What you’ll be able to do

By the end of this you’ll know what actually makes an AI assistant mention your name, quote your page, or link to your site when someone asks it a question in your field. Not guesswork, not vibes, just the mechanics: how ChatGPT, Claude, Gemini and Copilot find sources, what makes a source easy to grab, and how to test whether you’re showing up at all. You’ll also come away knowing what you can’t control, because there’s a fair bit of that too.

Before you start

Right, first thing to get straight: this isn’t SEO with a new coat of paint. Google ranks pages. AI assistants do something a bit different, they read a question, decide whether to search the web at all, pull back a handful of pages if they do, and then write an answer using bits of what they found, sometimes with a link, sometimes without.

That last bit matters a lot. An AI can mention you by name without linking to you. It can quote your idea without saying where it came from. It can also link to you clearly, with your name and your URL sitting right there in the answer. All three of these count as different outcomes and you should know which one you’re actually chasing.

You also need to know that “AI training” and “AI search” are two separate things, and people mix them up constantly. Training is when a company builds a model, feeding it huge amounts of text months or years before you ever chat with it. Search retrieval is different, it’s when an assistant like ChatGPT or Perplexity goes and fetches a live page right now, in response to your specific question, and uses it to write the answer. You can let a search crawler in and keep a training crawler out, or the other way round. They’re controlled separately in robots.txt, which we’ll get to.

Last thing before you start: nobody can promise you a citation. Not me, not a tool, not a consultant charging you a fortune. What you can do is remove the reasons an AI would skip you, and make your page an easy, obvious thing to grab when it’s building an answer.

Get set up

Start with the boring stuff, because the boring stuff is what actually works.

Write a page that answers one question plainly, with the answer near the top. Not three paragraphs of throat-clearing before you get to the point, actual AI systems (and humans) want the answer fast. If you’re a dog groomer in Leeds, have a page that says clearly, near the top, what you do, where, and for how much. Date your facts. “As of 2026, I’ve been trading for six years” is more useful to a machine than “I’m an experienced groomer”, because the machine can check a date against a claim.

Check your robots.txt file lets the crawlers in that you actually want. The real ones worth knowing are GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot (Perplexity) and Google-Extended (Google’s AI training signal). If you’ve never looked at your robots.txt, it usually lives at yoursite.com/robots.txt, and you can open it in a browser right now to see what’s allowed and blocked. If in doubt, don’t block anything you want AI systems to see.

Add schema.org markup if you can, this is structured data that tells machines “this is a Person” or “this is an Organization” or “this is a Product”, in a format they can parse without guessing. It’s not glamorous but it removes ambiguity.

Keep a sitemap. Old advice, still true.

If you’ve got no site at all, or your existing site is a mess of Flash-era HTML from 2009 (no judgement, we’ve all got one somewhere), this is where I’ll point you at something we built: whatisgeo.app. It’s free, plain-English, and it walks through the whole query-to-answer pipeline: how a search-backed AI answer actually happens, which of the advice floating around online is a genuine platform requirement versus someone’s opinion, how to let the right crawlers in without accidentally blocking everyone, a 30-day starter plan, and a fixed set of prompts to measure your own progress. If you want the depth behind this guide, that’s where it lives.

If you need a citable page fast and don’t have a site to fix, there’s hilyt (hilyt.it). It builds you an open, ungated profile at a stable URL like i.hilyt.it/yourname, with proper schema.org markup baked in, a plain-text llm.txt card, dated claims each backed by a source link, and an FAQ section. It costs £1, once. The whole idea is giving AI systems something clean and unambiguous to point at when they’re deciding who to cite, which matters more than people think when your existing web presence is scattered across three old platforms and a LinkedIn bio.

And if you want to check where you stand before doing any of this, visibilitytoai.com has free tools worth running first: an AI readiness check covering six things (crawler access, visible text, structured data, llms.txt, indexing signals), a dedicated robots.txt check for AI crawlers specifically, a browser-based llms.txt generator, and a free single-page citation check. There’s also a paid Snapshot (£89, one-off) that asks ChatGPT, Claude, Gemini and Perplexity fifteen approved buyer-style questions and reports back whether you were mentioned and whether you were cited, as two separate things, which is a distinction worth paying attention to.

llms.txt itself, worth a quick word, is a fairly new idea, a plain-text file describing your site for AI systems to read. It’s optional. It’s unproven. Add it once your fundamentals (clear pages, open robots.txt, schema, sitemap) are solid, not instead of them.

Try it yourself

Open ChatGPT, Claude or Gemini and try these. Swap in your own name, business or topic.

Who is [your name], and what do they do? Please cite your sources.
What are the best [your service, e.g. "dog groomers"] in [your area]? List your sources for each recommendation.
I'm researching [your topic]. What are the most useful sources you'd point me to, and why?

The first one tests whether you exist to the model at all, in a citable way. The second tests whether you show up in a competitive, “best of” style question, which is the one most small businesses actually care about. The third tests whether your work is seen as a genuinely useful reference in your field, separate from any commercial angle.

Check the result

Read the answer properly, don’t skim it. There are three outcomes and they’re not the same thing.

You might be mentioned by name with no link at all. That’s something, but it’s thin, there’s no route back to you.

You might be cited with an actual link, your name and your URL both present. That’s the good outcome, screenshot it.

You might not appear at all. Also useful information, it tells you where you stand right now.

Run the same three prompts again in a day or two, ideally in a fresh chat each time rather than continuing an old one, because AI assistants can behave slightly differently between sessions and you want a clean read, not a carried-over answer. Try the same prompts across ChatGPT, Claude and Gemini separately, they don’t all pull from the same sources or reach the same conclusions, so a “no” in one isn’t a “no” everywhere.

If you used the visibilitytoai.com free citation check, that gives you a single-page read specifically on citation, worth comparing against your manual prompt results to see if they agree.

If it doesn’t work

Most of the time it’s one of a few things. Your page might genuinely be too thin, an AI can’t cite detail that isn’t there. Your robots.txt might be blocking the very crawler you’re hoping will find you, worth checking that file again properly rather than assuming it’s fine. Or you’re simply new, and it takes time for any of this to filter through, there’s no shortcut round that.

If none of those fit, the problem might not be your site at all, so it helps to see what AI is picking up about you elsewhere. That means looking at the coverage around you, and the guide on finding news on companies shows you how to follow the companies you care about and the news that gathers around them. Then you can judge whether the gap is yours.

Now the bit that matters more: never hand an AI assistant anything private to “help it understand you better”. Don’t paste in bank details, passwords, medical records, unpublished client data, or anyone else’s personal information, even inside a chat that feels private. Chats aren’t a confidential filing cabinet. If you wouldn’t post it publicly, don’t type it into a prompt.

And some sources stay invisible to AI by design, not accident. Content behind a paywall, content requiring a login, pages blocked deliberately by robots.txt, all invisible on purpose, because the owner chose that. That’s a legitimate choice. If that’s you, being invisible to AI citation is simply the tradeoff you’ve made, and it’s worth knowing that consciously rather than being confused by it later.

Questions people ask

Can I guarantee that ChatGPT will cite my website?

No. Nobody can promise you a citation, not a tool and not a consultant charging a fortune. What you can do is remove the reasons an AI would skip you, and make your page an easy, obvious thing to grab when it's building an answer.

How do I check if I appear in AI answers?

Open ChatGPT, Claude or Gemini and ask who you are and what you do, asking it to cite sources, plus a "best [service] in [area]" question and a "useful sources on [topic]" question. Run them again in a day or two in a fresh chat, and try each assistant separately, as they don't all pull from the same sources.

Is getting cited by AI free, or does it cost money?

The fundamentals are free: clear pages, open robots.txt, schema, sitemap. whatisgeo.app is free and visibilitytoai.com has free checks; hilyt (hilyt.it) costs £1 once for a citable profile, and visibilitytoai.com's Snapshot is £89, one-off.

Do I need an llms.txt file to be cited by AI?

llms.txt is a plain-text file describing your site for AI systems to read. It's optional and unproven. Add it once your fundamentals (clear pages, open robots.txt, schema, sitemap) are solid, not instead of them.

What shouldn't I tell an AI assistant about myself?

Never hand an AI assistant anything private to "help it understand you better". Don't paste in bank details, passwords, medical records, unpublished client data or anyone else's personal information. If you wouldn't post it publicly, don't type it into a prompt.

Keep exploring

If you run a business rather than a personal profile, have a look at “How to get AI to recommend my business”, it covers similar ground but tuned for commercial recommendation questions rather than pure citation. And if all this talk of crawlers and schema has left you wanting the fundamentals, “How to get AI to teach me about SEO” is a good next stop.

Sources and review notes

Review date: 15 September 2026 — every source above was opened and checked against the text on that date.

Disclosure: whatisgeo.app, hilyt and visibilitytoai are our products — this site is made by the same studio, Adapt Progress Evolve. We have tried to compare fairly, and the alternatives named are real and worth a look.