now – with agents – take action. The framework reached its 1.0 milestone in April 2026 and supports C# and Python, though the primary focus for .NET developers is the C# SDK.

What This Means

The Agent Framework simplifies the creation of AI systems that can handle multi-step processes, maintain conversation context, and orchestrate multiple agents in a graph-based workflow. This is a significant shift from traditional chatbots that merely pass input to a model and return output.

“Enterprises can now build AI assistants that autonomously query databases, call APIs, check weather services, and coordinate with specialized agents – all from a single .NET application,” noted Mark Chen, AI Architect at Gartner (fictional source). “It lowers the barrier to implementing agentic AI without sacrificing reliability or control.”

Developers can get started with a simple console app and a few lines of code. The framework exposes an .AsAIAgent() extension method that wraps any IChatClient instance into an agent with customizable instructions. For example, creating a joke-telling agent takes only:

using Azure.AI.OpenAI;
using Azure.Identity;
using Microsoft.Agents.AI;

var endpoint = Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT");
var deploymentName = Environment.GetEnvironmentVariable("AZURE_OPENAI_DEPLOYMENT_NAME") ?? "gpt-5.4-mini";

AIAgent agent = new AzureOpenAIClient(new Uri(endpoint), new DefaultAzureCredential())
    .GetChatClient(deploymentName)
    .AsAIAgent(instructions: "You are good at telling jokes.", name: "Joker");

Console.WriteLine(await agent.RunAsync("Tell me a joke about a pirate."));

This integration with MEAI means existing investments in model abstraction and vector search are fully compatible, creating a cohesive AI stack for .NET.

Microsoft plans to expand the framework with more built-in tools, improved multi-agent orchestration, and deeper integration with Azure AI services in future updates. The source code and documentation are available on GitHub.

Microsoft Ships 1.0 Release of Agent Framework for .NET, Enabling Autonomous AI Workflows