Every brand has an AI biography

Somewhere right now, an AI system is answering a question about your business. Not from your website. From memory, and from whatever it has gathered about you across the wider internet. Someone asks it what you do, whether you are any good, who you serve, how you compare to the alternatives, and it answers with quiet confidence, drawing on a version of you it assembled without ever asking your permission.

That version is what I have come to call your AI biography. Every brand has one now, whether it has thought about it or not. It is the story machines tell about your business when you are not in the room, and for a growing number of your future customers it is the only version of you they will ever meet.

Most businesses have never read their own AI biography. Most do not know it exists. And that is exactly the problem, because a story you have never read is a story you are not shaping.

What is an AI biography?

Your AI biography is the composite understanding an AI system holds about your business, built from your entire digital footprint rather than from any single page you control. It is not stored in one place. It is not something you can log into and edit. It is inferred, continuously, from everything that exists about you across the web, and then reassembled on demand every time someone asks a machine about you.

Think about what that means. When someone types "how AI understands your business" into a search that is really about their own company, they are asking the right question in the wrong direction. The answer is not on their homepage. It is spread across dozens of sources, most of which they did not write, and the machine is doing the work of stitching those sources into a single narrative. That narrative is the biography. Your website contributes to it. So does everything else.

This is a genuine shift, and it is worth sitting with for a moment before rushing to fix anything. For twenty years, the goal was to control your message on channels you owned and rank the pages you published. The AI biography does not work like that. It is not authored by you. It is assembled about you. And the businesses that will do well are the ones that understand the difference and stop treating their website as though it were the whole story.

Why is your website no longer your whole story?

Here is the uncomfortable truth underneath all of this. AI does not rely on a single source, and it has good reason not to. A brand's own website is the least objective account of that brand available. Of course it is flattering. Of course it is on-message. The machine knows this, in the sense that its behaviour reflects a heavy weighting toward independent corroboration, and so it reads far more widely before it decides what to believe about you.

When an AI system builds its picture of your business, it is drawing on sources like these, and many more:

Your LinkedIn presence, and those of your leadership, which often carry more weight in professional queries than any marketing page you have ever published. Your Google Business Profile, with its categories, hours, location, and reviews. Podcasts you or your people have appeared on, and the transcripts that make those appearances machine-readable. Reviews across every platform that hosts them. News articles and press coverage. Industry associations and directories that list, or fail to list, your business. Government and regulatory databases where your legal entity is recorded. Structured knowledge systems and encyclopedic sources. And yes, your own websites, which matter, but as one witness among many rather than the whole account.

Notice what most of these have in common. You do not control them. You can influence some, earn your way into others, and correct a few, but the days when your owned channels were the primary input are over. The machine has learned to triangulate, and triangulation is precisely what makes the AI biography so much harder to fake than a well-optimised landing page.

How does AI build a composite picture?

So if AI is not simply reading your website, what is it actually doing? It is building what I call the composite picture, and understanding this framework changes how you think about the whole problem.

The machine does not read one source and stop. It compares information across many sources and looks for agreement. Where independent accounts of your business converge, saying the same thing about what you do, who you serve, and what you are known for, the system treats that convergence as signal. It becomes confident. Where the accounts diverge, contradicting each other or trailing off into silence, the system encounters noise, and noise makes it hesitant.

This is the mechanism people miss when they say AI "reads your website." It does not read a source so much as cross-examine a crowd of them. It is less like a reader and more like an analyst assembling a profile from many statements, weighting each by how independent and credible it seems, and paying closest attention to the sources that have no commercial reason to flatter you. When someone wonders how ChatGPT knows about their company, this is the answer. It knows what the ecosystem collectively and consistently says, filtered through its judgment about which parts of that ecosystem to trust.

The composite picture has a quiet consequence that is easy to underestimate. It means the machine can be more confident about you than your own website would suggest, if independent sources reinforce a clear story. And it means the machine can be far less confident than you would ever expect, even with a beautiful website, if the independent picture is thin or contradictory.

What happens when your AI biography goes wrong?

Now for the part that tends to make people sit up. When the composite picture is thin, inconsistent, or contradictory, the machine does not politely return an empty result. It fills the gaps. It takes the incomplete signals it has and synthesises the most plausible-sounding version of you it can construct, and it presents that invented version with the same confidence it would give a fact.

This is brand hallucination, and it is not a rare edge case. A Deloitte survey found that 77 percent of businesses using AI worry about hallucination issues, and that 47 percent of enterprise AI users admitted to making at least one major business decision based on hallucinated content. The concern is widespread because the phenomenon is real, and the version of it that should worry you most is the one aimed at your own brand.

In Become the Answer I describe five specific forms brand hallucination takes, and the everyday symptoms map straight onto them:

An outdated CEO, discontinued services, or an old product still presented as current is temporal distortion. The signals causing it are real; they are just no longer true. The ecosystem accumulates information over time without automatically discounting the old, so if the loudest, most repeated signals about you describe where you were five years ago, the machine will describe a historical version of you as though it were today.

Wrong services or a business placed in the wrong category is miscategorisation, the most commercially damaging form. A full-service firm read as a narrow specialist gets quietly excluded from every conversation about the wider range of problems it can actually solve. You never see the opportunities you were left out of, because exclusion is invisible from the inside.

Conflicting locations, and the machine attributing capabilities you do not have or missing ones you do, is expertise misattribution and, in its starkest form, fabricated specifics. When a business has so little consistent independent signal that the system has almost nothing solid to work with, it invents plausible detail to fill the void. Services that do not exist. Claims never made. If this is happening to you, it is the clearest possible sign that your independent presence needs urgent attention.

Inconsistent positioning and old reviews dominating the picture is what happens when you are not the clearest, most consistent voice defining yourself. Into that gap steps competitive misrepresentation, where the machine's account of you gets shaped by how competitors, critics, and stale reviews described you rather than by how you understand yourself today. If others are defining you more loudly and more consistently than you are, the machine uses their version.

And merged entities, where the system confuses your business with another of a similar name, is a categorisation failure in the structural knowledge systems that anchor identity. When those foundations are weak, the machine cannot cleanly tell where you end and someone else begins.

The most unsettling thing about all of this is not that it happens. It is that most businesses have no idea whether it is happening to them right now, because nobody ever told them to check.

Why does consistency matter more than volume?

If there is one message in this whole article I would tattoo onto the marketing plan of every business I work with, it is this. Consistency beats volume, and it is not close.

One hundred inconsistent mentions are worse than twenty highly consistent ones. This runs directly against instinct, because when people first understand the AI biography, their reflex is to produce more. More content, more posts, more presence everywhere, on the theory that volume must help. But volume of contradictory or off-message signal does not build a clearer picture. It builds a noisier one. Every mention that describes you slightly differently, positions you at a slightly different angle, or reflects a slightly different era of your business is another vote for confusion, and the machine counts the votes.

Twenty sources that all describe you the same way, in the same category, with the same core story, produce something a hundred scattered mentions never can: confidence. The machine can categorise you, and a business it can confidently categorise is a business it can confidently recommend. A business assembled from a hundred conflicting fragments is one it recommends around, reaching past the noise to a competitor whose story holds together.

This is why the "just publish more" advice does so much quiet damage. More owned content is just more things you have said about yourself, and it does nothing to make the independent picture more coherent.

Can you rewrite your AI biography?

Yes, and this is where it turns practical, though I want to be honest about the nature of the work. You cannot log in and edit your AI biography directly. What you can do is change the underlying reality the machine reads, and over time the biography follows. It is slower than editing a page and far more durable, because you are not gaming a description; you are correcting the record the description is drawn from.

The work that actually moves it looks like this. Publish original research and genuinely distinctive thinking, the kind of material independent sources want to cite rather than merely consume, because citation is what plants a clear, credible signal others build on. Update your entity information everywhere it is wrong or stale, from directories and profiles to the structural databases that anchor your identity, so temporal distortion has nothing to feed on. Hunt down and remove inconsistencies, the abandoned profile with the old positioning, the outdated bio, the conflicting category, every fragment casting a different vote. Earn trusted third-party mentions, the coverage and references and organic discussion that no amount of self-publishing can substitute for, because independent corroboration is the currency the machine values most. Improve your structured data so the machine can read your identity and expertise cleanly rather than inferring them from scattered clues. And clarify your positioning until it is sharp and singular enough that everyone describing you, including you, is telling the same story.

None of this fits on a checklist that promises results by Friday. All of it compounds. And unlike a tactic that stops working the moment a model updates, a coherent, well-earned biography gets stronger as the systems get better at telling coherence from noise.

What is the difference between visibility and recognition?

Underneath everything I have said sits a distinction I keep returning to, because it is the one the whole industry blurred and I spent years blurring myself. Visibility and recognition are not the same thing. Visibility is being findable. Recognition is being understood, well enough and by enough independent sources, that a machine forming a confident answer includes you without hesitation. Your AI biography is the difference between the two made concrete.

In my book I describe the three conditions that have to be present for recognition to form, and I call them the Recognition Triangle: Clarity, Credibility, and Earned Authority, with Recognition itself held in the centre by all three at once. The AI biography is the Recognition Triangle seen from the outside. Clarity is whether your biography tells a consistent, unambiguous story about what you stand for. Credibility is whether that story is backed by independent evidence rather than your own assertions. Earned Authority is whether the wider ecosystem has confirmed it, across the structural, expert, and community layers where real validation lives. When those three align, the biography holds, and the machine has something coherent enough to recommend. When they do not, the biography fractures into exactly the hallucinations we have been talking about.

So read your own biography. Ask an AI system about your category without naming yourself and see whether you appear and whether the description matches who you actually are. Then ask it about you directly and notice what it invents, what it gets wrong, and what era it thinks you are living in. That is your starting point, and almost nobody has looked.

Because here is the thing to carry out of all this. AI does not remember your brand because you published another blog post. It remembers the story the entire internet tells about you.