| The thesis: inside an AI model, your brand does not have an essence the way you imagine it does. It has a location, defined entirely by what it sits near. Your meaning is your neighbors: the companies, concepts, and categories the machine has learned to place beside you. To a model, you are the company you keep, literally. Which means you do not change what you mean by describing yourself. You change it by changing your neighborhood. |
We think of a brand as having an essence, a core identity that exists on its own and that we then communicate outward. It is a natural way to think, and inside a machine it is not how meaning works. A model does not store a little dossier of your essential qualities. It stores a position, and the position is meaningful only in relation to everything around it. Strip away the neighbors and there is nothing left. There is no you in there, independent of what you are near.
This sounds abstract until you realize it inverts the entire practice of positioning, and hands you a different lever than the one you have been pulling.
How a machine actually holds meaning
A useful way to understand what happens inside these systems is that concepts are represented as positions in a vast space, and a concept's meaning is expressed by where it sits relative to others. Things that mean similar things end up near each other; things that mean different things end up far apart. Meaning, in this picture, is not a definition attached to a word. It is a location in a landscape of other words.
The linguist J.R. Firth put the intuition in a line decades before any of this existed: you shall know a word by the company it keeps. He meant that a word's meaning is revealed by the words that tend to surround it. Modern machines take that principle further than he could have imagined and build their entire understanding of meaning on it. What a thing means, to a model, is largely a function of what it reliably appears alongside.
You have no essence in there, only a location
Follow that to its conclusion for your brand. In the machine, you are not a self-contained identity that happens to have some associations. You are the associations. Your meaning is constituted by your proximity to other things: your category, your competitors, the use cases, the problems, the adjectives, the named leaders and technologies you tend to appear beside. Remove those relationships and the model has nothing left to say about you, because there was never an essence underneath, only a position defined by its neighbors.
This reframes a lot. When we wrote about being compared to the wrong peers in The Wrong Kind of Famous, that was the tactical surface of a deeper truth. It is not that the wrong comparison is unhelpful. It is that in machine cognition, the comparison partly is your identity. Who you are placed beside is not a footnote to what you mean. It is a large part of the substance of it.
What decides your neighbors
If your neighbors are your meaning, the obvious question is what determines them. The answer is co-occurrence in the record the machine learned from: how often, and in what light, you show up together with other things. You become a neighbor of whatever you are reliably mentioned alongside, compared against, listed with, explained in terms of. Every time your name appears in the same breath as a category, a competitor, a concept, a use case, it tugs your position a little closer to that thing.
Which means your neighborhood is not chosen by you and it is not fixed. It is the accumulated result of every context you have appeared in, most of which you did not author. Left unmanaged, you drift toward whoever talks about you most and whatever they place you next to, which may be nothing like where you want to stand.
The inversion this forces
Here is the strategic turn, and it runs against instinct. The default way to shape perception is to describe yourself: refine the messaging, sharpen the copy, state more clearly what you are. That is describing your location by talking about the point itself. But a point has no meaning in isolation in this space. Describing yourself in a vacuum barely moves you, because meaning here is relational, not declarative.
You move by changing your neighborhood. You shift what you mean by shifting what you reliably appear beside. Get consistently discussed alongside the category you want to own, the problems you actually solve, the caliber of company you belong with, and your position migrates toward them. Keep getting mentioned beside the commodity players and the wrong use cases, and no amount of self-description will pull you out, because the machine is listening to your company, not your claims.
Good neighbors, bad neighbors
This gives you a sharper way to think about associations than mere reputation. Some proximities lift you: being placed beside the acknowledged leaders in the category you aspire to, beside the sophisticated use cases, beside the concepts that signal seriousness. Some sink you: being filed next to the discount options, the outdated approaches, the adjacent-but-lesser category you have outgrown. And some are simply wrong, neighbors you ended up beside by accident of who happened to mention you, that miscast you to anyone the model tells. The work is to court the first, escape the second, and correct the third.
How to change your neighborhood
You change your neighbors by earning genuine co-occurrence with the things you want to mean. Show up, credibly and repeatedly, in the contexts that define the position you want: the category conversations, the problems, the company you belong with, corroborated by sources the machine trusts. Give the machine structured clarity about what you sit beside, which is part of the work in Structured Identity, and sustain it, because neighborhoods drift, through the ongoing discipline of Model Response Optimization.
A caution, because this could be misread as a name-dropping strategy. You cannot fake your way adjacent to greatness. Manufacturing shallow proximity, stuffing your content with the names of category leaders you have no real relationship to, produces a weak and brittle signal a capable model discounts, and it can backfire by placing you beside them as the obvious lesser. Proximity has to be earned by actually operating in the neighborhood: real work, real customers, real corroboration in the right contexts. The goal is not to appear next to what you want to mean. It is to genuinely belong there, and to make sure the record reflects it.
Questions this raises
What does “you are your nearest neighbors” mean for a brand?
It means that inside an AI model your brand is represented as a position defined by what it sits near, not as a self-contained essence. Your meaning is constituted by your associations: the categories, competitors, concepts, and use cases you reliably appear beside. Change those neighbors and you change what you mean.
How do machines represent meaning this way?
A useful way to understand it is that concepts occupy positions in a large space, and meaning is expressed by relative location: similar things sit close, different things sit far apart. It echoes the linguist J.R. Firth's idea that you know a word by the company it keeps. Modern models build much of their understanding of meaning on that principle.
What determines which things my brand is associated with?
Co-occurrence in the record the models learned from: how often and in what light you appear alongside other things. Every mention beside a category, competitor, concept, or use case nudges your position toward it. Most of those contexts you did not author, so unmanaged, you drift toward whoever mentions you most.
Why doesn't better messaging fix my associations?
Because meaning in this space is relational, not declarative. Describing yourself in isolation barely moves your position, since a point has meaning only in relation to its neighbors. You shift what you mean by shifting what you reliably appear beside, not by restating what you are.
Can I just mention the category leaders to get associated with them?
No. Manufacturing shallow proximity produces a brittle signal that capable models discount, and it can backfire by casting you as the obvious lesser beside them. Real proximity has to be earned by operating in the neighborhood, with genuine work and corroboration in the right contexts, then reflected accurately in the record.
The bottom line
You have been managing your brand as if it had an essence the machines would eventually recognize if you described it well enough. In the machine, there is no essence to recognize, only a location in a field of other meanings, and that location is set by the company you keep. This is disorienting and also freeing, because it hands you a lever you were not using. Stop only polishing the description of the point. Start moving the point, by earning your way into the neighborhood you actually belong in. To a model, you are your nearest neighbors, so choose them, and earn them, on purpose.
Find out which neighborhood the machines have put you in. Request your Brand Interpretation Assessment, or read the Interpreter Architecture whitepaper for the full thesis.