Get cited in AI answers: what actually works

Nothing special is required: Google states that appearing in its AI-generated answers takes the same requirements as regular Search, namely an indexed, reachable page, with no new files or dedicated markup. It's one of the few certainties written down in a field where most of what circulates comes from vendors selling AI-visibility services, not from the companies that actually run those engines.

What Google says, in its own words

Google Search Central's page on AI features in Search is the most direct source available on this topic, because it doesn't come from a vendor selling rankings: it comes from whoever decides what to show. It states, word for word, that "there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary," that "you don't need to create new machine readable files, AI text files, or markup" to appear in those features, and that "there's also no special schema.org structured data that you need to add" (source: Google Search Central, AI features and your website, checked November 7, 2026).

What it does ask for is exactly what it asks for regular Search: a page that is indexed and eligible to show with a snippet, with crawling allowed in robots.txt, important content available as readable text, and the page reachable through internal links. A site that doesn't already show up in Search for the real reasons I listed in why isn't my website showing up on Google is already excluded from AI answers too.

The bots that decide what ends up in an answer versus what trains a model

Every AI company runs more than one crawler, each with a different job, and mixing them up means blocking the wrong thing. The distinction that matters for a business is between a crawler that gathers content to train a model and one that uses it to answer a search in real time.

OpenAI states that blocking GPTBot "indicates a site's content should not be used in training" its generative foundation models, but that each setting "is independent of the others": a site that blocks GPTBot can still appear in ChatGPT's search answers, governed by OAI-SearchBot, while one that blocks OAI-SearchBot will not show up in those answers even though it can remain a navigational link (source: OpenAI developer documentation on GPTBot, checked November 7, 2026). Perplexity follows the same logic from the opposite direction: it explicitly recommends allowing PerplexityBot "to ensure your site appears in search results," and states this crawler "is not used to crawl content for AI foundation models" (source: Perplexity's official guide to its bots, checked November 7, 2026). Google-Extended, finally, isn't a separate crawler but an extra permission layered on the one that already exists: it only covers training and grounding for Gemini models in Gemini Apps and Vertex AI, and Google states it does not touch a site's inclusion in Search or its ranking, AI Overviews included.

The proposal sold as a shortcut: llms.txt

Much of what circulates online about "getting read by AI" centers on a file called llms.txt: a markdown summary meant to sit at a site's root, proposed on September 3, 2024 by Jeremy Howard of Answer.AI (source: the specification at llmstxt.org, checked November 7, 2026). The document itself describes itself as "a proposal" open to community input, not a standard adopted by anyone.

Google says so directly, in its guidance on optimizing for AI in Search: "you don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search [...] as Google Search itself doesn't use them," adding that such a file "neither harms nor helps" ranking because it's simply ignored (source: Google Search Central's AI optimization guide, checked November 7, 2026). A different use does exist, but it's a separate matter: Chrome's Lighthouse tool added an experimental check for this file under its "Agentic Browsing" category, describing it as useful to AI agents that need to explore a site to complete a task, not to citation in search answers; the page itself calls it "optional at the moment" (source: Chrome Developers, the llms.txt audit, checked November 7, 2026).

FAQ markup isn't the shortcut it looks like

Another practice often sold as an AI-visibility lever is adding FAQPage structured data to a site's pages. Here the facts point in a clear direction: starting May 7, 2026, Google retired the visual FAQ rich result from Search, the expandable question-and-answer box that used to appear under a result (source: Google Search Central's page on FAQPage markup, checked November 7, 2026). Its AI features guidance, quoted above, never names this markup among the requirements to appear in AI Overviews or AI Mode. Anyone presenting it as a shortcut to AI answers is doing so without an official source behind it.

What you can actually control, in practice

Once you strip out what has no source, what's left is a short but concrete list. Allow crawling from the bots whose job is to get you into an answer (OAI-SearchBot, PerplexityBot, Googlebot), keeping that choice separate from training permission, which is a different decision. Keep important content as readable text, not only inside images or scripts. Link your pages to each other clearly, so a crawler can tell what connects to what. And if a site doesn't already appear in regular Search, fix that first: no extra file or markup substitutes for basic indexing.

What you can't control is the final choice: which source a generative engine decides to cite, out of everything it has crawled and indexed. No AI company publishes that criterion, and in the meantime the same phenomenon has a side effect worth knowing about, the clicks your site loses when Google itself answers in place of a link: I wrote about that in Google's AI answers and the clicks that stop arriving.

If you want an outside read on what's really keeping your site from being read, by people and by crawlers alike, there's the digital check-up. If the question is long-term visibility rather than a one-off fix, that's what the SEO and blog service covers. Otherwise get in touch and we'll look at where your site stands.