Microsoft Agent Framework Video

Microsoft’s latest agent pitch is no longer just about picking a model; it is about running agents inside a managed system with controls around them. (youtube.com) A Microsoft Mechanics video published April 22, 2026 said Foundry Agent Service lets companies deploy agents from local development, run them with secure identity and scoped permissions, and monitor each interaction. The same video framed the product as a path from prototype to production, not a demo tool. (youtube.com) Microsoft’s documentation describes Foundry Agent Service as a fully managed platform for building, deploying, and scaling agents, with hosting, scaling, identity, observability, and security handled by the service. It also says developers can use Microsoft Agent Framework, LangGraph, or their own code to deploy hosted agents into Foundry. (learn.microsoft.com) An AI agent is software that uses a large language model plus instructions and tools to take actions across several steps, not just answer with text. Foundry’s docs list built-in tools such as web search, file search, memory, code interpreter, Model Context Protocol servers, and custom functions. (learn.microsoft.com) Microsoft Agent Framework is the software layer for building those agents, while Foundry Agent Service is the hosted layer for running and governing them. Microsoft’s integration docs split that into two patterns: a code-first “Responses Agent” that does not create a server-side resource, and a versioned “Foundry Agent” managed through the Foundry portal or service APIs. (learn.microsoft.com) That split has become more concrete this month. Microsoft announced Agent Framework 1.0 on April 3, 2026, calling it the production-ready release for.NET and Python with stable application programming interfaces, long-term support, multi-agent orchestration, and support for protocols including Agent-to-Agent and Model Context Protocol. (devblogs.microsoft.com) Microsoft said in that April 3 post that Agent Framework unifies work that had been spread across Semantic Kernel and AutoGen. A separate Microsoft post says Agent Framework is now the successor to Semantic Kernel for building agents. (devblogs.microsoft.com, devblogs.microsoft.com) The tooling around that framework also moved closer to general release. Microsoft said April 16 that Foundry Toolkit for Visual Studio Code, formerly AI Toolkit, reached general availability, and a Foundry blog post on April 22 tied that toolkit to direct deployment into Foundry Agent Service. (techcommunity.microsoft.com, devblogs.microsoft.com) The company’s current message is that enterprises want the surrounding infrastructure as much as the model itself. In Microsoft’s own product pages and videos, the recurring features are versioned agents, role-based access control, tracing, managed authentication, virtual network isolation, stable endpoints, and publishing into Microsoft Teams, Microsoft 365 Copilot, and the Entra Agent Registry. (learn.microsoft.com, youtube.com) That makes the April 22 video less a product tutorial than a market signal: Microsoft is packaging agent creation, orchestration, deployment, and policy enforcement as one stack. The pitch is that companies do not just need smarter models; they need a governed place to put them. (youtube.com, learn.microsoft.com)