POINTER · FREE SCAN

Can an AI actually use your website?

Not "is it optimised" — can a real AI agent find what you do, what it costs, and how to take the next step? Twenty seconds. No signup. Try your own site, or a competitor's.

Scan a site — free See ours first

You'll get every check, pass or fail, with the raw request that produced it.

What the scan actually does

Most AI-readiness checkers look at your website and confirm the right files are present. This one calls the things those files promise. Four differences, and they're the whole gap:

Other AI-readiness toolsPointer
Check that you published a menu.Orders off it.
Check that your address is listed.Knocks on the door to see if anyone answers.
Check that you have a price list.Checks the price is a number a machine can read — not a picture, not "contact us".
Check that a page exists.Asks for a page that cannot exist, to see whether you admit it or say yes to everything.

19 live requests. Nothing scored from reading a document. Then the test that matters: we hand a blind AI model your site and a real buyer's task, and see whether it can get to a price and a next step without a human. It either can or it can't.

Don't take our word for it — run all three

Scan your site with all three of these and compare. It takes about a minute, and the disagreement between them is the entire point.

Cloudflare agent readiness Checks the standard files are present and well-formed. Free.
llmscan.dev Their own words: 8 signals, 100 points — whether AI can parse your homepage. Free.
3Dogs Pointer Calls what you advertise; tests whether an agent reaches a price and a next step. Free.

If the first two give you a high score and ours doesn't, that gap is not a stricter opinion about the same question. It's a different question — and it's the one an AI asks when someone tells it to find a firm like yours.

We scored 100 on Cloudflare and 90 on llmscan while our own test gave us 39 — because we were advertising things we hadn't finished building. We fixed it in an afternoon. That whole story, including the audit where six AI models said we'd rigged our own scoring →

What a failure actually costs you

Someone opens ChatGPT and types "find me a mortgage broker who does jumbo loans in Henderson" — or a family lawyer, or an agent who writes contractor liability, or a dentist taking new patients.

To put you on the shortlist, the AI needs three things off your site:

What it needsWhere yours usually isWhat happens
What you specifically doA headline image, or "full-service solutions"Can't tell if you do jumbo loans. Skips you.
What it costs, or who qualifies"Contact us for a quote"Nothing to compare. Skips you.
How to take the next stepA form a human must fill inCan't act. Skips you.

It doesn't guess and it doesn't ask. It builds the list from firms it can read. You're never told you were considered and dropped — the same way you're never told which callers hung up before you answered.

An honest note on timing. We're not going to tell you the sky is falling. Today, almost every client is still a person, and a person can find your phone number. What's changed is that buyers have started asking AI to shortlist — and a shortlist is decided long before anyone calls. Almost nobody in mortgage, insurance, real estate, law or private practice has done anything about this yet, so being early currently costs an afternoon rather than a budget.

If it finds something, fixing it starts at $199

Most of what's missing are files we generate and publish for your domain. Some need a change at your host. A few are decisions about what your firm actually offers — those we walk you through. Then we re-run this exact scan and give you before and after in writing, plus two follow-up scans across 90 days so it doesn't quietly break again.

Start with the scan. Decide after.

Free, twenty seconds, no email required. If nothing's wrong, you've lost twenty seconds.

Scan my site — free How Pointer works

Scores reflect a site as tested on the date shown; any site can change in an afternoon, and we hope they do. We publish our own score on every scan, including when it was bad.