Short answer: When ChatGPT, Perplexity and Google's AI answers search the web for a question, they name companies from the pages they find. If their search crawlers can't read your site, or your site doesn't say plainly what you do, where and for whom, you won't be named. Google, OpenAI and Perplexity have each published what they need, and none of it involves paying for a special "AI SEO" file.
Some buyers now start with a question typed into an assistant: "Who can automate our invoice approvals in Pune?" or "Which CA firms in Bengaluru handle GST for exporters?" The assistant answers with three or four names. Owners want to know how those names get picked. The companies behind these assistants have published a fair amount about it.
What happens when someone asks
When the assistant searches, it doesn't answer from memory alone. It runs web searches, reads the pages that come back, and writes an answer that cites some of them.
Google describes this on its page about AI features in Search. AI Overviews and AI Mode may use a "query fan-out" technique, issuing several related searches across subtopics and data sources to build a response. Google says this lets it show a wider and more diverse set of links than a classic search.
So the practical question for an owner becomes: when the assistant goes looking, can it reach your pages, and do those pages answer the question?
The three gates, in the vendors' own words
Google (AI Overviews and AI Mode). Google says a page must be indexed and eligible to show in Google Search with a snippet. It adds that there are no additional requirements and no special optimisation needed. You don't need new machine-readable files, AI text files or markup, and there's no special schema.org structured data to add. Its list of good practice includes keeping important content in text, making sure structured data matches the visible page, and keeping your Business Profile information up to date.
ChatGPT. OpenAI's crawler page says OAI-SearchBot is the crawler that surfaces websites in ChatGPT's search features. Sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. A separate crawler, GPTBot, relates to model training. OpenAI says the two settings are independent, so you can allow search and refuse training.
Perplexity. Perplexity's crawler page says PerplexityBot surfaces and links websites in its search results and is not used to crawl content for AI foundation models. It recommends allowing PerplexityBot in robots.txt and allowing its published IP ranges. It also warns that Cloudflare's Bot Fight Mode can't be skipped with custom firewall rules, so if that mode is on, it can still challenge Perplexity's bots.
Where companies shut themselves out
Check for an accidental block first. Some sites add robots.txt rules or firewall settings to keep AI companies from training on their content. A blanket "block all AI bots" rule can also block the search crawlers listed above. The site then drops out of the answers its buyers read, and Google traffic can look normal the whole time, so nobody notices.
Other ways companies hide from assistants:
- The offer sits only in a PDF brochure or an image. Google asks for important content in text form. A pricing table pasted as a picture can't be quoted.
- The site never says the plain facts. Which city, which services, what size of company you serve, and roughly what it costs. If a buyer asks for "a firm in Hyderabad that does X for 50-person companies" and none of your pages say those words, the assistant has nothing to match.
- Different facts in different places. Your website, Google Business Profile and directory listings describe you three different ways. The assistant may read any of them.
- Paying for shortcuts. If someone sells you an "AI visibility file" as the fix, compare it with Google's own statement that no such file is needed for its AI features.
Your pages and other people's pages both count
Because the assistant searches across sources, it reads more than your website. Directory listings, review sites, "top firms in Bengaluru" articles and forum threads can all come back in those searches. That means the way other sites describe you matters, and an outdated listing with an old price or old services can be the version the assistant repeats.
You can't control those pages directly, but you can keep your own listings current and consistent, and ask happy clients for honest reviews where your buyers look.
A 30-minute check
- Open yoursite.com/robots.txt. Look for lines naming OAI-SearchBot, PerplexityBot or Googlebot with "Disallow", or a rule that blocks every bot. If you want to keep AI training out, block GPTBot only, as OpenAI describes, and leave the search crawlers allowed.
- Ask whoever runs your hosting or Cloudflare whether bot protection is on, and whether it challenges these crawlers.
- Search Google for your company name and check that your main pages show with a normal text snippet.
- Ask the assistants your buyer's question. Type the exact question a customer would ask into ChatGPT, Perplexity and Gemini, with your city and service. Note which companies get named and which pages are linked.
- Fix what you control. Put your city, services, who you serve and a starting price in plain text on your site. Update your Google Business Profile and the listings the assistants linked to.
Then repeat step 4 once a month. OpenAI and Perplexity both say robots.txt changes can take up to about a day to register, but new pages and updated listings may take longer to be found and quoted.
Arthat AI
Working on something similar at your company?
Arthat AI builds these automations for Indian businesses every week. If you want to talk through whether this fits your situation, we'll give you a straight answer — even if the answer is "don't build it."
