The Internet Is About to Have Another iPhone Moment – And It’s Called MCP

I believe we’re on the brink of a fundamental shift in how the internet works—not just for users, but for businesses, applications, and AI. This moment feels like when the iPhone launched: it changed everything. And I think the next iPhone moment is the rise of AI-native interaction powered by something called the Model Context Protocol (MCP).

Let me explain why I think this will change everything.

What Is the Internet, Really?

At its core, the internet is a collection of websites and web applications. Think:

Blogs (like this one!)
Online stores (e-commerce)
Forums and communities
Tools and apps (like calendars, or in my case, Kinvivo.com – a personal and household management app)

Originally, the internet was just a bunch of pages. You had to know where to go. Then search engines changed everything—especially Google. Suddenly, the journey started with a query.

You wanted to buy a bike? You googled it. You read blogs, browsed shops, checked forums, gathered opinions, compared prices. Eventually, you made a decision and bought something.

Search made the web accessible. But it also made it fragmented. You, the user, had to do all the stitching together. Search didn’t do that for you.

Enter Chatbots – And Why They’re Different

At first glance, LLMs (Large Language Models) like ChatGPT seem like upgraded search engines. They crawl vast swaths of the internet, aggregate data, and return answers.

But there’s a fundamental difference: they don’t send you somewhere—they bring the answer to you. More importantly, they let you interact. Ask, refine, dig deeper. It’s not just search—it’s a conversation.

Still, LLMs alone are limited. They can’t take action. That’s where MCP comes in.

LLMs + Action = Magic

Let’s talk about a real example.

In Kinvivo, I wanted to let users say something like:

“I need four liters of milk, two apples, and three bananas.”

Using Whisper (for voice transcription) and OpenAI Functions, I turn that into structured data:

[
  { "name": "milk", "notes": "4 liters" },
  { "name": "apples", "notes": "2" },
  { "name": "bananas", "notes": "3" }
]

Then I post this to a backend endpoint to update a grocery list. It’s seamless. But I want to go further.

What if users could just chat with the AI and let it update their lists directly? Or book appointments? Or order items?

That’s the promise of MCP.

What Is MCP?

MCP is a protocol that lets AI agents interact with external systems—not just talk about them. It defines:

What functions are available
What the system can do on behalf of a user
How to structure and return data

Instead of crawling a page and trying to simulate a click, an AI can now say:

“Hey, add these items to my grocery list.”

“Book a craftsman for next Tuesday at 8 AM.”

“Order the bike that fits my preferences and budget.”

This is intent-driven computing, and it’s designed for machines from the ground up.

How E-Commerce Will Be Changed Forever

Let’s say I want to buy a bike.

Traditionally, I’d:

Google types of bikes.
Research crossover models.
Open five tabs of reviews.
Compare specs and prices.
Finally make a decision and buy it.

With an MCP-enabled world, here’s the new flow:

I tell a chatbot what I need: “A weekend trail bike that I can also ride to work.”
The LLM talks to MCP services to discover types, brands, specs.
It pulls structured data from multiple shops.
It cross-references reviews and availability.
It shows me 3 personalized recommendations with prices, reviews, images.
I choose one.
It books it, schedules delivery, and adds it to my calendar.

No search. No tab explosion. Just outcome.

Why This Is Bigger Than You Think

This isn’t just a better UI.

MCP means we stop building digital infrastructure for humans—and start building it for machines that serve humans.

APIs were always meant to expose functionality.
Headless e-commerce started moving in this direction.
MCP finishes the job: it standardizes interaction for AI.

In a world where every service exposes its capabilities via MCP, AI agents don’t need to crawl. They just ask.

The Coming Shift

I used to believe that chat apps wouldn’t replace search engines. I thought they’d live side-by-side, like desktop and mobile.

But now? I’m not so sure.

Because when chat apps:

Answer your question better than search
Remember what you said last
Let you act on decisions directly (via MCP)

… then why would you ever go back to search?

And if AI-first interaction becomes the default, then Google is no longer the starting point of the web.

That’s why OpenAI and others are investing so heavily. Because this is more than hype. This is a platform shift.

Final Thoughts

I’m not excited about AGI. I’m not worried about AI replacing developers. But I’m incredibly excited about building for AI, not just with AI.

And I think MCP is the infrastructure we’ve been missing.

If you’re building software today—whether it’s e-commerce, productivity tools, or anything else—think about this:

Are you building for humans to click?

Or are you building for AI to act?

Because the next iPhone moment might just be this:

The moment we stopped searching—and started doing.

Let me know what you think. Are you as hyped as I am?