Six Weeks to a Brand

Earlier this year I spent six Sundays with a furniture and object designer in his early twenties. He was home for the summer, due to fly back to school abroad, and he wanted to leave with something more than sketches: a brand that existed outside his own head. His father, who runs a consumer brand of his own, sat in on most sessions — which meant the room always had two teachers, one for AI and one for taste.

The awkward moment that ended up defining the whole project came in week four. He showed us photographs of a piece he had actually built — a chair with a printed transparency gradient running through it, so the center seems to dissolve. His father and I both assumed it was an AI render. It wasn't. It was a real object he had made with his hands, photographed on a white background.

For a designer, that's close to the worst compliment available. Your physical work reads as machine output. The obvious move is to fix it — add texture, shadows, staging, make the thing look reassuringly real. We talked about that for a while. Then the conversation turned somewhere more interesting: what if the confusion was the point?

His own view of AI imagery was that it's easy to mass-produce and short on life. What he was doing was the inverse — building, slowly and physically, objects that produce the effects people associate with AI. A thing that looks impossible until you lean in and see the wood grain. The weakness, stated carefully, was a position: real objects, made by hand, that earn the double-take. Click the post thinking it's a render; discover it's a chair.

Writing the brand down

That sentence didn't come from a brainstorm. It came out of a document we built over two sessions, in a way I now use with every founder I coach: the AI asks, the person answers.

He had tried the opposite first — dumping references and half-thoughts into a chat and asking for a brand definition. The results felt like nobody. So we reversed it. The AI interviewed him, one question at a time, and he answered by voice: why this name, who is this for, what do you refuse to be. Spoken answers, typos and all, went into a file — a brand definition the tools would read before doing anything else. What surfaced was specific: warmth over cold minimalism, playfulness without kitsch, customers who want a grin from their furniture rather than a lecture.

Then we attacked it. One prompt did most of the work: if you were a brand expert, how would you criticize this document — and what questions would you ask to sharpen it? The first pass produced a flood, so we made it ask one question at a time. We fed it his website and social accounts as evidence against the text. Contradictions surfaced; the document got carved down. It's still high-level, and it should be — a brand definition that specifies everything leaves no room for the work to surprise you.

Four strangers before every post

The last piece was deciding what to publish. He wanted to post regularly, but his confidence in judging his own work was — his words — unreliable. You can't see your own output from outside.

So we built a panel: four agents with distinct personas. A demanding furniture collector. A gallery curator. An interior designer. An ordinary consumer with good taste and no patience. Before anything goes up, each one reviews the image and caption from its own seat and answers three questions — would you feel something, would you follow, would you buy — in parallel, on screen, like a focus group that costs nothing and never flatters.

The panel isn't an oracle. AI personas drift toward the generic, and the fix is to keep feeding them real comments and messages as those accumulate, so their judgments stay tethered to actual people. When the panel raised a point he disagreed with, the disagreement itself went back into the brand document — which is the real loop: reactions revise the definition, and the definition shapes the next post.

Six Sundays earlier, none of this had existed outside his head — no account, no sentence describing the work, nothing a stranger could find. In the sixth week, the first post went up on the new account. First frame: the piece on white, looking exactly like a render. Following frames: the object in a room, and the hands that made it. Caption generated in his tone, because by then the folder itself carried the rules — he no longer had to explain himself to the machine at the start of every session.

What actually got built

Six weeks produced no factory contract and no sales. I want to be honest about scale: one account, one post, a handful of documents. What he flew back with is smaller and more durable than a launch — a definition of his work that survives cross-examination, a critic he can summon at will, four strangers to consult before publishing, and a daily routine that runs without me.

Most of the AI stories I tell are about improving work that already exists — faster reports, safer contracts. This one is categorically different: there was nothing to improve. What the six weeks produced is a brand that didn't exist before, out in the world where a stranger can find it. And what AI changed wasn't his ability — the chair proves that was already there. It changed the condition that quietly kills most one-person brands: being alone with your own judgment. An interviewer to draw the brand out of you, a critic to argue with it, four strangers to face before every post — no solo designer sustains that cast alone, week after week. He flew back with all of them in a folder.

The part I keep thinking about is the chair. Nothing about the object changed between the day it embarrassed him and the day it anchored his brand. The only thing that changed is that the accident got written down, defended against a critic, and claimed. Most weaknesses don't become brand cores. But you find the ones that can only by putting them in a document and letting something argue with you about it.