How to check what AI is already saying about you.
Being found by an AI agent is the easy half of the job. Here's how to check where you stand today, and why the check itself only ever tells you half the story.
5 min read
Nothing at all
Being mentioned by an AI agent is easy to get excited about and easy to stop at. It's the easy half of being found, and found is only the second link in a longer chain, covered in full elsewhere in this hub and in the Agentic commerce hub. This piece is about the practical two-minute version: checking where you stand today.
Open ChatGPT, Claude or Perplexity, and ask the question a real customer would ask, not the one that flatters you. "Tell me about [business]" just returns whatever's easiest to find. Someone deciding asks something closer to "is it good for two people who don't want a big group class". Type it in and read exactly what comes back, without smoothing over anything that reads badly.
What comes back falls into one of three shapes.
A blank answer, or the business simply not coming up, means the AI found nothing it was confident enough to repeat. It won't say so. It will move on to whichever competitor it does have something to say about. Being skipped quietly is still being skipped.
Something wrong
Say a cookery school in the Brecon Beacons moved its weekend course from Friday evenings to Saturdays two years ago. Ask an assistant about weekend availability and it might answer confidently and wrongly, describing the Friday class that hasn't run since the switch. Nobody told the sources an AI reads that anything had changed, so as far as it's concerned, nothing has. This is the most useful of the three results, because it points straight at the gap: a fact changed somewhere, and nothing told the places an AI reads that it had.
Right, for now
Say the same cookery school updated its own price list the week rates went up. Ask an assistant today and it might answer correctly, right down to the new price. That's not the finish line either: it's correct because someone happened to update the right page before the assistant last looked, not because it will still be right after the next change, or on a different assistant reading different sources.
An answer is not evidence
Typing a question into a chat window and reading the answer gives you a claim, not a verified fact, exactly the same as reading a business's own website ever did. It just sounds more neutral, because it obviously isn't the business talking about itself. Ask where the answer came from, when it was last checked, whether it's still true tonight, and the honest reply for most of what an AI says right now is that nobody knows, including the AI. Checking what it says about you only relocates the trust problem, it doesn't solve it: the chat window becomes the new thing you're trusting on faith, the same way the business's own website used to be.
Doing this properly, not just once
None of that makes the two-minute check pointless, it just puts it in its right place. Pick one question a customer would genuinely ask, ask it again on a rough schedule, monthly is plenty for most small operators, and try it on more than one assistant, because they draw on different sources and weight them differently. Read what comes back as a symptom worth acting on, not a job that's finished the moment the answer sounds right, and not the whole job even when it is right, because found was never the destination.
The real fix sits one layer back from the checking habit itself: the facts about a business living somewhere current, sourced and checkable, rather than scattered across a web page, a PDF, and whatever reception happens to say that day. That's the gap Selfe exists to close: turning a business's real information into something every AI can check and act on directly, not just repeat and hope is still true.
Which AI should I check?
More than one, if you have the two minutes. They draw on different sources and weight them differently, so the same business can come out right on one and wrong on another.
What if two assistants say different things about the same fact?
That's the finding, not a glitch to ignore. It usually means the true version of that fact lives in more than one place online, and the two places disagree with each other.
Does asking about my own business count as gaming the results?
No, you're reading what's already there, not shaping it. Keeping the underlying information accurate, automatically and everywhere, is the separate job that moves you from found to booked.
Structured data, explained without the jargon.
A returns policy that says "just get in touch if there's a problem" and one that says "returns accepted within 30 days, unworn, tags attached" can describe the same shop. Only one of them can be checked by something that isn't a person using judgement.
Being found is not the same as being bought from.
This is the distinction the rest of this hub, and most of this site, keeps coming back to. Being found by an AI is real progress. It is not, on its own, a booking, a sale, or a reason to stop.
The first thing an agent gets wrong about your business.
An AI agent rarely gets a business wrong out of carelessness. It gets it wrong because the fact that would have changed the answer was never written down anywhere it could check.