The AI Standard Hitting 97 Million Installs: Why Anthropic's MCP Could Change the Web for Your Business

Every so often a piece of technology crosses over from "thing engineers argue about" to "thing that quietly reshapes how everything works". In March 2026, Anthropic's Model Context Protocol, almost always shortened to MCP, hit that point. It passed 97 million installs, and the major software players, Microsoft and Google among them, have all adopted it. If you have never heard of MCP, that is completely fine. Most business owners haven't. But it is worth understanding the broad strokes, because it explains where the AI tools you already use are heading over the next couple of years.

We are going to keep this jargon-free. You will not need to write any code or change anything today. The goal is simply to help you recognise what is happening when your software vendors start talking about "AI integration" later this year.

What MCP actually is, in plain English

Until recently, an AI assistant like ChatGPT or Copilot was basically a very clever conversation partner. You could ask it questions and it would answer, but it lived in its own little box. It couldn't see your calendar, read your inbox, or look up a customer's record in your accounting software. If you wanted it to help with those things, you had to copy information in and out by hand.

MCP changes that. It is a shared standard for how AI tools connect to real-world apps, your calendar, your email, your databases, your accounting system, and act on what they find, in real time.

The cleanest way to think about it is the USB analogy. Before USB existed, every device had its own peculiar plug and its own cable. Connecting a printer was a small ordeal. USB created one standard socket, and suddenly anything could plug into anything. MCP is doing the same thing for AI. Instead of every AI tool needing a custom-built bridge to every piece of software, MCP gives them one common way to plug in.

What this looks like in practice

Imagine an AI assistant that can genuinely see across your business at once. You ask, "Has the Henderson job been invoiced, and is the client free for a follow-up next week?" and it checks your accounting software, looks at your calendar, and drafts the follow-up email, all in a single request, because it can reach all three systems through MCP.

That is the direction this is heading. Today's assistants mostly talk. MCP-connected assistants do. The protocol lets them take action across the tools you already pay for rather than living in isolation.

For a small business, the practical upshot is that the busywork sitting between your apps starts to dissolve. The constant copying of details from your inbox into your CRM, from your CRM into a quote, from the quote into your calendar, is exactly the kind of multi-step shuffling MCP is built to hand off.

Why 97 million installs is the number that matters

A new standard is only useful if everyone agrees to use it. A power socket that only fits one brand of appliance is worthless. The reason the 97 million figure is significant is not the raw number, it is what it signals: the whole industry has lined up behind the same approach.

When Microsoft, Google, and hundreds of software companies all start building "MCP servers", the little connectors that let their products speak the protocol, it stops being one vendor's bet and becomes the default plumbing of business AI. That is the moment a standard becomes safe to build on, because you can be reasonably confident it will still be here in five years.

For you, that means the AI features arriving in your tools through 2026 and 2027 are increasingly likely to share this common foundation. The accounting software, the booking system, the email platform, are all heading toward speaking the same language, which makes them far easier to wire together than the messy, one-off integrations of the past.

What Perth businesses should actually do

Here is the honest answer: nothing urgent, today. This is a "know where the road is heading" piece, not a "drop everything" one. You do not need to install anything, hire anyone, or learn a new system.

What is worth doing is paying attention. When a tool you already use announces AI integration or "connect your AI assistant" features over the coming year, MCP is very likely the technology underneath. Knowing that helps you ask the right questions: Does this connect to the other tools I rely on? What can the AI actually do once it's connected, and what can it not? Who can see my data when these systems talk to each other?

That last question matters most. The more your AI assistant can reach into your business, the more important it is to understand what it accesses and where that data goes. We covered the data-security side of AI tools in our earlier pieces, and the same caution applies here. Convenience and oversight need to move together.

For now, the smart move is simply to keep your core software modern and well-chosen. Businesses already on current, well-supported platforms will be able to switch these capabilities on as they arrive. Businesses limping along on ageing systems will find themselves locked out of the easiest productivity gains of the next few years. If you are unsure where your setup sits, that is exactly the kind of thing we help Perth businesses work through.