Microsoft merges AutoGen and Semantic Kernel into a unified agent framework

Microsoft Agent Framework is an open-source SDK and runtime designed to simplify orchestrating multi-agent systems. It converges the experimental multi-agent research of AutoGen with the enterprise-ready plumbing of Semantic Kernel into a single commercial-grade package.

What it actually does

Developers typically hit a wall when turning research demos into production agents. According to industry data, fragmented tooling costs 50% of developers more than 10 hours every week.

This framework lets developers prototype locally and deploy straight to Azure AI Foundry with built-in durability, observability, and compliance features.

Interoperability is a major focus here. It natively supports Anthropic's Model Context Protocol (MCP) for dynamic tool connections, OpenAPI standards, and the Agent2Agent protocol for communicating across different runtimes.

It also wraps advanced orchestration patterns like Magentic One directly into structured workflows, helping agents collaborate without constant manual babysitting.

The tooling and ecosystem

The framework is available right now in public preview as an open-source project, with Microsoft welcoming external contributions.

In the cloud, Azure is rolling out multi-agent workflows in Foundry Agent Service as a private preview. It adds a stateful workflow layer for long-running business processes, complete with error recovery and visual debugging via a dedicated VS Code extension.

Observability also gets a major upgrade. Working with Cisco's Outshift, Microsoft contributed agentic tracing standards to OpenTelemetry. Azure AI Foundry can now monitor workflows built not only on its new framework, but also on third-party stacks like LangChain, LangGraph, and OpenAI's Agents SDK.

At the same time, the low-latency Voice Live API has reached general availability. Early enterprise adopters including KPMG, Commerzbank, and Elastic are already testing the framework for financial auditing, avatar support, and search pipelines.

Why it matters

Orchestrating autonomous agents has felt like duct-taping incompatible libraries together. Merging AutoGen and Semantic Kernel while embracing open standards like MCP and OpenTelemetry shows Microsoft is building for real-world developer workflows rather than a walled garden.

If you are building multi-agent systems and tired of fragmented toolchains, it is well worth exploring in preview today. Keep in mind that full-scale cloud workflows with persistent state are still in private preview, so production rollouts will need to wait on broader cloud access.