For roughly thirty years, the internet has existed as a tool designed and built for human beings. It’s been largely visual, devised to catch our eyes, hold our attention, and ultimately lead us somewhere—down a path, into a funnel, toward a buy button.

The architecture of the internet has been visual and emotional and informed at every turn by human psychology. It’s existed to primarily to tell human stories—with color, with typography, with images. In both an economic and literal sense, good design has always been good business.

Until now.

The Birth of a Primary Agent

There’s an idea floating around Silicon Valley right now that we’re headed towards a future where each of us will have a “primary agent”—a verified AI that acts on our behalf across the digital world. These agents will negotiate, transact, research, and decide, all without our direct involvement. They’ll do our shopping for us. They’ll do our travel booking for us. They’ll do all—or at least a good portion—of the browsing for us.

When they do, the theory goes, they’ll need to interact with a different internet than the one we know, one that’s designed for machine comprehension, where metadata matters more than messaging, and clarity of schema outweighs the clarity of your brand story.

The way it’s being talked about right now, it’s almost as if we’re back in the early days of SEO, when brands buried text on their pages to try and entice crawlers to rank them higher. Generative engine optimization (GEO) and answer engine optimization (AEO) are forcing brands to grapple with whether or not they need two websites: one designed for humans and one designed for LLMs and agents.

As ever, the human web is visual and persuasive. It’s intended to influence. It thrives on design, and narrative, and a thousand small psychological cues that move us from curiosity to conversion.

On the other hand, the agent web is utilitarian. It parses and ranks. It looks for signal. It doesn’t get bored, or distracted, or impressed. It doesn’t care about design or UX. It operates in a world of vectors and APIs, about as far removed from the digital experiences we’ve become accustomed to as possible. And that notion means that organization have to think more deeply about their digital strategy than they have had to for quite some time.

What This Means for Brands

I’ve long argued that even in an agentic AI world, the rules of designing content for humans still apply. But for the first time, I’m wondering if this still true. Or, maybe I’m beginning to think that it’s true, but only half true, because the internet is bifurcating itself. We may be entering an era where brands have to ask who their real audience is: the humans they currently count as customers and prospects, or the agents who belong to them.

This is a complicated and nuanced question that feels impossible to answer right now. We’re in the early innings of this thing, but it feels like in the not-too-distant future, companies might need to maintain two layers of presence—a human-facing site rich with narrative and imagery, and an agent-facing layer designed for retrieval and reasoning. The balance between them could define how discoverable, credible, and competitive a brand becomes.

If this split keeps widening, I think strategy becomes the bridge. The brands that learn to communicate fluently in both languages—the emotional syntax of humans and the logical grammar of machines—will define what the next internet looks like.