AI Search
AI search is retrieval that ends in a written answer instead of a list of links. A model reads a set of sources, synthesizes them, and states a conclusion in plain language, sometimes naming the sites it drew from. ChatGPT, Google's AI Overviews and AI Mode, Perplexity, and Gemini are the main systems people mean when they say "AI search" today, and between them they now reach an enormous share of everyday queries.
Why AI search matters
The scale is no longer niche. Google's AI Overviews pass 2 billion monthly users across 200-plus countries, Gemini has surpassed 1 billion monthly active users, and ChatGPT reported 900 million weekly active users in early 2026. When that many queries end in a synthesized answer rather than a results page, being named in the answer starts to matter more than where you sit on page one. The effect on click behavior is measurable: Pew Research found people clicked through to a traditional result in only 8% of visits when an AI summary appeared, versus 15% without one, and Ahrefs found the #1 organic result now sees roughly 58% lower click-through on queries that trigger an AI Overview. Zero-click searches, where the user never visits a website at all, have climbed to over 68% of US Google searches. None of this means traffic disappears entirely, but it does mean a growing share of your audience only ever sees whatever the AI decided to say about you.
How AI search works
Most of these systems combine a large language model with a retrieval step: before answering, the system pulls in current web pages, a search index, or a private knowledge base, and feeds those fragments to the model as context. The model then writes an answer grounded in that material, sometimes with citations. This is why AI search rewards clear, well-structured, factual writing over keyword-stuffed copy, the model is trying to extract a defensible claim, not match a query string. It's also why the process is not simply "good SEO with extra steps": Ahrefs found that only 6 to 8% of URLs cited by ChatGPT overlap with Google's top 10 for the same query, and about 80% of ChatGPT's cited URLs don't appear anywhere in Google's top 100. Ranking and being cited are governed by different selection logic.
Common mistakes
The most common mistake is assuming there's a special file or markup that unlocks AI visibility. Google has said directly that no schema, no llms.txt, and no AI-specific content are required for AI Overviews or AI Mode, and it explicitly warns against writing separate pages "for AI." The llms.txt idea in particular hasn't held up: Google's John Mueller confirmed no Search system reads it, and an Ahrefs analysis of roughly 137,000 sites that published one found about 97% saw zero measurable referral traffic tied to it. A second mistake is chasing links while ignoring mentions, Ahrefs found brand mention frequency across the web correlates with AI citation rate at about 0.664, roughly three times stronger than backlinks at 0.218. A third is trusting AI answers uncritically: the Columbia Journalism Review's Tow Center found AI search tools misidentified basic facts about a source article, headline, date, or URL, in more than 60% of tested responses, so treat any AI-generated claim about your own brand as something to verify, not something to assume is correct.
Example
Picture a small orthopedic clinic in Rotterdam. Someone asks an AI assistant which clinic nearby takes new patients without a long wait and handles sports injuries. The assistant doesn't return ten links, it names two clinics and gives a reason for each. One clinic is named because its own site states its specialties, current wait times, and insurance partners in plain text that's easy to lift into an answer. A nearby clinic with an equally strong track record is skipped, not because it's worse, but because that same information sits inside a scanned PDF brochure and a hero image, where the model can't read it. The named clinic gets the call. The other never finds out why the phone didn't ring.
Common questions
Is AI search replacing Google Search?
Not replacing it, running alongside it and increasingly on top of it. Google's own AI Overviews and AI Mode sit inside Google Search itself, while ChatGPT, Perplexity, and Gemini operate as separate destinations. What's changed is that a large and growing share of queries across all of these now end in a synthesized answer rather than a list of links to click through.
Does adding an llms.txt file or extra schema markup help me show up in AI search?
The evidence says no on its own. Google has stated no special markup or AI-specific files are required, and an Ahrefs study of sites that published llms.txt found about 97% got no measurable referral traffic from it. Schema can help machines parse your page correctly, but it isn't a proven independent citation factor. What correlates more strongly with being cited is being mentioned, accurately and often, across the web.