The first thing an agent gets wrong about your business.

Known has to come before anything else works

Say a countryside inn switched from a set Sunday lunch menu to an à la carte one back in the spring. Ask an AI agent whether it does a traditional roast and it might answer confidently, and wrongly, describing a menu that hasn't existed for months. Nobody told the AI anything false. Nobody told it anything at all, which is a different problem with the same result: a customer who trusted the answer, and shows up expecting something that isn't there any more.

That's the shape of almost every first mistake an AI makes about a business. Not an invented detail, a missing one, filled in with whatever sounded plausible. The inn's website still describes the building, the rooms, the general feel of the place, accurately. What it doesn't say, because nobody thought to update it the way they'd update a blackboard by the door, is the one fact that would have changed this particular answer.

Being known, properly, means the real, current facts about a business live somewhere an AI can check, not just somewhere a person could eventually find them by asking. This is the first link in a longer chain: known enables found, found enables trusted, trusted enables booked. Skip straight to worrying about being found, or worrying about ranking well, while the underlying facts are still scattered across a website that hasn't been touched since a menu changed, and every later step inherits the same gap. An AI can only be as accurate as what it was told, and most businesses have told it less than they think.

Where the gap usually is

Rarely in what's written. Almost always in what's changed since it was written and never got corrected everywhere it needed to be: a menu, a price, an opening time, a policy that used to be generous and now has conditions attached. Each of these lives somewhere true, a booking system, a till, a manager's memory, but "somewhere true" and "somewhere an AI can check" are not the same place, and the gap between them is exactly where a confident, wrong answer comes from.

This is why the fix isn't "write a better website." A better-written page describing the old menu is still describing the old menu, just more persuasively: the gap sits in whether the true, current version of the fact ever reached the place an AI looks, not in how anything reads, which a website redesign does nothing to close on its own.

Why this matters more than a factual correction

An AI has no one to ask and no way to know its own answer has gone stale, so it states an old menu with the same confidence it would state the true one. A person calling to check the roast is still on gets a quick, human answer and a small apology if it's changed; the AI gets neither, because it never knows to ask. The customer arrives with an expectation a phone call would have corrected in ten seconds, and the business absorbs a complaint about something it never promised.

Getting the facts current, and getting them somewhere an AI can check rather than somewhere a person eventually would, is the whole of the Known step, and everything past it depends on getting this part right first. That's the specific gap Selfe exists to close: connecting a business's real, changing facts, menus, prices, policies, to something an AI checks directly, so the answer it gives is the one that's true today, not the one that was true when someone last updated a page.

Isn't this the same as keeping a website updated?

Related, but narrower and more specific: not "does the site look current," but "does every fact an AI might check reflect what's true today, in the specific place it's checking."

How would we know where our own gap is?

Ask an AI a question a real customer would ask about something that's changed recently. If the answer describes the old version, that's the gap, found directly rather than guessed at.