The artificial intelligence (AI) landscape is poised for significant transformation with the latest Model Context Protocol (MCP) update.
On March 26, 2025, developers behind the open standard finalized a new version that promises to revolutionize how AI agents interact with tools, data and interfaces across different platforms.
What is the Model Context Protocol?
Initially introduced by Anthropic in late 2023, MCP addresses a critical challenge in the AI ecosystem: Enabling AI agents to communicate seamlessly with various tools and applications. The protocol is a universal language, allowing different AI systems to interact consistently with real-world tools, regardless of origin.
Major Players Join Forces
This update is particularly significant because of its broad industry support. OpenAI, the industry leader in generative AI, announced immediate support for MCP across its products. CEO Sam Altman confirmed that MCP support is already available in OpenAI’s Agents SDK, with integration coming soon to ChatGPT’s desktop app and the Responses API.
Microsoft has also backed the protocol, launching a new Playwright-MCP server that enables AI agents like Claude to browse the web and interact with sites using the Chrome accessibility tree. AI systems can click, type, browse and engage with web content like human users.
Key Improvements in the Update
The March 26 update introduces several crucial enhancements:
- OAuth 2.1-Based Authorization Framework: Adds robust security standards for agent-server communication, vital for HTTP-based transports.
- Streamable HTTP Transport: Replaces the older HTTP+SSE setup, enabling real-time, bidirectional data flow with improved compatibility.
- JSON-RPC Batching: Allows clients to send multiple requests simultaneously, reducing latency and improving efficiency in agent- tool interactions.
- Tool Annotations: Adds rich metadata describing tool behavior, enabling more sophisticated discovery and reasoning by AI agents.
The protocol’s modular design, built on JSON-RPC 2.0, allows developers to implement only the necessary components while maintaining compatibility across the ecosystem.
Practical Applications
Microsoft’s Playwright-MCP server demonstrates the practical power of this protocol. AI agents can perform complex web interactions by wrapping browser automation capabilities in the MCP standard. They can navigate websites, input data, take screenshots and interact with page elements through accessibility descriptors.
This transforms AI agents into versatile assistants capable of test automation, quality assurance and data navigation tasks that previously required human intervention.
The Bigger Picture: Standardization Wins
The growing support for MCP represents a significant win for standardization in the AI industry. With major players like Anthropic, OpenAI, Microsoft and LangChain aligning behind a standard protocol, we’re witnessing the emergence of what Alexander Doria, co-founder of AI startup Pleias, calls “the protocol era of AI.”
This standardization addresses one of the most significant barriers to AI advancement: Interoperability. MCP enables more sophisticated, capable and reliable AI applications by creating a common language for AI systems to communicate with tools and interfaces.
What’s Next?
According to Mitch Ashley, VP and practice lead, DevOps and Application Development at The Futurum Group, “Model Context Protocol (MCP) is today’s best option for bridging the gap between AI models and products and other data, websites and systems, but it has not yet become a de facto standard. OpenAI, Microsoft and Amazon take different paths in supporting MCP. OpenAI is adding the MCP protocol across their products, Microsoft extended the Playwright server (built on the Playwright browser automation library), and Amazon added MCP support within Amazon Bedrock. Widespread adoption of MCP would push Meta and Apple to use MCP, and deepen its integration in OpenAI, Microsoft and Amazon.”
The question now is whether other major tech companies like Meta, Amazon and Apple will follow suit. If they do, MCP could become the universal standard for AI actions, fundamentally changing how AI systems interact with the digital world.
As we enter this new era of AI interoperability, the possibilities for AI applications expand dramatically. The Model Context Protocol may be remembered as the breakthrough that enabled truly integrated, capable AI systems to become a practical reality.
