When AI Introduces Your Company,
Which Version of You Does It Use?

We're fortunate to live in this era of computers, the internet, and AI.

It also means your company has picked up an introducer you've never met.

Picture this: a potential customer gets your business card passed along by a friend. Before calling, they type your company name into an AI tool and ask, "What does this company do?" A few seconds later they have a paragraph. It sounds authoritative, and it's mostly correct. It just describes you as you were three years ago: your main line of business back then, the area you served, and the tagline on your website that you've since changed. Or it could be worse. It mixes you up with another company with a similar name, or simply says it can't find anything.

The customer won't know anything is wrong. They'll just come to talk to you with that paragraph in mind. Or they won't come at all.

"If You Can't Say Who You Are in One Line, a New Logo Won't Help" was about getting "who you are" down to one line. This piece asks the follow-up: once that sentence is written, is it the same everywhere? An earlier piece, "When a Customer Asks AI 'Who's Best at This,' Are You in the Answer?", asked whether you show up in the answer at all. This one asks whether the you that shows up is the right one.

01Consistency started as a discipline for human readers

Brand consistency is an old topic. People in branding have been talking about it for decades, and the reasons have always been clear.

One framework that gets used a lot, the brand identity prism, breaks a brand into six facets: physique, personality, culture, relationship, reflection, and self-image. It sets out to answer just one question: how a brand gets seen, understood, and felt by customers "consistently." Those six facets eventually land on four kinds of touchpoints: visual, verbal, content, and experience, in that order.

Another way of breaking down a brand lists four things under the "image" cell: visual identity, brand voice and language, brand story, and brand experience. "Voice and language" here means slogans, ad copy, tone, and style.

Both frameworks put "how you describe yourself" on the same level as your logo. The reason is practical: a person's impression of you gets pieced together slowly, over many contacts. They look at your Facebook page once, browse your website once, hear your sales rep pitch once. If those three contacts give them three different stories, the impression blurs.

But people have one saving grace: they look at one place at a time, and they fill in the gaps themselves. Your Facebook page is a bit casual, your website a bit formal, and the directory listing was filled in five years ago. People read the gap as "the company is growing" and don't look into it much.

02Machines read the same material, but all at once

AI isn't that forgiving.

When it's asked "what does this company do?", it won't necessarily look only at your newest page, the one you're happiest with. Every description of you it can read may get used to build the answer. Say your website says you do "branding and websites," a directory lists you under "graphic design," and an old news story calls you a "freelance studio." It has to choose one of the three versions, or blend them into a version that looks like none of them.

The more scattered your descriptions, the blurrier the you it pieces together. The older your descriptions, the more outdated that you is.

An infographic making the rounds in marketing circles, an "AI search optimization checklist," turns this into routine work. It lists a category called "brand clarity" among the monthly tasks: check how AI describes your brand and your competitors, find the gaps in perception, update your company description across platforms. Every quarter there's one more item: "correct inaccurate AI answers."

When something gets put on a monthly and quarterly schedule, it means the people doing it already expect AI to get brands wrong, and to keep getting them wrong.

The graphic closes with a formula: discoverability × clarity × authority × trust = AI visibility. You don't have to take the formula itself seriously, but it uses multiplication signs. However well you do on the other three, if clarity is close to zero, the product is still close to zero.

This piece covers clarity alone, because of the four it needs the least technology and budget. A business owner can do it themselves.

03Step one: list every place you've introduced yourself

Most companies have never done this. These descriptions were written in different years, by different people, and nobody has ever been responsible for putting them side by side.

Make a table and fill in, row by row, every place where the company "introduces itself":

This table is only a starting point; your company may have other places too. Once it's filled in, you'll usually see three situations.

Outdated. The sentence was written when the company was still doing something else. The business moved on long ago, and the bio didn't move with it.

Divergent. One company, with marketing writing one version, sales writing another, and the contractor who built the website years ago writing a third. Each sentence is fine on its own. Put them together and they don't sound like the same company.

Blank. There are places where you've never written anything. The description there was pulled in automatically by the platform, or written for you by someone else. Wherever you aren't speaking, someone else is.

04Step two: ask AI yourself, and not just once

Once the table is filled in, open the AI tools you use most and ask the way a customer would, for example "What does this company do?" and "Who are this company's services a good fit for?"

Ask each question at least twice, and ask again at a different time. AI's answers vary from one time to the next; look only once and it's easy to mistake a fluke for a verdict.

Then put the answers next to the one sentence you wrote in "If You Can't Say Who You Are in One Line," and mark them up, paragraph by paragraph:

• ✓ Gets it right, and it's the you of today
• ✗ Describes an old business, old address, or old positioning
• ✗ Gets it wrong, or mixes you up with another company
• ✗ Leaves out something you care about a lot

While you're at it, ask once how it introduces two or three of your competitors. That checklist includes this item too. Seeing how others are described often makes things easier to spot than looking at yourself: so when customers compare, these are the two paragraphs they're handed.

Every ✗ you mark is worth taking back to the table from section 03 to hunt for. AI sometimes makes things up out of nothing, but the thing worth checking first is whether it read that version of you in some corner you never cleaned up. You can't verify that inference for every item, but the ones you can trace are the places to fix first.

05Step three: fix what you can change first

Order the fixes by how much control you have.

Fix your website first. It's the only place where you have the final say, with enough room to spell that one sentence out properly. At many companies, the "About us" page was written the year the site launched. Since then the site has had a new layout and a new key visual, but no one has touched the words on that page. Getting your homepage's first paragraph, your About page, and your footer line to say the same thing is the bare minimum.

Next come the platforms where you have an account and can make changes yourself: Google Business Profile, your Facebook page, all the various bio fields. These are quick to change. It's just that nobody normally thinks to.

After that come the ones you have to ask others to change: directories, trade association pages, introductions on partners' websites. Send an email with the new paragraph attached, and most will swap it in for you.

Last are the ones you can't change: old news stories, articles other people have written. Leave those be. What you can do is let the new, consistent description pile up until it outnumbers the old ones.

At the technical layer there are also things like structured data that make a site easier for machines to understand, but that comes later. When the descriptions themselves don't agree, finer markup only labels the disagreement more clearly.

06Consistent doesn't mean pasting the same paragraph everywhere

At this point it's easy to swing to the other extreme: copy one standard description into every location.

Do that and your Facebook page reads like an official notice, and your quotes read like ads. The "personality" facet of the brand identity prism already allows for speaking in different tones in different settings, as long as it still sounds like the same person.

What you need to hold onto is the core sentence: who you are, who you work for, and what makes you different. The tone can shift with the setting. The meaning of that sentence can't.

And to be honest: even after these three steps, AI may still get you wrong. The old stories are still out there, other people's retellings are still out there, and what the model reads and how it weighs it are out of your hands. Consistency can't guarantee the outcome. What it can do is two things: everywhere you control says the same sentence, and the next time you're described wrong, you know where to look for the source.

That one sentence has to be the same everywhere

Before Brand consistency was a discipline for human readers. Miss a spot or two and customers would fill in the gaps themselves.

Now It's also raw material for machines. Every spot you miss could become the version a customer hears first.

"If You Can't Say Who You Are in One Line" wrote that one sentence; this piece puts it back in every place. Do both, and when a customer asks AI, they're more likely to get the you of today.

Root the brand so it grows deep. A brand that describes itself inconsistently has roots in several places, and none of them go deep.

If, halfway through the audit, you find that what needs changing goes beyond a few lines of bio, because the sentence itself isn't thought through yet or your website as a whole can't carry it, then it's time to start over, from brand strategy or from a website audit.

Rooted in craft. Built for the new wild.