One of the most important patterns in the history of the internet is that the companies controlling discovery tend to accumulate disproportionate power.
- Google controlled discovery through ranking.
- Amazon controlled product discovery through marketplace search and recommendation.
- Instagram controlled attention through the feed.
- TikTok controls attention through algorithmic recommendations.
- App stores controlled mobile distribution.
Every major internet platform that inserted itself between users and businesses eventually became extraordinarily powerful. Why? Because controlling discovery means influencing decisions. And influencing decisions means shaping the market underneath.
AI assistants are widely hypothesised to become the next version of that layer.
The interface is becoming the intermediary
Historically, the web worked through navigation.
Users clicked links, browsed websites, compared pages, opened tabs. Learned how to navigate different digital environments. Even when Google became dominant, users still spent most of their time inside destination experiences after discovery.
But AI changes the shape of the interaction. Increasingly, users ask questions directly.
“What’s the best…”
“Compare…”
“Help me choose…”
“Plan…”
“Recommend…”
The AI system no longer simply points somewhere. It interprets, filters, frames, compares and recommends.
That sounds subtle, but it is strategically enormous. Because once the intermediary starts shaping decisions rather than merely routing traffic, the intermediary accumulates influence over the entire market underneath it.
AI is becoming a super-aggregator
Historically, aggregators became powerful because they helped users navigate abundance.
Booking.com helped navigate hotels. Skyscanner helped navigate flights. Amazon helped navigate products. Google helped navigate information. Comparison sites, marketplaces and search engines all became valuable because the web became too large and fragmented for users to navigate efficiently on their own.
But AI systems are themselves becoming navigation layers. And that creates a fascinating shift.
For years, aggregators sat between users and inventory owners. They controlled filtering, comparison, discovery and recommendation. That position gave them enormous leverage over merchants, publishers, hotels, airlines, restaurants and brands underneath them.
But now AI systems increasingly sit above the aggregators themselves. Which raises an important question: what happens when the aggregator gets aggregated?
If an AI assistant can compare hotels, summarize reviews, understand preferences, explain tradeoffs, shortlist options and recommend the best fit - then why does the user necessarily visit the comparison site at all?
The same question applies across huge parts of the internet.
- Why navigate twenty tabs if the AI system synthesises the landscape for you?
- Why manually filter products if the AI already understands your intent?
- Why browse complex category trees if the AI can route you directly toward likely decisions?
AI systems are not just another interface layer. They are potentially a new meta-layer sitting above much of the existing internet.
Not all businesses are exposed equally
What makes this shift particularly interesting is that different types of businesses face very different risks.
Some businesses primarily won because they controlled navigation and aggregation. Others won because they controlled inventory. Others won because they built trust, taste or emotional connection. AI affects each of these differently.
Aggregators may be the most exposed. If a business’s primary value was helping users compare, filter and navigate abundance, AI increasingly performs many of those functions natively.
Travel comparison sites are an obvious example. Historically, they helped users navigate complexity: flights, prices, routes, dates, reviews and availability. But AI assistants are believed to be naturally good at exactly this kind of synthesis. The danger is that many aggregators become invisible infrastructure underneath the AI layer itself.
Merchants and inventory owners face a different problem. They risk commoditization. If products increasingly appear as generated recommendations inside somebody else’s interface, then businesses can become reduced to specifications, availability, pricing, ratings and structured metadata.
In other words, inventory. That is dangerous because many brands spent decades building emotional differentiation above and beyond the underlying product itself.
But there is another category of company that may become more valuable in the AI era, not less: Curators. Editorial brands. Taste-driven businesses. Experience-first companies. Because while AI is excellent at summarising information, it is much weaker at replicating trust, identity, aspiration, emotional resonance and point of view.
The Economist is not valuable simply because it contains information about politics and economics. Its value also comes from editorial judgment, worldview and trust. A luxury travel brand is not valuable simply because it can technically book hotels. Its value comes from curation, taste and confidence. A great retailer is not just a logistics system. It helps people imagine who they want to become.
These are fundamentally more emotional and experiential forms of value. And that matters because AI tends to compress informational differentiation faster than emotional differentiation.
Search ranked links. AI adds interpretation.
Google mostly organised the web through retrieval. AI systems increasingly organise the world through interpretation, which is a much deeper layer of influence.
A search engine might return ten hotels, ten publications or ten SaaS vendors. The user still performed most of the reasoning. But AI systems increasingly collapse comparison, summarisation, prioritisation and recommendation into the intermediary layer itself.
The assistant does not merely retrieve options - it increasingly shapes what matters, which tradeoffs are important, which products feel trustworthy and which brands even enter consideration. That changes how value flows across the internet - the battle shifts upward.
For years, digital businesses obsessed over traffic acquisition: SEO, paid acquisition, landing pages, funnels and conversion optimisation. But AI shifts the strategic bottleneck, with the critical question increasingly becoming: who shapes the decision before the click even happens?
Because if the AI assistant already frames the shortlist, the comparison and the recommendation, then much of the market influence has already been captured before the user ever reaches the destination experience.
The experience layer becomes strategically important again
There is an important nuance here: Controlling discovery is powerful. But controlling experience still matters enormously. Even if AI systems increasingly shape awareness, comparison and recommendation, brands still own what happens after intent is routed toward them. And paradoxically, that layer may become even more important.
Because if every AI assistant can summarise products, services and information, then the actual experience surrounding the decision becomes one of the few remaining places where brands can still create differentiated value. That includes trust, emotional connection, confidence, guidance, taste, storytelling, interaction quality and identity.
In other words, AI systems may control more of the decision layer, but brands can still control how decisions feel. And for many important purchases, that matters enormously.
The next decade of the internet may be shaped by two layers
I increasingly think the next version of the internet consists of two connected but distinct layers.
The first layer is machine-facing. This is where businesses structure products, services and knowledge so AI systems can interpret them accurately.
The second layer is human-facing. This is where brands help people explore, compare, understand and feel confident in decisions.
The companies that win will likely understand both. They will become legible to AI systems underneath, while building deeply differentiated experiences on top.
Because in the AI era, becoming discoverable is not enough. The harder challenge is remaining meaningful after the intermediary has already summarised you.