Two of the most quoted lines in modern management were never meant as decoration. In 1954, Peter Drucker wrote that the purpose of a business is to create a customer. In 1997, Clayton Christensen reframed the same idea from the buyer's side: customers do not buy products, they hire them to do a job. For decades both lines circulated as slogans, printed on slides and forgotten by the next agenda item. They were not slogans. They were instructions. And the instruction is the same in both: the work that decides who wins happens before anyone is ready to buy.
Drucker and Christensen as Instructions
Read literally, Drucker is not describing a transaction. He is describing the manufacture of a want that did not previously exist in usable form. To create a customer is to install, in someone's mind, the conviction that a problem is worth solving and that solving it is worth paying for. Christensen sharpened the lens but kept the direction. The job a product is hired for is defined upstream of the product. It is the shape of the dissatisfaction, the standard for what counts as progress, the reason a person reaches for one category and ignores another.
In both readings the frame precedes the purchase. What counts as a problem, what counts as a credible solution, what is worth paying for, who is considered serious: all of it is decided before the market opens. The frame is the set of assumptions a buyer carries into the room. By the time a vendor is being evaluated, the most important decisions have already been made, silently, somewhere upstream.
For roughly seventy years most companies could afford to ignore this. Demand capture was hard enough on its own. Rank higher. Target better. Convert faster. Close tighter. If you were competent at the mechanics, you survived, and the firm that ran the mechanics best usually won. The frame mattered, but it was a luxury good. Execution was the bottleneck, so execution was the moat.
The Mechanics Collapsed
That bottleneck has dissolved. Discovery, copywriting, research, segmentation, targeting, and optimization were the disciplines that separated winners from also-rans, and they are now broadly cheap. The tools that used to require a department now require a prompt. A competent operator with current models can produce in an afternoon what a marketing function once produced in a quarter. This is not a forecast. It is the observable state of the work.
When a capability becomes abundant, it stops being a source of advantage. If everyone can rank, target, and convert at a high level, then ranking, targeting, and converting no longer sort the field. The mechanics still have to be done, but doing them well is now table stakes rather than a differentiator. Advantage does not vanish when this happens. It relocates. It moves to whatever remains scarce.
What remains scarce is the layer above the mechanics: the pre-demand layer. This is the mental model people adopt before they shop. What counts as a problem. What counts as premium. Which metrics matter. Who is considered credible. A firm that owns this layer does not have to win the conversion fight, because it has already shaped the criteria by which the fight is judged. Drucker saw the territory. Christensen mapped it. Category design sold it as a method. The AI era is what turns it from one advantage among many into the only defensible surface left.
Production vs Distribution of Belief
Here is the turn that changes the strategy. The same forces that collapsed the mechanics also collapsed the production of frames. Generating a category manifesto, a point of view, a renamed problem, a manifesto-grade narrative, is now trivial. Any founder with a laptop can generate one before lunch and three more after it. If the frame were the moat, the moat would already be gone, because frames have become as cheap as the copy they used to compete with.
The bottleneck is no longer narrative production. It is narrative adoption. Producing a frame and getting a market to wear one are different problems with different economics. Adoption requires trust. Trust requires a track record. A track record requires time, consistency, and a public identity that a model cannot manufacture on demand. You can generate a thousand plausible frames in a day, and not one of them is believed, because belief is not a property of the text. It is a property of the relationship between the source and the market over time.
You can LLM your way to a frame. You cannot LLM your way to being believed.
So the real moat is not authorship. It is distribution of belief. Not who writes the frame, but who gets the market to adopt it as its own. The compact version: the production of belief has been commoditized, and the distribution of belief has not. Anyone can state a position. Almost no one has earned the standing to make a market reorganize its assumptions around that position. That standing is slow to build, hard to fake, and impossible to download, which is exactly what makes it defensible.
Reframing in the AI Discontinuity
The obvious objection is that this favors incumbents and first movers, and the objection is correct. In a market whose frame is already set, a better frame usually loses to an entrenched one, because the entrenched frame has the trust and the track record that the better idea lacks. Most of the time, the firm that defined the category earliest keeps it, even when a sharper definition arrives later.
Which is precisely why discontinuities matter. A discontinuity is the rare moment when an entrenched frame loses its grip, because the world it was built for no longer exists. AI is the current one. Every category whose frame was constructed for a pre-AI world is, for a short period, open for reframing. The assumptions about what is hard, what is valuable, what is credible, and what is worth paying for were calibrated to a reality that has shifted under them. The market is briefly willing to reconsider. Briefly is the operative word. These windows close as the new assumptions harden into the next default, and the firm that set them becomes the next incumbent.
This is where a firm's public posture stops being branding and becomes strategy. The right to frame a category is not claimed by asserting a frame. It is earned by standing behind a position in public, long enough, and consistently enough, to be believed when the market finally looks up. A thesis-led posture is the mechanism for converting a point of view into standing. It is slow on purpose, because the slowness is the proof. A position held across cycles, through inconvenient quarters and easy ones alike, accumulates the one asset a model cannot generate: a record the market can check. That is the discipline behind how we publish at BOST, and it is the same discipline we look for in any firm that intends to lead rather than follow a category.
None of this makes the mechanics optional. The frame defines what counts as the outcome; the work still has to deliver it. Both compound, and neither is fully automatable, which is why both are worth holding. But the order of operations has inverted. The strategist's job is no longer to capture demand more efficiently than competitors, because efficiency at demand capture is now widely available and therefore no longer scarce. The job is to define what the market considers worth demanding, then stand behind that definition long enough to be believed.
The companies that shape the question will beat the companies that only optimize the answer.
Drucker told us to create the customer. Christensen told us to understand the job. AI is telling us which of the two instructions was the real moat. The mechanics of capturing demand have become a commodity that anyone can buy. The act of shaping demand, of deciding what a market counts as a problem and standing publicly behind that definition until it is believed, has not. For a short window, while the old frames are loose, that act is also available to challengers. The firms that move into the pre-demand layer now will define the questions the rest of the market spends the next decade answering.