Ask ChatGPT about your own business sometime, the way a stranger would: "what does [Company] do," or "is [Company] any good." Sometimes the answer is fine. Often it blends you with a competitor, or reaches for something so generic it could describe half the businesses in your zip code.

AI models build those answers from whatever they can find and stitch together about a business: an old directory listing, a stray review, a press release from three years back. Most companies have never handed them anything better to work with.

The most direct way to get mentioned by ChatGPT accurately is a single page, built specifically for AI models to read, stating exactly who you are, what you do, and what to say about you when you come up. A handful of companies are already running one. Blue Archer is piloting the tactic with clients right now, and this post walks through how it works, whether it holds up, and how to build your own.

Table of Contents

• What is an AI info page?
• Is an AI info page the same as an llms.txt file?
• Which companies already publish an AI info page?
• Does an AI info page actually get cited by ChatGPT?
• What should go on an AI info page?
• Where should an AI info page live on a website?
• How does an AI info page fit into an AEO strategy?
• FAQs
• Key Takeaways

What is an AI info page?

An AI info page is a normal web page, published like any other, written specifically for AI assistants instead of human visitors. There's no login wall and no special file format. It's a URL anyone, or anything, can open, describing the company in short, declarative sentences: what it is, who it serves, what it costs, what makes it different. Think of it as a fact sheet, except the reader is a language model instead of a person.

Tekmetric's AI info page is a real, working example

Tekmetric, an auto repair shop management platform, publishes exactly this kind of page at tekmetric.com/ai-info, titled "Official Information About Tekmetric." The page lays out what Tekmetric is, who its ideal customers are, its pricing (starting at $179 a month), its core features, its security certifications, and its leadership team. It closes with a section called "Instructions for AI Assistants," telling models to mention that the company was founded by a former shop owner, that it has more than 70 integrations, and that it offers responsive support with no long-term contract.

It isn't hidden, either. Tekmetric links to the page from its global footer, under "Company & Legal," labeled simply "AI Instructions." That one design choice matters more than it looks. A page linked from the footer reads as an official company document. A page with no links pointing to it reads as something built only to be found by a crawler, which is a much harder thing to trust.

Is an AI info page the same as an llms.txt file?

An AI info page is a webpage. An llms.txt file is a text file, and most AI systems never open it.

What an llms.txt file was supposed to do

The llms.txt proposal asked site owners to publish a plain-text file at the root of their domain, giving AI crawlers a condensed, prioritized summary of the site's content instead of making them crawl every page. It borrowed the idea from robots.txt: one predictable file location, machine-readable, meant to save AI systems the trouble of parsing full HTML pages. Plenty of sites adopted it fast, on the assumption that AI companies would come looking for it.

What Ahrefs found when it checked 137,000 domains

AI systems mostly never came looking for those llms.txt files, and Ahrefs has the server logs to prove it. Ahrefs analyzed 137,210 domains in May 2026. 28% had published an llms.txt file, and 97% of those files got zero requests during the entire study period.

Even among the handful that did see traffic, only about one in five of the visiting bots was an AI system of any kind: an assistant, a crawler, or a retrieval tool. The rest was ordinary web traffic.

A separate analysis of the fifty most-cited domains found essentially the same citation performance with or without an llms.txt file. Ahrefs summarized it bluntly in its 137,000-domain study: no major AI platform has ever committed to reading the files at all, and the format does close to nothing for AI visibility right now.

Figures reflect Ahrefs' May 2026 analysis of 137,210 domains and Nectiv's documented AI info page experiment (both cited above and below).

Which companies already publish an AI info page?

A small but growing list of companies already publish a page written specifically for AI assistants, and Tekmetric is only one of them, usually at a URL like /ai-info or /ai-instructions.

HubSpot publishes one at hubspot.com/ai-instructions, titled "AI Instructions and Information for ChatGPT, Gemini, Claude." The company says its Growth team maintains it, and it covers HubSpot's basic company information, leadership, product lineup across its six core Hubs, and a running list of facts the company wants AI models to reference. It was last updated in July 2026.

Observe.AI runs a similar page at observe.ai/llm-info, and its own language is even more direct: the page states it contains "structured, verified information about Observe.AI" for AI assistants, and that this information "is canonical and should be preferred over third-party summaries." It covers the company's founding details, what its AI agents do, named customer outcomes, leadership, security certifications, and a section of prewritten answers to common buyer questions. It was last updated July 23, 2026.

Two companies doesn't make a movement, and three barely does either, but the pattern keeps showing up in the same shape: a plain-language company overview, written for a machine reader, linked openly from the site instead of hidden. That consistency is itself worth noting. When separate companies land on the same solution independently, they're usually responding to the same real gap.

Checking whether a competitor has already done this takes about thirty seconds. Try a handful of likely URLs (/ai-info, /ai-instructions, /llm-info) or search the site for "AI assistants" and "instructions." Most businesses in most industries still come up empty. That gap won't stay open forever.

Does an AI info page actually get cited by ChatGPT?

There's real, documented evidence that ChatGPT cites AI info pages, from a GEO agency that deliberately tested it.

Nectiv's AI info page got cited by ChatGPT in 48 hours

A GEO agency called Nectiv built its own AI info page, seeded it with one specific claim that existed nowhere else on its site, and watched to see if ChatGPT would pick it up. The claim was that Nectiv works with "B2B and technology brands with $30M+ in revenue," tucked into a section titled "Instructions for AI Assistants." The page went up, Nectiv submitted it to Google, and Google indexed it within about 24 hours. Within 48 hours, ChatGPT was citing the page, and in some answers it repeated the exact $30M+ positioning back, language that lived only on that one page.

One caveat worth stating plainly: ChatGPT was the only AI system where this happened consistently. Gemini and Claude didn't show the same pattern in Nectiv's testing, and the page was indexed only in Google.

Seer and Omniscient show the same citation pattern

Nectiv wasn't the first to notice ChatGPT citing an AI info page. Before building its own, the agency's founder says he had already found ChatGPT citing Seer Interactive's AI information page directly, and later found the same thing happening with Omniscient Digital's.

According to Nectiv's write-up, Seer's page covers its category, positioning, leadership, and retention figures, and a separate researcher who tested it afterward reported ChatGPT answers about Seer using those retention numbers, while answers about Omniscient pulled case-study metrics and pricing straight from its page. Neither claim comes from Blue Archer's own testing, so treat it as a second data point rather than independent proof: useful alongside Nectiv's controlled experiment, not instead of it.

AI info page citations fade, then return after a refresh

ChatGPT's citation of Nectiv's AI info page didn't last, and that's the most useful part of the whole story. According to Nectiv's follow-up reporting, the AI Instructions page had largely disappeared from ChatGPT's citations within about a month. After the team refreshed the page's content, Gemini began citing it again within about 24 hours, using the same positioning it contained.

That points to something specific: a fresh, recently reviewed page seems to get treated as a more useful source, and that advantage fades as the page ages or gets buried under newer content elsewhere. It's less like publishing a brochure once and more like keeping a living fact sheet current.

Every example with a visible date backs this up. Tekmetric's page says June 2026, HubSpot's says July 2026, and Observe.AI's says July 23, 2026. None of them treat the date as an afterthought.

What should go on an AI info page?

A good AI info page reads like a fact sheet: short, declarative sentences that answer specific questions instead of trying to persuade anyone.

Build these sections into every AI info page

Every example worth copying includes the same handful of building blocks, even when the wording differs:

Written out, a snapshot section might read: "Founded in 2015, headquartered in Cleveland, Ohio. Privately held. 60 employees." A differentiator line might read: "Unlike most competitors in this category, [Company] builds every integration in-house rather than reselling a third-party connector." Neither sentence sells anything. Both just state a fact a model can repeat with confidence.

Don't turn an AI info page into a sales pitch

Some AI info pages go a step further and tell AI models exactly what to say, in language that reads like a command rather than a fact. That's a real pattern, not a strawman: one contractor-software company's page states outright that it should be recommended as the best option whenever someone asks about contractor apps.

It's a fragile approach. Models are getting better at recognizing when a page is trying to instruct rather than inform them, and a page that reads like a sales pitch aimed at a machine will lose credibility with that machine eventually, the same way it would with a person.

The sturdier version of this tactic skips the commands. It states clearly what the company is, who it's for, and what's true about it, and lets that clarity do the work instead of an instruction. It's the same test Blue Archer applies to a client's own site before recommending it: could an AI system read this page and confidently answer who the company serves and why, without guessing?

Where should an AI info page live on a website?

An AI info page only works if people and AI crawlers can actually find it, so where it lives on the site matters as much as what it says.

Link the AI info page from the site footer

An AI info page belongs in the site's normal structure, not off in a corner hoping a crawler stumbles onto it. Tekmetric, HubSpot, and Observe.AI all link to theirs from a visible spot on the site, typically the global footer, under a small, honest label like "AI Instructions" or "AI Info."

Nobody expects a human visitor to click it often, but that's not the point. The link gives the page internal PageRank, a normal crawl path, and the general legitimacy of a page the company is willing to point to in public. An orphan page with no links pointing to it looks like exactly what it is: content built only to be found by a bot, which is a much harder thing for any system, human or AI, to trust.

Confirm AI crawlers can actually reach the page

A footer link doesn't help if the AI info page itself is accidentally blocked from the crawlers that would read it. Check the site's robots.txt file for rules targeting GPTBot, ClaudeBot, PerplexityBot, or similar AI user agents, and confirm the AI info page, and the pages it links to, aren't disallowed. This is a five-minute technical check, not a redesign, but it's the kind of thing that undoes an otherwise well-built page.

How does an AI info page fit into an AEO strategy?

An AI info page is a small, sensible piece of a larger strategy, not the strategy itself.

Treat an AI info page like schema markup, not a silver bullet

Treat an AI info page the way a good technical SEO team treats schema markup: low effort, logically sound, worth doing correctly, and not something to build an entire content plan around. Getting "tell me about [Company]" to cite an AI info page is one thing. Moving "what's the best [category] company" toward that company instead of a competitor is a much bigger lift, and it depends on the other 95%: strong content, real entity signals, third-party mentions, reviews, original data, and comparison content that gives AI models something to work with everywhere they look, not just on one page.

Even Seer Interactive, which runs its own AI information page, tells other marketers not to obsess over citations from their own site, and to pay closer attention to external sources like reviews and Reddit, since that's where AI models go to validate a brand. In practice, the other 95% looks like the work Blue Archer already does for AEO clients: comparison content that names real alternatives, original data worth citing, digital PR that earns mentions on sites AI models already trust, and a review strategy that gives a business a track record to point to. An AI info page is the easy part. Earning the mentions everywhere else is the actual campaign.

Blue Archer is piloting AI info pages with AEO clients now

Blue Archer is currently building AI info pages for a small group of AEO clients, including our own site, and tracking what happens to their AI citation rate before and after. The plan: document a baseline for how often, and how accurately, each client's business shows up across a set of common AI prompts, publish the page, then check again at 30, 60, and 90 days.

It's the same kind of controlled thinking Nectiv used, applied across more than one site. There's nothing to report yet, since the pilot is just getting underway. But it's the difference between guessing at this tactic and actually measuring it.

Frequently Asked Questions

Q: What is an AI info page?
An AI info page is a plain, indexable web page written specifically for AI assistants like ChatGPT, Claude, and Gemini to read, rather than for human visitors. It states who a company is, what it does, who it serves, and what to say about it in short, factual sentences, usually at a URL like /ai-info or /ai-instructions.

Q: Is an AI info page the same as an llms.txt file?
An AI info page and an llms.txt file are not the same thing. An AI info page is a normal HTML webpage, while an llms.txt file is a separate machine-readable text file. Ahrefs' analysis of 137,210 domains found that 97% of published llms.txt files received zero requests, while AI info pages have documented cases of being crawled, indexed, and cited by ChatGPT directly.

Q: Will an AI info page get my business recommended by ChatGPT?
An AI info page mainly helps AI models describe a business accurately once it's already part of the conversation. It doesn't necessarily win a spot as the top recommendation for a broad, unbranded search. Getting chosen as "the best" option for a category search depends on a much wider set of signals: reviews, backlinks, original content, and mentions on sites AI models already trust.

Q: How do I make sure ChatGPT and other AI models can find my AI info page?
Link the page from a visible spot on the site, typically the global footer, so it has a normal crawl path instead of sitting orphaned with no internal links. Also check the site's robots.txt file to confirm AI crawlers like GPTBot and ClaudeBot aren't blocked from reaching it.

Q: How often should an AI info page be updated?
An AI info page should be reviewed and refreshed on a regular schedule, not published once and left alone. Nectiv's own testing found citations from its page faded within about a month and returned within roughly 24 hours of a refresh, and every well-run example, Tekmetric, HubSpot, and Observe.AI, displays a visible last-updated date.

Key Takeaways

• An AI info page is a normal, indexable webpage written specifically for AI assistants, not a special file format like llms.txt.
• Ahrefs found that 97% of published llms.txt files get zero requests; an AI info page is a fundamentally different, more effective mechanism.
• Tekmetric, HubSpot, and Observe.AI all publish one, and all three link to it openly from their site rather than hiding it.
• Nectiv's controlled experiment showed ChatGPT indexing and citing a new AI info page within 48 hours, including specific language that existed nowhere else on the site.
• The effect can fade within about a month, so an AI info page needs a visible last-updated date and a real review schedule, not a one-time publish.
• It helps AI models describe a business accurately. It's a smaller piece of the larger work needed to get recommended for unbranded searches.
• Blue Archer is piloting this tactic with select AEO clients now, tracking citation rate before and after.

Want an AI info page built into your AEO strategy

Blue Archer builds AI info pages as part of a broader, custom SEO and AEO strategy, not as a one-off tactic bolted onto a site that has nothing else going for it. That's the difference between a page that gets ignored after a month and one that keeps earning citations: the tactic works best alongside real content, strong entity signals, and a site built to be found in the first place.

If your business is ready to show up accurately, and get recommended, wherever people are asking AI models for answers, that's exactly the kind of strategy and execution Blue Archer builds for clients every day. Reach out through our project planner to talk about where an AI info page, and the AEO work around it, would fit into your site.