Note: This is the first of a series of five posts I’ll be making this week on MCP and how I see its transformational impact on the architecture for the next generation of AI systems.

Post 1 in the Series: Introducing MCP to Marketing Tech Leaders

By Dan Jaye, CEO

I’ve been thinking a lot about Model Context Protocol (MCP) for months now, and I keep returning to one thought: MCP is to generative AI what Open API and Swagger were to the modern web. If that analogy doesn’t immediately resonate, let me paint you a picture of why I believe MCP is fundamentally reshaping how we architect AI systems.

Cast your mind back to the early 2000s web development landscape. Every integration was a bespoke nightmare – custom APIs, proprietary formats, incompatible systems that required armies of developers just to make two applications exchange basic data. Then Open API emerged and gave us something revolutionary: a standard language for digital systems to communicate. Suddenly, we had this elegant, federated web where applications could seamlessly integrate with hundreds of services without reinventing the wheel each time.

In advertising technology, I witnessed this transformation firsthand with Open RTB – a specialized standard built on Open API principles that now processes trillions of programmatic transactions daily. That same pattern is now unfolding in AI, but with compressed timelines that would make your head spin.

Today, Trapped in Boxes (but not for long)

Right now, we’re trapped in what I call “the most backwards user interface imaginable” – a text box. Think about that for a moment. We have these extraordinarily sophisticated AI systems, capable of reasoning, analysis, and complex problem-solving, and our primary interface is… typing into a box and hoping for the best. It’s like owning a Ferrari and being permanently stuck in first gear.

This fragmentation isn’t just inconvenient – it’s architecturally limiting. Before MCP, connecting an AI model to your Google Drive, customer database, and Slack required three different custom implementations, each with its own quirks and maintenance overhead.

How MCP Transforms the Game

Here’s where MCP changes everything, and I mean everything.  Imagine an MCP client as your AI’s web browser.  Just as your browser today elegantly handles all the complexity of rendering websites, downloading resources, and managing interactions, an MCP client orchestrates the intelligence layer – navigating AI systems, accessing tools, and coordinating sophisticated workflows.  The MCP client becomes your interface to large language models, where that’s GPT-4, Claude, Mistral or Gemini.

On the flip side, MCP servers function like websites in this ecosystem. They expose resources, tools, and prompts that clients can discover and access. But here’s the elegant part: because everything adheres to the same standard, any MCP client can seamlessly work with any MCP server. We’re transitioning from isolated AI applications to a federated ecosystem of interoperable components.

The Momentum is Undeniable

What I’m witnessing in the market is acceleration that frankly surprises even me. We’re already seeing over 5,000 active MCP servers as of May 2025, according to Glama’s public directory. Major players like Salesforce, Block, and Apollo have integrated MCP into their platforms, while development powerhouses including Zed, Replit, Codeium, and Sourcegraph are enhancing their tools with MCP integration.

Direct competitors are embracing this standard because they recognize what I’ve recognized: this isn’t a convenient add-on protocol – it’s the foundational architecture for the next generation of AI systems.

At Aqfer, I’m finding that client conversations have become dramatically more productive. Instead of explaining complex custom integrations, I can articulate how our solutions integrate into this emerging standard ecosystem. Use case after use case maps perfectly to MCP components. “Oh, that’s just an MCP server. Oh, that’s an MCP resource.” The framework is so intuitive it’s reshaping how we think about AI solution design.

The Strategic Window Is Open

We’re still in the early chapters of this story. Authentication protocols are evolving rapidly. Security standards are being hammered out. Most MCP applications currently run locally, and the governance frameworks for federation are still emerging. But that’s precisely why now is the moment to engage deeply.

The companies that recognized Open API’s potential early dominated the web economy. The companies that embraced Open RTB early owned programmatic advertising. I’m convinced the same dynamic will play out with MCP in AI – but faster and with higher stakes.

Move to Post 2 in the Series Here: Sampling, Elicitation, and Completions: The New Vocabulary of AI Orchestration