By Scott David, Founder & Director

A few years ago, every technology company seemed to be having the same conversation.

How do we use AI?

Today, the conversation feels much more practical.

How do we connect AI to the systems we already have?

That's a very different challenge.

Building an impressive AI model is one thing. Helping that model interact with documents, databases, internal tools and business applications reliably is something else entirely. It's where many organisations are now focusing their attention, and it's one of the reasons the Model Context Protocol, or MCP, is generating so much interest.

If you haven't come across it yet, that's understandable. It isn't the sort of technology that dominates headlines in the same way a new language model does. Yet, in many ways, it could have a much bigger influence on how AI is adopted across enterprise software.

The easiest way to think about MCP is as a common language.

For years, software teams have spent significant amounts of time building bespoke integrations. Every application speaks slightly differently, every API has its own conventions and every new connection introduces another layer of maintenance. As AI becomes part of more business workflows, that complexity only increases.

MCP offers a different approach. Rather than every AI application building its own way of talking to external systems, it defines a standard for how models, tools and data sources communicate. The protocol is designed to make those interactions more consistent, more portable and easier to scale across different environments.

In many respects, it reminds me of what happened with USB-C.

Before it became widely adopted, every device seemed to require its own cable. Nothing quite connected in the same way and every manufacturer had its own approach. Eventually, the industry realised that a common standard made life easier for everyone.

MCP feels like it could play a similar role for enterprise AI.

That doesn't make it exciting in the traditional sense. Standards rarely are. They're rarely the headline announcement at a conference or the feature that gets people talking on social media.

What they do create is confidence.

When businesses know their AI systems can connect to different tools using a common protocol, they can spend less time worrying about integration and more time thinking about the problems they're trying to solve.

That has interesting implications for engineering teams too.

As AI becomes embedded across more products, software engineers are increasingly being asked to think beyond models. Understanding APIs, distributed systems, security, permissions and data architecture is becoming just as important as understanding the AI itself.

It's another example of how software engineering is evolving. The engineer's role isn't becoming narrower because of AI. It's becoming broader. Building intelligent products increasingly means understanding how lots of different systems interact rather than focusing on a single technology in isolation.

For hiring managers, I think that's an important point.

The conversation around AI recruitment often centres on finding people with experience of the latest models or frameworks. Those skills absolutely matter, but they're only one part of the picture.

The businesses likely to benefit most from AI over the next few years will be the ones capable of integrating it effectively into existing products, workflows and customer experiences. That requires engineers who understand systems, architecture and interoperability every bit as much as machine learning.

Whether MCP becomes the dominant standard remains to be seen. Technology has a habit of evolving in unexpected directions, and no protocol is guaranteed long-term success.

What's already clear, though, is that enterprise AI is entering a different phase.

The focus is shifting from experimenting with models to connecting them with the tools businesses rely on every day.

If that trend continues, protocols like MCP may end up being remembered as some of the most important technologies of this generation of AI.

Not because they were the most visible.

Because they quietly made everything else possible