# How to show up in AI search: becoming a source in ChatGPT, Gemini and Perplexity > To show up in AI search, your site must first be indexed by search engines, allow AI search crawlers in robots.txt, and publish content that answers each question in its first sentence. Question-shaped headings, first-hand information, visible authors and dates, structured data and llms.txt all help. Nobody can guarantee a citation, but these steps improve the odds. - URL: https://radkod.com/en/blog/show-up-in-ai-search - Publisher: RadKod - Published: 2026-09-03 ## In short - AI search visibility sits on top of classic SEO. A page that is not indexed is rarely cited. - AI tools quote sentences, so each section should open with a direct answer. - First-hand information, a named author and a current date set a page apart. - Allow AI search crawlers in robots.txt. Training crawlers are a separate decision. - llms.txt is cheap to add, but no platform promises to use it. ## What does it mean to show up in AI search? To show up in AI search means your site is cited as a source when ChatGPT, Gemini, Perplexity or Google's AI Overviews answer a question. The user reads a written answer instead of ten blue links, with a handful of source links attached. The goal is to be one of them. People call this [GEO](/en/glossary/generative-engine-optimization) (generative engine optimisation) or AEO (answer engine optimisation). The names are new. Most of the groundwork is not. ## How do AI tools choose their sources? AI tools mostly pull candidate pages from a search index, then quote the ones that answer the question most clearly. Which index they use and how they rank it varies by platform and is not fully disclosed. Some things are documented. Google's AI Overviews are part of Google Search and use the same index; Google says no special optimisation is needed beyond normal SEO. OpenAI uses OAI-SearchBot to crawl for ChatGPT search. Perplexity builds its index with PerplexityBot. When a user asks the assistant to open a specific link, separate agents such as ChatGPT-User or Perplexity-User fetch it. The first conclusion is simple: if your page is not indexed, it is unlikely to be cited. Start with the basics in [what makes an SEO friendly website](/en/blog/seo-friendly-website). ## How should content be written to get quoted? Content gets quoted when the heading asks the question and the first sentence answers it. AI tools lift one or two sentences, not the whole page, so that sentence must stand on its own. - **Question headings:** not "Pricing" but "What determines website cost?", close to how people prompt. - **Answer first:** the sentence under the heading gives the answer. Context and detail follow. - **Self-contained paragraphs:** "as mentioned above" makes no sense once quoted. - **Concrete facts:** numbers, durations, thresholds and steps instead of "it should be fast". - **Lists and tables:** comparisons and steps are easier to parse than prose. - **A summary up top:** a few lines that show what the page answers. ## Why does first-hand information matter so much? First-hand information matters because an AI tool can find generic facts on a hundred sites, but your experience only on yours. Your process, your measurements, the problems you see often and the questions clients actually ask are what make a page worth citing. We write, for example, that our quotes fit on one page, that the definition of done is signed on day one and that we fix bugs free for 30 days after delivery. Nobody else can publish that. Show who wrote the page and when it was last updated. A named author, a company address and contact details signal that a real person stands behind the content. Google's guidance on helpful content stresses experience, expertise and trust. ## Does structured data help with AI search? Structured data helps indirectly: it states what the page is, who wrote it and when, in a form machines do not misread. How AI platforms use it is not fully documented, but search indexes read it. Use Article for posts, Organization for the company and FAQPage for questions, and keep the markup identical to what the page shows. ## What is llms.txt, and do you need it? [llms.txt](/en/glossary/llms-txt) is a Markdown file at the site root that gives language models a short summary of the site and a list of key pages. It was proposed in 2024 and is not an official standard. To be honest, no major AI search platform has said it uses llms.txt for ranking, and Google says no special file is needed for its AI features. It is cheap, though, and some coding tools and agents do read it. We add it, and we do not promise anyone results because of it. ## Which crawlers should you allow? If you want to show up in AI search, allow the crawlers that fetch pages for search answers. Crawlers that collect content for model training are a separate decision, and blocking them usually does not affect search visibility. - **OAI-SearchBot:** crawls for ChatGPT search. Block it and you may not appear in ChatGPT search answers. - **ChatGPT-User:** fetches a page when a user asks ChatGPT to open it. - **GPTBot:** OpenAI's training crawler, controlled separately from the search crawler. - **PerplexityBot and Perplexity-User:** Perplexity's index crawler and its user-triggered fetcher. - **Claude-SearchBot, Claude-User, ClaudeBot:** Anthropic's search, user and training agents. - **Google-Extended:** a robots.txt token, not a crawler. It controls use of your content for Gemini models. Google states it does not affect Search inclusion or ranking. AI Overviews follow Googlebot rules. A common mistake is pasting a ready-made "block all AI bots" list that also blocks the search agents. Read each platform's documentation and decide bot by bot. ![Classic SEO next to being cited in AI answers (GEO), with the foundation they share.](https://radkod.com/images/content/seo-ve-geo.svg) *Speed, structured data, clear answers, author and date, llms.txt: they serve both.* **Classic SEO compared with AI search** | Topic | Classic SEO | AI search (GEO) | | --- | --- | --- | | Result | Ranked list of links | Written answer with a few source links | | Goal | Rank high and get the click | Be cited as a source in the answer | | Unit | The page | A sentence or paragraph inside the page | | Content style | Page that covers the topic | Question headings, answer first, self-contained paragraphs | | Crawlers | Googlebot, Bingbot | Plus OAI-SearchBot, PerplexityBot, Claude-SearchBot | | Measurement | Search Console, rankings, clicks | Limited. Referral visits and crawler logs | | Shared base | Speed, indexing, structured data, trust, helpful content | The same base | ## How does the RadKod site apply this? We built everything in this guide into our own site, running automatically. We share it as an example, not as proof of results. - **llms.txt and llms-full.txt:** a site summary and key content list at the root, plus the full Markdown text of included content in one file. Both update when we publish. - **Markdown per page:** add .md to any post, service or glossary URL and you get a plain Markdown version that points to the original as canonical, so it does not create duplicate content. - **Crawler policy:** robots.txt groups crawlers into search, AI search and AI training, each switched on or off from the admin panel. - **Crawler log:** we record which bot visited which page and when, so we see AI crawler activity instead of guessing. - **Structured data:** each page carries organisation, article, breadcrumb and FAQ data in one graph. We set up the same foundation on client sites. See our [web design](/en/services/web-design) service. ## Steps to show up in AI search 1. **Confirm you are indexed** — Check key pages in Google Search Console and Bing Webmaster Tools. 2. **Read your robots.txt** — Make sure OAI-SearchBot, PerplexityBot and Claude-SearchBot are not blocked. Decide on training bots separately. 3. **Collect real client questions** — Each question from calls, emails and meetings can become a heading or a post. 4. **Rewrite headings as questions** — Make the first sentence of every section the answer, starting with your most important pages. 5. **Add first-hand detail** — Your process, measurements and examples, with a visible author and update date. 6. **Add structured data and llms.txt** — Generate both from code so they stay current. 7. **Ask and monitor** — Ask ChatGPT, Gemini and Perplexity your clients' questions regularly, see who gets cited, and watch crawler visits in your logs. ## Can anyone guarantee AI search visibility? No. Nobody can guarantee a citation in AI answers. The same question asked twice can cite different sources, and platforms change their methods without notice. Be wary of any offer to put you "first in ChatGPT". The good news is that nothing here is AI-only. A fast page with a clear answer and a named author also does well in classic search. For the speed side, read [site speed and Core Web Vitals](/en/blog/site-speed-core-web-vitals). For the pages a company site should have, see [pages every company website needs](/en/blog/pages-every-company-website-needs). ## Frequently asked questions ### Is GEO the same as SEO? Not quite, but they overlap heavily. SEO aims for a ranked link, GEO for a citation in a written answer. Both rest on speed, indexing, trust and content that answers the question. ### Do I need to do anything special for Google AI Overviews? Google says no. AI Overviews are part of Search and normal SEO best practices apply. The page must be indexed and its text crawlable. ### If I block GPTBot, will I disappear from ChatGPT? According to OpenAI, GPTBot is for training and OAI-SearchBot is for ChatGPT search. They are controlled separately, so you can block one and allow the other. ### Does blocking Google-Extended hurt my Google rankings? Google says it does not. Google-Extended controls use of your content for Gemini models and has no effect on Search ranking. ### Will llms.txt get me into AI answers? There is no guarantee. It is not an official standard and major search platforms have not said they rank with it. It is harmless to add, but do not expect results from it alone. ### Can a small company site be cited? Yes. For narrow, specific questions the clearest answer tends to win over the biggest site. First-hand content on local services, process and pricing factors is a real advantage. ### How do I measure traffic from AI tools? Look for referrers such as chatgpt.com, perplexity.ai and gemini.google.com in your analytics, and check server logs for AI crawler visits. To see whether you are named in answers, ask the questions yourself. ## Sources - [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features), Google Search Central - [Overview of Google crawlers and fetchers](https://developers.google.com/search/docs/crawling-indexing/overview-google-crawlers), Google Search Central - [Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content), Google Search Central - [Overview of OpenAI Crawlers](https://platform.openai.com/docs/bots), OpenAI - [Perplexity Crawlers](https://docs.perplexity.ai/guides/bots), Perplexity - [The /llms.txt file](https://llmstxt.org/), llmstxt.org - [Schema.org](https://schema.org/), Schema.org **Get your site ready for AI search** [Get a quote](https://radkod.com/en/get-a-quote)