Will AI Agents Kill the Web as We Know It?
The way we interact with the web today is surprisingly manual. Want to book a flight? You’ll probably head to a familiar airline’s website or open Google and type in your dates. If that site also offers hotel and car rental options, great—you might stay and book everything in one place. But more likely, you’re picky. So you go off searching for that perfect boutique hotel or the restaurant you’ve read about. Click by click, tab by tab, you stitch your trip together.
Ordering food works the same way. You might order directly from a local takeaway you trust, but more often than not, you'll use a delivery platform like Deliveroo. These aggregators simplify the process—storing your address, payment details, and past orders—making it easier to go through them than deal with individual sites.
But all of this is may be about to change (if you believe the agentic hype).
Instead of navigating to websites or tapping through apps, we’ll increasingly delegate these tasks to AI agents. Soon, I’ll simply say:
“Book me a flight to Barcelona for next weekend, boutique hotel in the Gothic Quarter, and a dinner reservation for Saturday night. Make sure it’s walkable.”
That’s it. No more URLs. No search boxes. No forms. Just a prompt—and everything gets done.
Today, the agent might simulate a user—scraping content, clicking buttons, filling out forms. But very soon, it will bypass the interface entirely. It will plug directly into APIs and structured data formats, authenticate itself, and complete the transaction behind the scenes.
One of the key enablers of this future is the Model Context Protocol (MCP)—a new standard that lets websites expose structured actions and interfaces designed specifically for AI agents to consume. Instead of trying to reverse-engineer a checkout flow, the agent can simply ask the site what actions it supports, and then execute those actions with the right permissions and context.
This unlocks a new kind of machine-to-machine commerce. And it will have profound implications.
First, it undermines traditional search. If AI agents are fulfilling intent directly, nobody’s Googling “best tapas bar in Barcelona.” That means fewer impressions, fewer clicks, and far less advertising revenue.
Second, it hollows out the role of websites. If agents are transacting via MCP or similar protocols, the front-end—the polished UI, the clever copy, the brand storytelling—may never be seen.
We’ve been here before.
Think back to the rise of e-commerce. As online shopping became easier and more convenient, foot traffic to high streets and town centres declined. Many brick-and-mortar stores shuttered. To survive, retailers had to reinvent themselves. Shopping became an experience—with immersive store design, knowledgeable staff, in-person exclusives, and tactile engagement that online couldn’t replicate.
We’ll likely see a similar shift with agentive commerce.
When AI agents handle the utility—finding products, comparing prices, making purchases—online stores will need to counter with something agents can’t replicate: a compelling, meaningful user experience.
Things like:
Interactive product demos
Vibrant communities and social proof
Content-rich environments that blur the line between brand and entertainment
In short, websites will need to become more than just places to transact. They’ll need to offer reasons to explore, to engage, and to feel something.
The web is about to split in two.
One layer will be invisible: efficient, transactional, built for machines—powered by protocols like MCP.
The other will be human-facing: emotional, experiential, and deeply brand-driven.
AI agents won’t kill the web. But they will radically reshape it. And just like with retail, the winners will be the ones who make that second layer worth visiting.