AI assistants can only cite what they can crawl and understand. Let their crawlers in, put your business facts in plain HTML and in LocalBusiness structured data, answer buyers' questions in the first sentence under clear headings, and keep the same facts on your profiles and listings. Nobody can guarantee a mention.
How assistants answer local questions
Ask an assistant “who picks up scrap metal near Dayton” and it has to find businesses, judge which are relevant, and explain why. Depending on the product, it draws on a search index, its own crawler’s copy of web pages, map and business data, or a mix. When it shows sources, it links to pages that stated the answer clearly.
The details differ by product and change often. The principles don’t: a page that can’t be crawled can’t be cited, and a fact that’s only in an image or a script-rendered widget is easy to miss.
Google’s AI features
Google’s guidance for its AI features (AI Overviews and AI Mode) is that there are no additional requirements beyond normal Search best practices. The page needs to be indexed and eligible to appear with a snippet. The same guidance lists checking that your Business Profile information is up to date, and says you don’t need special AI text files or extra schema markup to appear. In other words: good local SEO is the preparation.
Google also has a separate robots.txt token, Google-Extended, that controls whether content can be used for training Gemini models and for grounding in Gemini Apps and Vertex AI. Google says it doesn’t affect inclusion or ranking in Google Search.
Other assistants’ crawlers
Each major assistant documents its crawlers and how robots.txt controls them:
- OpenAI runs OAI-SearchBot (for search features in ChatGPT) and GPTBot (for training), among others.
- Anthropic documents ClaudeBot and related user agents.
- Perplexity documents PerplexityBot.
If your robots.txt blocks these, or blocks everything except Googlebot (some security plugins do this), those assistants can’t read your pages directly. Open yoursite.com/robots.txt and look for Disallow: / under any user-agent you didn’t mean to block.
Put the facts where machines can read them
Right-click your homepage, choose View Source, and search for your phone number. If it isn’t there, it’s being drawn by JavaScript or sits in an image. Assistants and crawlers vary in how much JavaScript they run; plain HTML always works.
The facts to have in text on your site:
- business name, address (if customers visit) and phone
- the services you offer, by name
- the areas you serve
- hours
- what makes you different in one plain sentence (what you accept, how fast you deliver, the sizes you stock)
Then state them again in LocalBusiness structured data. See our schema guide.
Answer questions the way buyers ask them
Assistants are asked questions, so pages that answer questions get quoted. Use headings shaped like the question (“What can go in a 20 yard dumpster?”) and answer it in the first sentence underneath. Then add the detail. This is also how featured snippets have long worked.
Don’t write fake Q&A or stuff pages with questions nobody asks. Use the ones your office staff actually hear on the phone.
Keep facts identical everywhere
When your website, Google Business Profile, Bing Places and directories disagree about your hours or phone, any system pulling from several sources has to pick one, and it may pick the old one. Consistency is one of the most useful things you can do for AI answers, and it’s the same work as NAP cleanup.
What about llms.txt?
llms.txt is a proposed convention: a plain-text Markdown file at the root of your site that summarises what’s there for language models. It isn’t a standard that any search engine requires, and no major assistant has said it ranks sites by it. It’s cheap to add and harmless. This site has one at /llms.txt if you want an example.
What nobody can promise
No one can guarantee that an assistant will mention your business. Anyone selling “guaranteed ChatGPT placement” is selling something they don’t control. What you can do is remove the reasons it wouldn’t: blocked crawlers, hidden facts, vague pages and conflicting listings. The free grader checks the on-page part in seconds.