Contents:

• Introduction

• MCP and API Delivery

• Interesting Content for this Week

• Conferences

MCP and API Delivery

The hype around the Model Context Protocol (MCP) is not abating. Almost daily, new innovations are being announced. In this post, I’ll touch on what MCP means for API delivery—the design, development, deployment, and operation of APIs.

MCP standardises how large language models (LLMs) connect to services (via APIs) and tools. It is meant to replace the previously fragmented approach to integrating LLMs with various data sources by offering a unified protocol. MCP was announced in November 2024 by Anthropic. More recently, OpenAI announced that its Agents SDK now supports MCP. The OpenAI team is also working on adding MCP support to both the ChatGPT desktop app and the OpenAI API in the coming months.

For developers, MCP enables IDEs to connect to data sources—such as GitHub repositories—to provide rich AI assistance. For example, using the GitHub MCP server, developers can update files, automate pull requests, track issue tickets, and search for code. Popular agentic coding IDEs that support MCP include Cursor and Claude Code.

This richer AI assistance reduces context switching between tools. For example, agentic IDEs, using MCP clients, can read data from a Notion MCP server to guide feature implementation, create PRs and branches in GitHub, and message teammates via a Slack MCP server. The MCP servers continue to interact with the APIs of their respective applications—and in fact, MCP may drive increased API usage.

Your API templates, catalog and style guide, are useful contexts to provide via an MCP server, so that developers can receive in-line assistance to design APIs that conform to organisational standards.

Context management is important for agentic AI. Agentic AI refers to systems that use autonomous AI agents to make decisions and take actions in pursuit of specific goals—such as automating workflows that previously required human intervention. Agentic AI can transform the API delivery process. It can reduce lead times by accelerating API design and development. From a governance perspective, it also unlocks tools to enforce API standards more systematically. I will explore this further in future posts.

Regarding API discovery, the emergence of GitHub MCP servers highlights the importance of treating OpenAPI and Arazzo files as code and storing them in version-controlled repositories like GitHub. Organisations should also consider creating MCP servers for their internal API catalogs.

In summary, MCP provides a standardised interface for AI applications to interact with external systems. Its client-server architecture allows it to benefit from a rapidly growing ecosystem of MCP servers. Ultimately, MCP has significant potential to enhance both API delivery and governance.

Interesting Content for this Week

This week, I am going to focus on some recent content on the model context protocol and agentic AI.

How Agentic AI and APIs Are Powering the Next Wave of Enterprise Innovation: Karl Fankhauser discusses composable APIs in the Agentic AI era. Karl discusses opportunities and challenges for agentic AI and APIs to improve complex flows in healthcare and other industries.

MCP: The Differential for Modern APIs and Systems: Steve Manuel shares how MCP improves system integration reliability by mitigating disruptions caused by API changes.

MCP: The Ultimate API Consumer (Not the API Killer): Kevin Swiber writes that with the introduction of MCP, APIs are encountering a transformative and are poised to redefine data interchange. He discusses MCP misconceptions and ramifications.

Is MCP the New API? How Model Context Protocol Is Changing the Way Banking Apps Talk to Data: Suman Devarasetti discusses the ascendancy of Generative AI, characterised by the development of intelligent assistants, and it is concurrent with the emergence of the Model Context Protocol (MCP), which is set to revolutionise API interactions.

The Model Context Protocol (MCP) — A Complete Tutorial: Nimrita Koul provides a great, comprehensive introduction to MCP.

Kong Mesh 2.10: Simplified Provisioning and Policy Management: Kong Mesh 2.10 introduces enhancements to improve the provisioning experience and simplify policy management for its users.

Is Your API Ready for the AI Agents? : Michal Trojanowski, highlights the increasing prevalence of AI agents necessitates that service providers develop accessible APIs and comprehensive documentation, thereby facilitating integration.

Conferences

Postman's annual user conference: POST/CON 25. Date: June 3rd & 4th 2025, Location: JW Marriott Los Angeles L.A. Live, Los Angeles, CA Register Here 

APIdays Helsinki: Theme: “APIs for Innovation, Intelligence, and Impact” Date: June 3rd & 4th 2025. Location: Pikku-Finlandia, Helsinki Register Here

APIdays Germany: Theme: “Accelerate AI Use Cases with APIs” Date: July 2nd & 3rd, 2025. Location: Smartvillage Bogenhausen, München, Germany. Register Here

APIdays London: Theme: “No AI Without APIs” Conference Date: September 22nd - 24th, Location: Convene 155 Bishopsgate, London EC2M 3YD