For over two decades, search meant one thing: Google. You typed in a few keywords, hit enter, and scrolled through a list of blue links (find out why they’re blue here). Entire industries were built on this premise — keyword stuffers, backlink brokers, on-page optimizers, and technical auditing. Online visibility was always defined by page rank, and the rules were clear: get indexed, climb the rankings, win the clicks, the attention and the purchases.

But today, search is starting to feel different.

If you ask Google where the nearest Italian restaurant is, you’ll still see a map dotted with pins. But if you type something more open-ended like “What’s the best vitamin for men?”, you’re greeted by an AI Overview at the top of the page — a generated synthesis of results, compiled for you so you don’t have to wade through ten different health blogs or magazines.

It’s subtle (since you’re still using Google), but it signals something bigger: search is shifting away from traditional results pages and toward AI-native experiences.

A shifting target: Page rank to reference

The core expectation for search has changed. Users no longer want a list of possible answers; they want the answer and more — contextual, synthesized, and decision-ready. That is what large language models (LLMs) like GPT-4o, Gemini, and Claude now provide: the ability to reason, remember, and refine responses across multi-turn conversations.

Instead of tweaking search queries with advanced search operators, users today engage in a natural dialogues. They ask follow-up queries. They can even push the model for depth. They seek not just information, but outcomes.

And the business model underneath this shift is just as transformative. Classic search engines monetized attention through ads. Most LLMs, in contrast, are subscription-driven, with far less incentive to surface third-party content unless it truly adds value. For those imagining a future where LLMs are filled with ads — yes, it is possible. But even then, the rules of the game will be different. Even then, what matters most won’t be ranking high on a results page; it will be whether your brand is referenced directly in the model’s output.

Put simply: in the LLM era, SEO evolves into Generative Engine Optimization (GEO).

The domino effect  

This evolution is shaking the foundations of the $80B+ SEO industry. With Apple announcing AI-native search engines like Perplexity and Claude will be built into Safari, Google’s distribution chokehold suddenly looks less secure. Perplexity itself has gone further, launching Comet, its own browser and making recent headlines with a $34.5B unsolicited all-cash offer for Google Chrome.

For brands, the ground is moving. The game is no longer about click-through rates — it’s about reference frequency. How often is your content cited in model-generated answers? How are you remembered, framed, and perceived in the AI layer?

As the founder of Vercel, Guillermo Rauch put it on X: “ChatGPT now refers 10% of new Vercel signups, which have also accelerated.” Referral traffic, once the domain of Google’s links, is now flowing through LLMs.

The implication is clear: brands must start asking not just “How do humans see us online?” but “How do models see us?”

GEO: Optimizing for models

Here’s the twist: optimizing for AI isn’t entirely foreign. If your brand already produces high-quality, authoritative content, builds on backlinks, and stays relevant to evolving intent — congratulations, you’re already doing much of what GEO demands.

The nuance lies in understanding how LLMs consume, filter, and cite data. Models don’t pull obscure sources from page 600 of Google. They prioritize authority, relevance, and trustworthiness — the same fundamentals of good SEO.

But there’s new work to be done. Teams will need to:

• Track how often their brand is mentioned in AI Overviews or LLM answers.

• Understand how they’re being described — not just product features like “waterproof” or “warm,” but whether their brand is remembered at all.

• Respond quickly to emerging mentions in model outputs.

In other words, GEO is less about gaming an algorithm and more about earning a place in the model’s memory. The best way to achieve this is still up for debate, and multiple schools of thought are emerging. Some tactics are relatively well understood — for example, ensuring your brand is mentioned in the source documents LLMs tend to cite. Others remain more speculative, such as whether today’s closed models like GPT-5 give preference to journalistic outlets over social media, or how open-sourced models responses shift with increased transparency in their training sets.  

Act III: The Agentic web

If Act I was Google SEO, and Act II is GEO in LLM interfaces, then Act III may already be emerging: search done by agents, not humans.

Imagine handing off a task to an AI agent in ChatGPT’s Agent mode or using Claude’s Computer Use: “Find me the best running shoes for flat feet under $200 and order them by Friday.” The agent will search, compare, and transact — often without you ever seeing a traditional search page.

But here’s the catch: agents don’t experience the web the way humans do. They don’t see beautifully designed hero sections or neatly ordered menus. They wrestle with sprawling DOM trees, clunky buttons, and drop-downs. Many rely on screenshots, augmented metadata, or API shortcuts just to get through a navigation flow.

The current internet or Web 2.0 (for the Web 3.0 folks) was built for human eyes, not agentic workflows. That will arguably change in the short-term as proposed concepts like Agentic Web Interface (AWI) advocates for website owners to upgrade their sites that allows them to maintain control over what they expose as tools for agents to use — but let’s leave that discussion for another day.

For businesses, the next search optimization challenge isn’t just making content easy for people to read, but making it easy for agents and sub-agents to act upon and achieve human outcomes. Visibility will mean being seamlessly referenced in an agent’s sequence of workflow — not just being seen, but being used to get you those pairs of shoes.

A new competitive frontier

According to venture capital firm, A16z, every technological wave reshapes marketing and advertising. In the 2000s, it was Google’s Adwords. In the 2010s, it was Facebook’s targeting engine. In 2025, it’s LLMs and agentic platforms. The competition now is to get into the model’s mind — to be the trusted, cited, remembered source when answers are generated and when tasks are executed.

Search isn’t dying. It’s evolving. From page rank, to answer rank, to agent workflows, we’re entering a new era where the future of brand visibility could well live inside the intelligence layer of the Internet.

And that changes everything.

Enjoyed the read? Let’s take it further.

Connect to unlock exclusive insights, smart AI tools, and real connections that spark action.

Schedule a chat to unlock the full experience