Can an AI Read Your Services?

Short answer first: if your page does not say what you do and where, in text, an AI has to guess. There is no special markup that fixes that, and no file you can upload instead. Google says so in its own documentation.

That is the good news. The work is plain words in the right place, and you can test a page in about ten seconds without any tools.

What does Google actually say about AI features?

Google’s AI features guidance for site owners — last updated 10 December 2025 — says there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimizations necessary. The eligibility conditions it does list are the ordinary ones:

• The page has to be indexed and eligible to be shown in Google Search with a snippet.
• Crawling has to be allowed — in robots.txt, and by whatever CDN or hosting sits in front of your site.
• The important content has to be available in text. Not baked into a picture, not trapped in a PDF.

Read the second half of that honestly: if no optimization exists, then nobody can sell you one. Anyone offering “AI Overview optimization” as a product is selling something Google says is not a thing.

What does Microsoft say?

Microsoft Advertising published guidance on 8 October 2025 about content being included in AI search answers. Its advice is structural rather than technical:

• Ask the question, then answer it. Direct questions with clear answers, phrased the way people actually search.
• Treat headings as chapter titles. Each one should mark off a clear slice of content.
• Do not hide the answer. Content tucked inside tabs or expandable menus may not be rendered at all.
• Use HTML for anything that matters. Not a PDF, not an image of text.

Both companies are describing the same thing from different angles: say it plainly, in text, under a heading that matches the question.

The ten-second check

Open any service page on your own site and do this. It works on a phone.

• 1. Read the heading and the first two lines out loud. Then say what you do and where. If you cannot say it from those words alone, neither can a machine.
• 2. Search the page for the last four digits of your phone number. Ctrl-F on Windows, Command-F on a Mac. Chrome on Android: the three dots, then Find in page. Safari on iPhone: the share button, then Find on Page. If the browser finds nothing, your number is part of an image — and an image cannot be read the way text can.
• 3. Give each service its own page, titled in the words a customer would use. “Plumbing repair in Petersburg” beats “Our solutions.”

What does good look like?

Take an invented example. A plumbing firm’s page headed “Wexford & Co. — quality you can trust” tells a machine nothing. The same page headed “Plumbing repair in the Tri-Cities” tells it everything it needs.

• The heading names the service and the town. One line, no slogan.
• The first two lines say what you do, for whom, and where. Before any story, any history, any mission statement.
• The phone number and address are real text on the page, not part of a graphic.

Wexford & Co. is an invented name. A search on 19 September 2026 found no plumbing business by that name in Richmond or the Tri-Cities.

What about robots.txt?

This is where most advice on the subject goes wrong, so here it is stated carefully. Your site can have a file at yoursite.com/robots.txt that asks crawlers to stay out of certain places.

• A whole-site block looks like this: the line Disallow: / underneath User-agent: *. That asks everyone to stay away from everything.
• Lines like Disallow: /wp-admin/ are normal and fine. Nearly every WordPress site has one. It is not a problem and it is not blocking your services.
• A block aimed at one named crawler is often deliberate. Somebody may have chosen to keep a specific AI crawler out. That is a decision, not a fault.
• It is a request, not a lock. Well-behaved crawlers honour it; it is not a security control.

Ask your web person before changing anything in that file. A wrong line there can remove a site from search entirely.

Which crawlers are we even talking about?

The AI companies publish their own crawler names. Two worth knowing, both read on 21 September 2026:

• OpenAI documents OAI-SearchBot, which surfaces websites in ChatGPT’s search features; ChatGPT-User, which visits a page when a person asks something; GPTBot, used for training foundation models; and OAI-AdsBot.
• Perplexity documents PerplexityBot, for surfacing and linking sites in results, and Perplexity-User, which fetches a page to answer a specific question.

Neither page carries a visible date, and both have changed before without announcement. If a decision turns on what one of them says, open it yourself on the day.

I ran Artifex Web through this

I found the problem by running the same check I am asking you to run. Three of my four service pages did not say what or where in the main heading. They do now. That is not a confession dressed up as marketing — it is the reason I trust the check, because it caught something on a site built by someone who does this for a living.

What this does not do

Passing this check does not guarantee that an AI will mention you, cite you, or send you anyone. Google states there is no optimization for AI features, which cuts both ways: nothing gets you in, and nothing is being withheld from you. What plain text does is remove the reason a machine would have to guess.

Your turn: read your own heading out loud today.

Every source below was opened and re-checked on 21 September 2026. Google’s and Microsoft’s pages carry dates; the OpenAI and Perplexity crawler pages do not, and change silently.