Your website is ready for AI search when an assistant can do three plain things with it: find your pages, read the words on them, and see that you allow it to use what it reads. Most small sites pass one or two of those and quietly fail the rest, and the owner never finds out, because nothing looks broken. The page still reads fine to a person. It is just invisible to the machine writing the answer.
This matters because a lot of searching now ends in that answer instead of a click. About 46% of people now read Google’s AI answer instead of a link (Bain and Company, 2026). An assistant can only name a business it can read, so being readable is the entry ticket, not a nice-to-have.
The clearest way to see where you stand is the free Agent Readiness scan at isitagentready.com, from Cloudflare. It checks five plain areas. I run it on my own site (it passes) and on every site I build. Here is what each area means in owner words, and what fixing it looks like.
| Area | The question | A pass looks like |
|---|---|---|
| Find you | Can an assistant reach your pages at all? | Nothing blocks the crawler, and a sitemap lists your pages. |
| Read you | Is your text in the page, or hidden? | Real text in the HTML, not locked behind JavaScript or baked into an image. |
| Permission | Do you say what AI may do with your content? | Your robots file names the assistants and states the rules. |
| Discovery | Can a tool pick up your key facts cleanly? | Structured data and a plain-text file lay out your hours, services, and location. |
| Commerce | Can software buy from you directly? | Only relevant if you sell online. Most small sites skip it. |
Area 1: Can assistants find you
An AI crawler is a piece of software that visits pages and collects the text. If yours cannot reach a page, nothing else in this list matters. Two things usually get in the way: a robots.txt rule that blocks the crawler, and a missing sitemap, the simple file that lists your pages so a machine can find all of them instead of stumbling on a few. Fixing it means clearing the block and publishing a clean sitemap. On the sites I build, both ship by default.
Area 2: Can they read you
Reaching a page is not the same as reading it. If your words only appear after a browser runs JavaScript, or if your headline is really a picture of text, a crawler often sees an empty page. The fix is to serve the actual text in the HTML, so it is there the moment the page is fetched. You can check this yourself: right-click the page, choose View Page Source, and search for a sentence you see on screen. If it is not in the source, a crawler probably cannot read it either.
Area 3: Do you say what they may do
This is the area almost no one has handled. You can state, in your robots file, which assistants may read your site and what they may do with it. These are called content signals. Barely anyone sets them: only about 1 in 25 of the web’s 200,000 busiest sites tells AI what it may do with their content (Cloudflare, 2026). Getting it right is a genuine head start, and it is a small, one-time change.
Area 4: Can tools discover your key facts
Structured data is your hours, services, and location written in a format a machine reads without guessing, so it describes you correctly instead of inventing something. Worth being honest here: structure helps a machine read you accurately, it does not buy you a spot in an answer. A controlled test of 1,885 pages found that adding schema produced no citation lift on its own (Ahrefs, 2026). Do it for accuracy, not as a lever.
The same area covers llms.txt, a plain-text summary file for assistants. Treat it as future-proofing, never a traffic source: about 1 in 48 sites publishes one today (Web Almanac, 2025), and no consumer AI search engine reads it yet. It costs almost nothing to add, so I ship it and move on.
Area 5: Commerce (most small sites skip this)
The last area asks whether software can buy from you directly, the way an agent might place an order on a customer’s behalf. It applies to online stores, not to a service business that wants a call or a booking. If you do not sell online, a clean pass here is simply “not applicable,” and that is fine. The scan does not hold it against you.
What fixing it looks like
None of this is exotic. It is a handful of small, permanent changes: clear the blocks, serve real text, name the assistants, publish your facts. On a rebuild I do all of it at once, so the site launches ready instead of catching up later. If you would like to see your own site scanned and a faster, readable version built before you spend anything, I will put together a free preview. For the bigger picture of how being found changed, start with The Art of Being Found.
Common questions
What is AI search? It is when you ask an assistant like ChatGPT, Claude, or Google’s AI answer a question and it writes a reply built from web pages, instead of handing you a list of links to click.
Is being AI-ready the same as ranking on Google? No. Ranking is about where you sit in the list of links. AI-readiness is about whether a machine can read your site at all. They overlap, because both start with a page a crawler can reach, but you can rank decently and still be unreadable to an assistant.
Will making my site AI-ready get me recommended by ChatGPT? No, and it is worth being clear about that. An assistant can only name a business it can read, so readability is necessary. Being named is earned off your site, through reviews, mentions, and a reputation worth repeating. Readiness makes you eligible, not chosen.
How do I check my own site? Run it through isitagentready.com for the five areas above, and Lighthouse for speed and technical SEO. Both are free and take about a minute each. Then run a competitor and compare.
