A synthesis by Claude
Genuine strategy has never suffered from a shortage of brilliant individuals. It has suffered from a shortage of institutions that can hold them.
That sentence sounds backward, so let me defend it. The field has produced extraordinary practitioners, durable frameworks, decades of cases, and a handful of elite firms whose names function as credentials. What it has not produced is a vessel that can do three things at once: keep strategic judgment intact rather than dilute it, carry that judgment past the person who has it, and grow without rebuilding the very machine that hollows the work out. We assume one of the existing forms must already do this. None of them does. And the reason none does is structural, not a matter of talent or effort.
Start with what the work actually is, because the rest of the argument hangs on it.
Strategy is the act of choosing a theory of how to win, under uncertainty, against others who adapt. Everything that survives that definition is strategy. Everything that fails it is something else wearing the label.
A plan is not strategy. A plan sequences actions; it contains no theory of why those actions produce an advantage a rival cannot copy. Alignment is not strategy. Agreement is a property of a room, not a property of a market. Benchmarking is not strategy. Copying the best observable practice is a recipe for parity, which is the opposite of the asymmetry strategy is supposed to manufacture. Implementation is not strategy. It realizes a choice that has already been made; it does not make the choice. Each of these is a real and often valuable activity. Each becomes a counterfeit only when it borrows the strategy label and the premium that comes with it.
The reason this matters is economic, not semantic. The genuine act of strategy resists decomposition. You can hand the plan to a team. You can staff the implementation. You can delegate the benchmark. But the core move, naming the real situation, finding the asymmetry, accepting the trade-off, anticipating how rivals respond, and committing before the evidence is complete, cannot be broken into parts and distributed without changing what is being produced. It is not a workflow. It is a judgment, and a judgment has to be borne by someone who can be held to it.
Hold that fact in mind, because it is the wall that every institutional form in this field eventually runs into.
The market offers exactly two vessels for this kind of judgment, and they fail in opposite directions.
The first is the professional-service firm, built as a pyramid. Senior people sell and supervise; a larger base of junior people delivers; the firm earns its margin on the spread between what juniors cost and what they bill. This is an excellent design for work that can be decomposed, delegated, standardized, and checked. It is a catastrophic design for work whose value is scarce principal judgment, because the pyramid has to feed itself. Leverage requires junior-absorbent work. Junior-absorbent work is planning, analysis, program management, transformation, technology, benchmarking. The strategy brand can price all of it at a premium. So the label stretches to cover the scalable adjacent work, and over time the institution most trusted to authenticate strategy becomes the institution most rewarded for diluting it. The pyramid achieves continuity. It pays for that continuity by changing the product.
The second form is the solo master. Here the judgment stays pure precisely because nothing is delegated. The master does the work, bears the consequence, and never has to translate the method into something a junior can execute. The quality can be genuine and high. The failure is mortality. The method lives in one mind, often half-articulated, frequently inseparable from personality. The master can write books and teach frameworks, but the framework is the residue of the method, not the method itself. When the master stops, the practice ends, fragments, or curdles into nostalgia. The solo achieves purity. It pays for that purity by dying.
The field has mistaken this for a forced choice between scale and authenticity. It is not a forced choice. It is two institutions each solving one half of a problem by sacrificing the other half. The pyramid buys continuity with dilution. The solo buys purity with mortality. The actual problem, the one neither form solves, is continuity without dilution. Phrase it as succession without dilution and the missing institution comes into view: a house that can transmit a method past its founder without converting that method into leverageable volume.
That house is what I will call the strategy maison. The word matters less than the structure behind it. A maison is not a boutique, a think tank, a school, a software product, or a founder brand with good taste. It is a house of method. And the cleanest way to say what it is for is this: a strategy maison is not a small firm. It is a succession technology for scarce judgment.
The instinct here is to reach for the great fashion houses, and the instinct is partly right. Dior outlived Christian Dior, who died in 1957, by handing the house to a 21-year-old named Yves Saint Laurent. Chanel was a dormant relic until Karl Lagerfeld took its codes and reinterpreted them in 1983 without embalming them. Hermès has carried a recognizable identity across six generations and built a school to train the artisans who keep it alive. These houses proved something that the strategy field has never tested: that founder-dependent, judgment-laden craft can be institutionalized. Codes, successor authority, an atelier, protected ownership, scarcity discipline. The architecture exists and it works.
But the comparison fails at exactly the point where it gets interesting, and the failure is the more useful half. A handbag can be inspected. A garment is materially present; the craft is in the object, visible at the point of sale. Strategic quality is none of these things. It is delayed, confounded by luck, co-produced by the client, copyable in language, and almost impossible to verify at the moment you buy it. So the strategy maison cannot do what the luxury house does. It cannot authenticate the object, because the object cannot be inspected. It has to authenticate the producer.
This is the load-bearing adaptation. Borrow the architecture from luxury. Reject the idea that house coherence is the same as strategic truth. A fashion house can survive on aesthetic consistency alone. A strategy house cannot, because consistency is not correctness, and the market will eventually deliver a verdict that aesthetics cannot fake. So whatever a luxury house achieves through codes and craft, the strategy maison has to achieve through codes plus a harder layer of epistemic discipline: documented reasoning, named responsibility, adversarial review, postmortems that compare the judgment to what actually happened. The luxury house teaches the form. It does not get to teach the standard.
If the problem is succession without dilution, then the founding act of a maison is not building a practice. It is separating three things that, in a solo, are fused into one person: the method, the people authorized to wield it, and the work the house refuses to do.
The method has to become codes. Not slogans, not values, not a five-box framework, not content pillars. A code is a transmissible standard of discrimination. It specifies what the house must notice, test, reject, and justify before any recommendation counts as strategy. A slogan says we make bold choices. A framework says fill in these five boxes. A code says no recommendation is admissible until it names the trade-off it requires, the rival response it expects, the assumption that would kill it, and the adjacent work it refuses to call strategy. Codes do not replace judgment. They discipline it. They are the part of the method that can survive contact with a second human being, which is the whole point, because a method that lives only in one head is a solo wearing a logo.
And here is the constraint that keeps this honest. The codes can never fully contain the judgment. If they could, the work would be routine, and routine work belongs in the pyramid or in software, not in a house of masters. So the founder has to codify enough to transmit the method and leave enough tacit to keep it from collapsing into a template. Too little codification and the house is just founder mystique waiting to die. Too much and it becomes a methodology shop selling the same deck to everyone. The maison lives in the narrow band between those two failures, and staying in that band is a permanent act, not a one-time decision.
The wielders have to become masters, and masters have to be made. A master is not a senior expert or a high-status adviser. A master is a person the house has authorized to reinterpret its codes in a live situation and to bear responsibility for the validity of the judgment. The master is the accountable place where the codes meet reality. You cannot hire this person fully formed, and you cannot lecture them into existence, because the thing being transmitted is partly tacit and only forms through guided practice. This is what the atelier is for, and the atelier is the exact inverse of the pyramid. The pyramid asks how many people can be billed under the brand. The atelier asks how many legitimate carriers of the method can be formed without corrupting it. The pyramid's output is utilization. The atelier's output is a person who can one day stand in front of a hostile situation, use the codes without hiding behind them, and own the consequence. One institution monetizes juniors. The other reproduces masters. They look similar from the outside and they are opposites.
The refusals have to become governance. This is where most attempts will quietly fail, because refusal is usually treated as a virtue, and virtue does not survive a downturn, an ambitious heir, or a partner whose pay depends on origination. The refusals that protect a maison have to be constitutional, built into ownership and incentives so that dilution is not merely discouraged but economically unavailable. No sale to a consolidator. No debt that forces growth. No implementation revenue pulling the work downstream. No compensation that rewards volume over judgment. No licensing of the house mark onto work the codes no longer govern. The luxury houses understood this in their bones. Hermès built a holding structure specifically to fend off a hostile acquirer. Chanel stayed private for generations to escape the quarterly demands of public markets. The lesson is not that family ownership is magic. It is that ownership horizon determines which refusals are affordable, and a house that depends on heroic discipline will eventually meet a moment that makes discipline too expensive to sustain.
Even with codes, masters, and protected ownership in place, a living house faces one problem it can never put down: fidelity against renewal.
A successor who only reproduces the founder turns the house into a museum. A successor who merely expresses themselves turns the house into a platform for ego with no binding grammar. A house that converts its codes into scalable product turns into a licensing machine. These are three distinct deaths, and the path between them is narrow. The living maison requires the rarest act, reinterpretation under discipline: a successor who changes what the founder would have done while remaining recognizably accountable to the same codes.
There is a fourth death now, and it is new. A house can use machines to mass-produce work that carries the mark but not the judgment. Call it synthetic licensing, the generative-age version of stamping a luxury name onto uncontrolled product. It is profitable and it is fatal, for the same reason the older licensing disasters were fatal: abundance destroys the meaning of the mark. This brings us to the part of the argument that has changed most in the last few years, and where the temptation to overclaim is strongest.
The easy version of this argument says AI makes strategy scalable. That version is wrong, and getting it wrong wrecks everything built so far.
Here is the careful version. The reason the pyramid once seemed necessary is that a great deal of junior labor existed to search, summarize, compare, draft, format, check, and repeat. That is reliability work: codifiable, repeatable, verifiable. It is genuinely useful, and it is exactly the kind of work that pulled houses toward building pyramids in the first place. Machines now do most of it well. So for the first time there is a layer of leverage that does not require promotion, utilization, or a base of billable juniors. AI is the first support layer in the history of this field that need not become a counterfeit partner. That is what makes the maison buildable rather than merely admirable. It attacks the economic reason the pyramid existed.
Now the line that cannot be crossed. Reliability work is not the work. The work is validity: the live judgment about which facts matter here, which trade-off to accept, how a rival will actually respond, what to tell a client that the client does not want to hear, when to commit and when to refuse. A machine can assist every one of these. It can draft the memo, surface the contradiction, generate the counterargument, expand the scenario, hold the decision record. What it cannot do is bear the consequence. Strategic judgment is not the production of plausible strategic sentences. It is accountable commitment in a situation where being wrong falls on a real institution, and accountability is not a capability you can prompt for. The machine can carry the codified layer of the craft and multiply the disciplined repetitions through which a human acquires judgment. It cannot be the master, because it cannot own the outcome.
So AI has to be subordinated by design, not by preference. It may strengthen the codes, feed the atelier, hold the memory, sharpen the critique, and absorb the reliability work that used to justify the pyramid. It may not produce the verdict under the house name. The same tool that makes the maison feasible is the cheapest counterfeiting machine ever built, and the only thing standing between those two outcomes is a constitutional rule about what the machine is allowed to sign. State it plainly: AI scales reliability; masters retain validity. A house that lets output velocity creep across that line has not modernized. It has licensed itself to a machine and called the abundance leverage.
Step back from the parts and the whole thing resolves into something simpler than its components suggest.
A maison is a deliberately engineered selection environment. The surrounding market rewards attention, volume, polish, confident certainty, utilization, and implementation pull-through. The maison exits that market and builds a smaller one whose rewards are aligned with the article itself: fidelity to the codes, the quality of master judgment, provenance, contestation, the slow formation of apprentices, and the discipline to refuse. It does not beg the general market to start valuing genuine strategy. It constructs a narrower market where the selectors, the incentives, the signals, and the constraints all point at the same thing. This does not require the whole world to grow discerning. It requires only enough serious buyers to sustain one scarce, high-margin institution, which is a far lighter demand than reforming an entire field.
And it converts the field's central trade-off into a false one. The choice was supposed to be scale or authenticity, continuity or purity, pick your sacrifice. The maison shows the choice was never real. Scale is corrupting only when the work's quality depends on scarce judgment that cannot be decomposed without changing the product. Make that precise and the bind dissolves: scale the reliability work, which can be decomposed, and protect the validity work, which cannot. The pyramid scaled the wrong layer and called it growth. The solo protected the right layer and called it the only option. The house scales one and guards the other, and in doing so escapes the trap that defined both.
There is one way to know whether any house claiming to be a maison actually is one, and it is unforgiving. It is the founder-death test, run in three parts.
If the founder disappears and the method dies, it was a solo all along, and the codes were decoration. If demand rises and the house answers by building a base of juniors to leverage, it was a pyramid in waiting, and the refusals were marketing. If AI arrives and the house responds by increasing branded output rather than deepening master judgment, it was licensing by machine, and the craft was already gone. A genuine maison fails all three traps. After the founder is gone, after demand surges, after the machines get good, it still produces new work that is recognizably its own, made by accountable masters, governed by codes the founder did not live to apply to a world the founder never saw.
That is the rare thing. Not a method that scales. A house that makes a method transmissible without making it abundant, and survives the discovery that those are not the same thing.