How to get your business to appear in ChatGPT and in Google’s AI answers

More and more people ask ChatGPT, Gemini, Perplexity or Google itself "which company does this in Curitiba" and get a ready-made answer, sometimes with two or three names and links. The question small business owners bring us is direct: how do I make my company one of those names?

The honest answer starts with a sentence few agencies like to say: nobody controls what AI cites. What you can do is make your site accessible to the right bots, easy to understand and full of information worth citing. In this article we use reiko.top itself as a practical example: we show what is configured here, what each piece is for and, above all, what none of it guarantees.

There is no ChatGPT "sign-up": where AI gets its answers

There is no form to get your business into ChatGPT answers, or into Google’s AI Overviews or AI Mode. Google is explicit in its documentation for site owners: there are no additional requirements or special optimizations to appear in these features. To be eligible as a supporting link, a page must be indexed and eligible to show in Search with a snippet. On the map the rule is similar: according to Google, there is no way to request or pay for a better local ranking.

In simplified terms, AI answers come from three places:

• What the model learned in training. Public content collected up to a certain date. It is what the assistant "knows" without searching, and it is often outdated.
• A search run on the spot. When ChatGPT, Claude, Perplexity or Google search before answering, they query an index of pages and cite sources with links. This is where your site can get in today, if it is accessible and indexed.
• Pages opened at the user’s request. When someone pastes a link or asks the assistant to read a site, a bot visits that page at that moment.

In practice, the part you can influence is the second and third: being in the index the AI checks and having pages that answer the question well. That is why GEO (generative engine optimization, the name being used for "optimizing for AI engines") is, for the most part, well-done SEO with a few extra precautions.

The technical side that matters: bots, llms.txt, structured data and speed

robots.txt: who can enter your site

robots.txt is the file that tells bots what they may crawl. Each AI company uses different bots with different jobs, and blocking the wrong one removes you from answers without anyone noticing. According to each company’s official documentation:

The practical consequence: you can refuse training use of your content (GPTBot, ClaudeBot, Google-Extended) and still appear in AI search (OAI-SearchBot, Claude-SearchBot, PerplexityBot). They are separate decisions. OpenAI states that robots.txt changes take about 24 hours to take effect in its search.

llms.txt: a site summary for AI to read

llms.txt is a proposal published in September 2024 (llmstxt.org): a Markdown file at the site root with the company name, a summary and links to the pages that matter. The idea is to give an AI assistant a short, clean map, without menus, footers and scripts in the way. It is a proposal, not an official standard, and Google says plainly that you do not need to create AI text files to appear in AI Overviews. It costs little and helps agents that read it; it is not a ranking factor.

Structured data (schema.org)

These are code blocks that describe, in machine format, who the company is, where it is, what it sells and what each page is. They remove ambiguity: "Reiko" stops being just a word and becomes an organization with an address, phone and services. Google also notes there is no special schema for AI, and that marked-up content must match the visible text of the page. Markup that says one thing while the page shows another is worse than no markup.

Indexing and speed

If the page is not indexed, it does not appear in AI Overviews, full stop. A correct sitemap, error-free pages, the right canonical tags, content that loads without heavy JavaScript and a fast mobile site are still the foundation. Details are in our technical SEO 2026 guide.

The reiko.top example: what is configured and what it does not guarantee

Everything below can be checked live, and it is the same setup we apply in the SEO projects we manage:

• robots.txt with a single group (User-agent: *). It opens the whole site to all bots, AI bots included, and blocks only internal areas. We deliberately did not create a separate group for GPTBot: a bot with its own group ignores the general rules, and that once cancelled our blocks here. Anyone opening the file will see a Crawl-delay: 10 that is not in our file; the hosting company adds it on delivery. Google ignores that line; Anthropic says it respects it.
• /llms.txt and /llms-full.txt. The first summarizes the company (founding year, address, Brazilian company ID, contact, services, main pages); the second carries the content in long form.
• Structured data. Organization and ProfessionalService on every page, Service on service pages, FAQPage where there are FAQs, Article on each post and BreadcrumbList.
• RSS feed at /feed.xml, declared in the site header, so readers and aggregators find new posts.
• IndexNow pinged on every publication. The protocol is supported by Bing, Naver, Seznam, Yandex and Yep; Google is not on the list, so for Google the sitemap does the job.
• Same name, address and phone on the site, in the schema, in llms.txt and in the footer.

What this does NOT guarantee: that ChatGPT recommends Reiko when someone asks for a software house, that AI Overviews cites this post, or that any assistant repeats our data without errors. These pieces remove obstacles and keep correct information available. The choice of what to cite belongs to each company’s system, changes without notice and nobody outside controls it. Anyone selling a "guaranteed position in ChatGPT" is selling something they cannot deliver.

Content AI cites: a direct answer, your own data and authorship

When AI searches, it looks for a passage that answers the question clearly. Pages that ramble for three paragraphs before saying something useful lose to pages that answer in the first line. Google recommends unique content, not a copy of what already exists everywhere. In practice:

• Answer in the first sentence. If the page is about "how long does an air conditioner installation take", the first sentence answers that. Details come after.
• Your own data. A number only you have (how many jobs per month, which problem shows up most, what changed in your industry) is worth more than the generic list anyone can write.
• One question per section. Headings that are real customer questions, with the answer right below, make it easier to extract the right passage.
• Authorship and date. Who wrote it, with what experience, when it was updated. This helps readers and gives context to whoever evaluates the source.
• One page per service and per city. "We do everything across the region" answers nobody’s question. A page for each main service, stating what it is, for whom and where, does.

The same text AI can cite is the one that converts the visitor who arrives through the link. For that second part, our anatomy of a converting landing page is worth reading.

Consistent name, address and phone outside your site

AI does not look only at your site. It cross-checks what it finds in several places, and inconsistent data turns into wrong answers: an old phone number, the previous office address, a trade name different from the registered one. Google says local ranking depends on relevance, distance and prominence (how well known the business is), and consistency across sources helps all three.

• Google Business Profile fully filled in: correct category, services with descriptions, hours, real photos, replies to reviews.
• Directories and networks you are already on (Bing Places, Apple Maps, LinkedIn, industry associations, local guides): same name, address, phone and website on all of them.
• Third-party mentions: local press, partners, suppliers and customers who cite your business with a link. It is the off-site version of "prominence".
• A review every six months: search your company name on Google and ask two or three AI assistants what they know about it. Note what is wrong and fix it at the source.

The local SEO work we describe in the article on SEO in an election year is the same work that feeds AI answers to "near me" searches.

How to measure what comes from AI

Without measurement, any conversation about AI becomes opinion. Two free tools cover the essentials.

Google Search Console

According to Google, sites appearing in AI Overviews and AI Mode are counted in overall Search Console traffic, in the Performance report, under the "Web" search type. That documentation does not describe a separate AI filter: clicks and impressions from these features are added to regular search. What you can do is track long, question-style queries, which are typical of these features, and compare impressions and clicks month by month.

Google Analytics 4

When someone clicks a link cited by an assistant, the visit usually arrives with the assistant as its source. In GA4, under Traffic acquisition, filter session source by:

• chatgpt.com
• perplexity.ai
• gemini.google.com
• claude.ai and copilot.microsoft.com

Not every visit arrives labeled: apps and some browsers do not pass the source, and those visits fall into "Direct". So treat the number as a floor, not a total. And, as with any channel, what matters is leads (WhatsApp clicks, forms, phone calls), not visits.

What not to do

• Block every AI bot in a panic. Blocking OAI-SearchBot, Claude-SearchBot or PerplexityBot removes you from search answers. If training is the concern, block only the training bots.
• Create a separate robots.txt group without understanding the rule. "User-agent: GPTBot / Allow: /" makes that bot ignore every block in the general group, internal areas included.
• Publish mass AI-generated text without review. Generic, repetitive content with no real experience behind it is exactly what neither Google nor the assistants have any reason to cite.
• Mark up in schema what the page does not show. Reviews, prices or questions that exist only in the code go against Google’s guidelines.
• Hide text "for the AI to read". Invisible blocks of keywords or instructions for bots are manipulation and can sink the site in regular search.
• Pay for a "ChatGPT listing" or a guaranteed position. That product does not exist.

"There is no shortcut to being cited by AI. There is an accessible site, correct data and a better answer than the competitor’s. It is the same work that has always made a business show up on Google."

Conclusion

Appearing in ChatGPT and AI Overviews is not a product you buy or a sign-up you complete. It results from three things: the site being open to the right bots and indexed, information about the business being correct and consistent everywhere, and pages answering your customers’ questions better than the others.

Reiko provides monthly SEO for small and mid-sized businesses with this focus: technical work in order, content reviewed by people, local SEO and measurement in Search Console and GA4. If you want to talk about your case, see our websites and SEO service or reach us through the contact page.

Frequently asked questions

What is GEO (generative engine optimization)?

It is the name being used for the work of getting a site found and cited by AI answer engines such as ChatGPT, Perplexity and Google’s AI Overviews. In practice, the foundation is the same as SEO: a crawlable, indexed site, correct structured data, content that answers directly and off-site reputation.

Is SEO dead because of AI answers?

No. Google says that to appear in AI Overviews and AI Mode, a page must be indexed and eligible to show in Search, with no extra requirement. And assistants that search before answering depend on an index of pages. What changes is the format: more ready-made answers and fewer lists of links, which makes clear content and your own data even more valuable.

Should I block AI bots?

It depends on what you want to avoid. If the concern is your content being used for training, you can block only the training bots (GPTBot, ClaudeBot and the Google-Extended token) and keep the search bots allowed (OAI-SearchBot, Claude-SearchBot, PerplexityBot). Blocking everything removes the business from search answers. For most small businesses that want to be found, allowing them makes more sense.

How long does it take to appear in ChatGPT?

There is no timeline anyone can promise. It depends on the search bot having access to the site, on the pages being indexed, on how much better your content answers the question than competitors’ and on how each company’s system chooses sources, which changes without notice. OpenAI only states that robots.txt changes take about 24 hours to take effect in its search.

Do I need an llms.txt file?

It is not mandatory. It is a 2024 proposal, not an official standard, and Google states you do not need to create AI text files to appear in Search’s AI features. It can help agents that read it understand the site quickly, and it is cheap to maintain. We have one on reiko.top, without expecting miracles from it.