AI search isn’t an upgrade – it’s a rebuild

In Part 1, we examined the impact of AI on search and how many incumbents are well-positioned to take advantage of this technology to improve their existing propositions.

But the truth is, AI won’t just improve the way we currently experience search. It will rebuild the entire experience. We see this happening in a few ways:

• For one, AI will fundamentally challenge cost-per-click ad models, as users will have fewer clicks and see fewer results. That necessitates a total rethink of the way search results are presented and the way companies monetise search.
• Add to that the rapid rise in AI agents autonomously navigating the web, and the whole rulebook as we know it goes out the window.

Faced with a transformation of this pace and scale, we see a number of opportunities for startups:

1. Vertical search: Where AI actually works today

We’ve seen numerous billion-dollar outcomes in vertical search (Kayak, Indeed, Zillow, TripAdvisor) over the last 20 years. But as AI-native vertical search players emerge, we believe these incumbents are particularly vulnerable to disruption compared to more generalist players. There are a couple of key reasons for this:

Low barrier to entry for startups: Vertical AI search can more easily overcome the intent cold start problem. Users typically arrive with a specific, high-intent query (e.g., “I want to book a trip to Canada”), and the universe of possible results is relatively constrained. This makes it easier for new entrants to deliver relevant results quickly, without requiring vast amounts of data scale and compute, advantages that are key to the success of generalist platforms like Google or Meta.

Traditional advertising models go out the window: For many existing leaders in vertical search, cost-per-click ads are at the core of their business model. As AI streamlines the user journey (with fewer, more relevant results and clicks), monetisation will shift from clicks to outcomes, necessitating a total rethink of revenue streams.

So, where do we expect winners to emerge?

Today, informational vertical search is ahead. Companies like Perplexity, Consensus, and Open Evidence have grown quickly, benefiting from AI’s ability to aggregate and synthesise information far more efficiently than traditional search. Monetisation remains early, with most players currently pursuing premium models, but questions around margins and scale remain.

That’s why Perplexity and others are beginning to experiment with ads. Just like pure-play transactional AI vertical players like Mindtrip or Jitty, they are still in very early stages. While AI is clearly providing faster scanning, aggregation and insight delivery for informational search, we’re yet to see the same level of disruption and transformation in transactional search.

Still, consumers also use GenAI tools to enhance their shopping (see above) and we remain confident that winners will emerge in transactional vertical search. As user experiences improve, we expect to see breakout companies define this next wave. Down the line, these vertical winners could gradually expand to more general-purpose search engines over time, challenging the incumbents here too.

2.  Beyond search – agents will replace the entry point to the web

We believe AI agents will fundamentally reshape how people navigate the web — not just through better search, but by redefining the entire user interface layer.

Two major shifts are underway:

1. The collapse of the search vs. discovery distinction

Traditionally, search was goal-driven (Google, Amazon), while discovery was passive and feed-based (TikTok, Twitter). The line between the two is getting increasingly blurred.

Feeds are now being used for search: 40% of Gen Z already prefer TikTok or Instagram for everyday search queries. As AI further personalises these experiences (and potentially the interface), the distinction between discovering and searching will continue to blur.

2. AI as an execution layer, not just a navigation tool

Agents can go beyond fetching pages to interpreting and acting on information. As agents get better at understanding preferences, interpreting intent, and executing workflows, we will see a much more efficient and personalised web experience. Another powerful motion that can be seen in this example is that AI search engines don’t need to start from scratch, but can be built on top of existing recommendation and personalisation systems to deliver powerful results without requiring massive upfront investment in data and compute.

How will this play out?

We see that this can become a reality on three different interface levels:

Device/ OS-level integration

The device level has the most amount of context and distribution advantage, which is why incumbents are investing heavily in this area. For example, Apple’s “Apple Intelligence” seeks to tightly embed AI into the user’s workflow at the OS level. So far, however, results have been mixed, opening the door to do at different levels.

AI-native browsers

Browsers like Strawberry and Deta.surf are building AI-first browsers that help users automate tasks, not just view pages.

Personal agents

Agents like Proxy operate across apps, interpret data, and complete multi-step tasks. This has the potential to replace not just browser sessions, but full workflows. So far, these specialised agents are by far the most capable.

User alignment / better monetisation models

While incumbent search players (and feeds) are incentivised to drive transactions and optimise for ad revenue, agent-based models can be more naturally aligned with the user, only recommending necessary purchases. We expect there to be a large group of users who prefer this over classical models and who would be willing to pay for power features, better automation capabilities, and deeper integrations.

Opportunity to become the gateway for the entire web traffic

We think that whoever controls this interface will become the new gateway for discovery, commerce, and productivity on the web. The shift is already underway –  and we believe some of the most valuable AI-native consumer companies of this decade will be built in this space.

3. AI SEO: A new stack for a new funnel

As consumer behaviour shifts toward AI-powered search, companies and marketers need to think about how to get discovered when users no longer “Google” but ask ChatGPT, Perplexity, or Claude.

A new generation of AI SEO or GEO (Generative Engine Optimisation) tools is emerging to help brands understand what users are asking, how LLMs respond, and optimise their presence across these models.

In the first wave of SEO, the size of outcomes was somewhat limited (e.g.. Semrush reached a $2B market cap), with much of the downstream value captured by agencies. AI-native tools now aim to also capture a much larger portion of the value chain, automating content creation analysis, prompt testing, and ranking strategies.

This unlocks an opportunity to build a new category of vertical SaaS and agent-enabled platforms designed for AI discoverability. As LLMs become the default discovery layer, brands will need continuous optimisation to ensure they’re part of the conversation. AI agents may well be the most capital-efficient way to get there.

Conclusion

AI is set to reshape search, discovery, and the broader interface layer of the internet. While incumbents have scale and data advantages, they also face deep structural challenges. We believe this creates space for a new wave of category-defining companies across vertical AI search, agent-powered navigation, and AI-native SEO.

If you’re a founder building in this space, we’d love to hear from you — feel free to reach out at [email protected].