TLDR: Microsoft’s Azure AI Foundry Agent Service has reached General Availability (GA), now offering full support for the Model Context Protocol (MCP). This integration significantly simplifies the development of AI agents by enabling seamless connection to APIs and enterprise systems without the need for custom functions, fostering an ‘open-by-design’ AI ecosystem.
Microsoft has announced that its Azure AI Foundry Agent Service is now generally available, featuring comprehensive support for the Model Context Protocol (MCP). This pivotal advancement aims to revolutionize how developers connect Artificial Intelligence (AI) agents to diverse enterprise data sources, workflows, and specialized knowledge bases, eliminating the traditional complexities of custom integrations.
The Azure AI Foundry Agent Service, which initially became publicly available in May, serves as a unified platform for designing, deploying, and scaling enterprise-grade AI agents. The newly integrated Model Context Protocol, an open, JSON-RPC-based standard originally proposed by Anthropic, acts as a universal translator for AI integrations. It allows a ‘server’ to publish tools and contextual resources once, which any compliant ‘client’—such as the Foundry Agent Service—can then automatically discover and invoke.
This ‘connect once, integrate anywhere’ paradigm drastically simplifies the process of building and maintaining AI agents. Developers can now connect their AI agents to APIs and enterprise systems without the cumbersome need for hand-rolling Azure Functions, managing OpenAPI specifications, or developing bespoke plugins for each backend system. The service now functions as a first-class MCP client, enabling developers to integrate any remote MCP server—whether self-hosted or SaaS—and have its capabilities imported within seconds, with automatic updates and robust enterprise-grade security.
Kapil Dhanger, a Principal Cloud Solutions Architect at Microsoft, aptly described the MCP as ‘USB-C for AI integrations,’ underscoring its promise of universal connectivity. Echoing this sentiment, Solution Architect Abbas Ali Aloc remarked on LinkedIn that ‘MCP sounds like the Rosetta Stone for AI agents – excellent work in promoting interoperability!’
Microsoft’s commitment to an ‘open-by-design’ AI ecosystem was a key theme at Microsoft Build 2025, where CEO Satya Nadella highlighted the partnership with Anthropic to establish MCP as a foundational standard across various Microsoft platforms, including Windows 11, GitHub, Copilot Studio, and Azure AI Foundry. This move positions MCP as a critical enabler for a more interconnected and efficient AI development landscape.
Also Read:
- Microsoft Azure Welcomes Grok AI Models Amidst Shifting AI Landscape
- Microsoft Unveils Open-Source GitHub Copilot Chat for Enhanced AI Transparency in VS Code
While Azure AI Foundry Agent Service offers a unified platform with built-in MCP support, other major cloud providers are also embracing the protocol. Google Cloud, through its Vertex AI Agent Builder, Agent Development Kit, and ‘MCP Toolbox for Databases,’ provides a structured environment for MCP-based agent connections. Similarly, AWS leverages Amazon Bedrock Agents and a growing suite of open-source MCP servers to integrate real-time contextual information into tools like Amazon Q Developer, further solidifying MCP’s role as an industry-wide standard for AI agent interoperability.
