How to Show Up in ChatGPT, Claude, and Google's AI Answers
The question I get most often this year is some version of: "How do I get my business to come up when someone asks ChatGPT?" It is a fair question, and it is being asked because customer behavior actually moved. Someone looking for a Korean dentist in LA increasingly types that into an assistant and reads one answer with two or three businesses named — instead of scrolling a page of links and deciding for themselves.
This is not a separate discipline from SEO, and anyone selling you a special "AI optimization" package is mostly reselling fundamentals. But there are a few decisions that genuinely control whether an assistant can cite you at all — and one of them is a single line in a file most business owners have never opened. That line decides whether ChatGPT, Claude, and Perplexity can cite you. Google's AI Overview is a separate surface, earned by ordinary Googlebot ranking work — and one concrete result from my own client work on that surface: ID Dental is listed first of three in Google's AI Overview for 한인치과 엘에이 (Korean dentist in LA), with the site cited as a source — checked live on September 2, 2026.
A lot of websites are quietly blocking the assistants
Every site has a file at /robots.txt that tells automated visitors what they may read. For years the standard advice — baked into WordPress plugins, agency checklists, and a lot of well-meaning blog posts — was to block AI crawlers to protect your content from training. For a publisher selling subscriptions, that logic holds. For a local business whose website exists to be found, it is self-defeating: a marketing site has nothing proprietary on it, and blocking those crawlers makes you uncitable in exactly the place customers are starting to ask for recommendations.
The important thing to understand is that these crawlers are not one group. They split into three, and they deserve different answers:
- Answer engines — ChatGPT-User, OAI-SearchBot, PerplexityBot, Perplexity-User, Claude-User, Claude-SearchBot. These fetch or index your pages to answer a question someone is asking right now, and they cite the source. These are the ones that name and link the source — the crawlers that can put your business in front of someone mid-decision.
- Training crawlers — GPTBot, ClaudeBot and their older names. These collect text to train future models. They send nobody today, but a model that learned your business name during training can recommend you later without searching at all. I allow these on my own site, for exactly that reason.
- Bulk scrapers — CCBot, Bytespider and similar. These pull whole pages into corpora that get redistributed, with no citation and no way to know where your text ends up. That is a different bargain from a named AI company crawling under its own name, and I decline it. I block these, and I would block them on your site too.
The Google-Extended decision everyone gets backwards
Google's crawler permissions have one token that causes more confusion than the rest combined: Google-Extended. Two things about it are worth knowing before you or your developer touch it.
First, blocking it does nothing for your Google Search ranking, and nothing to keep you out of AI Overviews. Those both run on regular Googlebot. People block Google-Extended believing they are opting out of Google's AI, then wonder why their competitor still appears in the AI Overview. All they actually did was remove themselves from Gemini.
Second, Google bundles two different things into that one token: grounding answers inside the Gemini app, and using your text for model training. You cannot accept one and refuse the other. So it is a genuine trade, and I made it deliberately on my own site — Google is the platform this business runs on, so being visible inside Gemini is worth the training use. That is a judgment call, not a rule. What matters is making it on purpose instead of inheriting a plugin's default.
Being allowed in is the floor, not the win
Unblocking the crawlers only makes you eligible. What decides whether an assistant actually names you is closer to plain editorial work than to anything technical, and it is the same work that earns a Google ranking.
An assistant answering "who is a good Korean dentist in LA" is looking for pages that state facts it can safely repeat and attribute. That means your services, your neighborhood, your hours, and your prices written as real text on your own site — not baked into a photo, not locked in a PDF menu, not living only on Instagram. It means structured data so a machine can read your address and hours without guessing. And it means being specific: "implant dentist in Koreatown" is citable, "quality you can trust" is not.
One more, cheap and worth doing: a file at /llms.txt — a plain-language summary of what the business is, what it costs, and where it operates. I keep one on this site. Be honest about its status, though: it is an emerging convention, not a ratified standard every engine promises to read. It costs an hour and it cannot hurt you, which is a fair reason to have one and a bad reason to pay someone much for it.
What nobody can honestly promise you
I will not tell you I can put your business into ChatGPT's answer for a given question, and you should be wary of anyone who does. Assistants do not publish rankings, the answers vary between users and change week to week, and the one dashboard that exists is Google's own: since August 31, 2026, Search Console's generative AI performance report shows impressions from AI Overviews and AI Mode — impressions only, no clicks, and nothing from ChatGPT, Claude or Perplexity. Anyone quoting you a guaranteed AI ranking is selling something they cannot measure.
What can be done is unglamorous and it compounds: open the right doors, write the facts down plainly, keep the site fast and readable, and keep publishing pages that answer the questions your customers actually ask. Opening the right doors is what makes you citable in ChatGPT and Perplexity at all; the rest is what earns the Google ranking that a first-of-three spot in an AI Overview sits on. That is the work behind a dental clinic showing up first there — no trick aimed at the AI, just a site built so the answer was easy to find and safe to repeat.
If you want to know what your site currently tells the assistants — including whether it is blocking them — I will look and tell you plainly, at no cost. It is a short check. After that, the care plan is where it stays true: I re-run the questions your customers actually ask and read the answers by hand, keep the crawler list current as these engines rename their bots, and keep publishing pages they can cite — month after month.
Free ConsultationFind this useful? See more of brian-yang.com in your Google Search and AI results.
Add to preferred sources on Google