A brand enters the machine wearing a very elegant coat.
The machine asks what it is.
“An innovative solution for modern life,” says the brand.
The machine places it in a cupboard marked MISCELLANEOUS.
This is the new positioning problem. As AI assistants begin to filter, compare, and recommend choices, a brand must be understood before it can be considered. If its language is vague, its products are poorly described, or its claims cannot be checked, it may disappear before a person sees it.
Nobody rejected it. The machine simply ran out of places to put it.
Naming is part of seeing
In a 2007 experiment, Russian speakers—whose language separates lighter and darker blues—were faster to distinguish shades when they crossed that linguistic boundary. When researchers disrupted verbal processing, the advantage disappeared. The finding is modest but useful: names can affect how quickly a difference becomes available to thought.
Brand positioning has always done this. A category tells people what kind of thing they are looking at. A clear name gives the product a shelf in the mind. A useful difference gives them a reason to take it down again.
AI adds another observer.
Accenture’s 2026 survey of 25,590 people across 16 countries found that consumers are increasingly willing to delegate parts of buying to AI agents. The agent may compare prices, narrow the options, or create the shortlist before a brand meets the human at all.
The customer remains human. The machine has become the doorman.
The doorman has a clipboard.
BUREAU SYSTEM NOTE 12: Your brand possesses emotional depth. The category field is blank.
Poetry needs paperwork
The vault idea needs one amendment.
For people, a vivid name can open perception. For a machine, the name must also connect to evidence.
Google’s guidance for merchant listings asks for explicit product information such as the brand, offer, price, availability, and identifiers. This is not glamorous. Neither is a passport, until the border.
Kantar makes the broader brand argument: AI systems respond to clarity, consistency, and corroborated truth. A beautiful story on the homepage cannot compensate for contradictory descriptions, missing product facts, or reviews that tell another tale.
The future brand needs two kinds of language:
- Human meaning: a clear category, memorable point of view, and reason to care.
- Machine proof: structured facts, precise claims, consistent descriptions, and outside evidence.
Meaning creates desire. Proof allows the doorman to recommend it.
The three-label test
Before commissioning a grand strategy film involving mist, ask three smaller questions:
- What are we? Use the category a customer would actually request.
- Why this one? Name the difference without using innovative, seamless, or leading.
- Where is the proof? Show the facts an assistant could verify elsewhere.
Then test the answers with both a person and a machine. Ask each to describe the brand, compare it with three alternatives, and explain when it should be chosen.
The answers need not be identical. They should be recognizably about the same organization.
There is also a useful limit. Gartner found that consumers were more open to AI helping narrow options than making the final purchase decision. The machine may arrange the shelf. A person still brings memory, trust, taste, and a curious attachment to the red one.
Write for the person. Then make the facts clear enough for the machine carrying the message.