Nothing Your Business Does Is Novel

In 2014, Oliver Gassmann and two colleagues at the University of St. Gallen published the result of a multi-year effort to catalogue how companies actually reinvent themselves. They had gone through decades of business-model innovation hunting for the genuinely new: the moves that let a company change the game rather than just play it harder. What they found, and published in The Business Model Navigator, should have reframed every strategy offsite held since: roughly 90 percent of business-model innovations are not new. They are recombinations of 55 recurring patterns that already existed somewhere else. Razor-and-blades. Freemium. The two-sided marketplace. Subscription. Pay-per-use. Fifty-five patterns in total, mixed and matched across industries, and nine times out of ten the “innovation” is one of them lifted out of one context and dropped into another.

Read that number again, because it is doing more work than it looks. Ninety percent is not a claim that business is unoriginal. It is a claim about where the originality lives. The 55 patterns are the components: the known parts, sitting in a collective playbook, available to anyone who reads a case study. The innovation is never in the pattern itself; it is in the recognition that a particular pattern, moved into a particular market under a particular set of pressures, suddenly works where it hadn’t before. The novelty in business was never in the component. It lives in the constraint (the specific market, the specific adversary, the specific commitment you can’t take back) which is the only part that keeps changing.

What follows: why component-level novelty in business went nearly extinct without anyone sending the memo, where the real novelty relocated to, and why that relocation decides which half of your own work you should be handing to a machine and which half you should be guarding with your life.

The strongest version of “business is still full of genuine invention”

Let me give the invention-is-everywhere view its best shot, because a weak version is too easy to knock down. Business has produced genuine new categories inside a single working career. Software sold as a service instead of a shipped disc. The low-cost airline that made flying cost less than the train. The smartphone-era marketplace that put a stranger’s spare room or spare car into your hand in ninety seconds. The streaming bundle that unwound a century of how entertainment was sold. Someone had to see each of these, and seeing them was hard, and calling that work “recombination” to the face of the person who did it feels both wrong and a little insulting. Concede all of it. It is real invention, and it is not going away.

The part that does not survive a close look is the assumption that the invention lived in a component. Take the low-cost carrier. Southwest recombined a handful of known operational choices (one aircraft type, no assigned seats, secondary airports, fast gate turns) and invented not one of them. What it saw was the constraint. Southwest had already proven the model flying intrastate in Texas from 1971, where it sat beyond the reach of the federal regulators who dictated fares and routes everywhere else; the U.S. Airline Deregulation Act of 1978 then tore down that forty-year regime and let the proven pattern expand nationwide: deregulation that Southwest’s own success had helped bring about. Two decades later Ryanair copied the playbook almost line for line and became Europe’s largest airline by passengers: the same components, the same recognition, a new constraint as Europe liberalized its skies through the early 1990s. The pattern was portable. The seeing was the work.

Or take software as a service. Renting software by the hour was not new when Salesforce launched in 1999; it existed as time-sharing in the 1960s and again as the “application service provider” model in the late 1990s, and the ASP wave mostly failed. What changed was not the component. It was the constraint: always-on broadband and a browser on every desk finally made “software you never install” hold together, where a decade earlier it hadn’t. Nobody invented hosted software. Someone recognized the exact moment the old pattern would finally carry weight.

So the honest version of the steelman is this: invention in business is alive and well, and it almost always lives one level up from where we keep looking for it: in the constraint, not the component.

The category error

The mistake that follows is small to state and expensive to make. If you believe the scarce, valuable human contribution is inventing novel business components (a new pricing model, a new org design, a new funnel) you will staff for it, hire for it, promote for it, and spend your own scarce attention on it. And you will be optimizing for a supply that a recombination engine now produces faster and cheaper than any team you can assemble.

Because that is what a large language model is, at its core: the most capable recombination engine ever built, trained on essentially every business pattern anyone has ever written down: every pricing page, every org chart, every case study, every 10-K, every marketing playbook, every strategy deck that ever leaked. Ask it for a pricing model for a product, or an org structure for a forty-person go-to-market team, or three positioning angles for a crowded category, and it returns a competent recombination of the known patterns in seconds. That is not the hard part. It is the part the machine is structurally best at, and, per St. Gallen, the part that was already 90 percent recombination when humans did it by hand.

The genuine novelty, the part that was never in the training data because it did not exist yet, sits in four places, and all four are constraints. Your specific market: the actual customers, their actual willingness to pay, the channel that actually reaches them. Your adversaries: the competitor who reprices the day after you launch, the counterparty across the negotiating table, the regulator who can rule against you, the fraudster who adapts to every control you ship. Your irreversible commitments: what you have already signed, published, promised, hired, and positioned around, none of which the machine knows, all of which bind your next move. And your definition of success: what “good” actually means for your company, which is almost never the generic version.

None of those four is a thing you invent. Each is a thing you recognize and specify. The scarce skill was never conjuring a new business component out of nothing. It was seeing the constraint clearly enough to say what the recombination has to survive.

You can watch the category error cost real money whenever a proven component meets the wrong constraint. In November 2011, J.C. Penney hired Ron Johnson, the executive who had just built Apple’s retail stores into the most profitable square footage in retail, and Johnson imported the Apple playbook almost intact: everyday low prices instead of coupons, boutique shops-within-the-store, a premium in-store experience. The component was real and proven; he had run it himself. It was deployed against the wrong constraint. J.C. Penney’s actual customers were coupon-driven bargain hunters, and pulling out the coupons pulled out the reason they walked in. In the fiscal year that followed, sales fell about 25 percent (roughly $4 billion) and the company booked a $985 million loss. Johnson was gone seventeen months after he started. The playbook was not wrong. The constraint was ignored.

What this means in practice

If the recombination is the cheap half and a machine is better at it than you are, the correct move is not to guard the recombination. It is to give it away (deliberately, at volume) and spend the attention you recover on the constraints.

For an executive that means a specific inversion of what usually gets rewarded. Stop paying your best people to produce the plan, the deck, the model, the campaign; the machine produces those now, in minutes, in a dozen variants. Start paying them for the sharpness of the constraint and the honesty of the definition. The person who can say “here is the customer we are actually serving, here is the move our competitor will make the week after we launch, here is what we have already committed that we cannot unwind, and here is precisely what good looks like” is doing the work that does not commoditize. The person producing the fifth polished version of the board deck is doing what tokens do now.

Not the deck. Not the model. The constraint.

It also changes what a good meeting looks like. Bring the recombination already done (cheap, disposable, several versions of it) and spend the scarce human hour on which constraint each version has to survive and which one quietly breaks it. The output was the easy part. It was always the easy part; production merely hid how easy, because a competent employee used to discover the real constraints incrementally, in the middle of the work, and adjust without telling anyone. Hand the production to a machine and that silent adjustment stops happening; the ambiguity a person would have resolved by judgment comes back to you as a confident, polished, wrong answer. That failure mode is a later post in this series. For now it is enough to notice that the judgment was never a byproduct of the typing. It was the job.

This is the same reframe AOCyber was built around. The company’s own line for it is that capability commoditizes; trust infrastructure compounds: a recombination is a depreciating asset the next model reproduces next quarter, while the constraints and definitions wrapped around it are what accumulate. AODex, its multi-model AI workspace, exists to make the recombination cheap and governed (a hundred-plus models behind one interface, with the persistent memory, knowledge bases, and cited sources that keep the output reviewable) precisely so your scarce attention is never spent producing the draft, and stays free for the part of the job that is actually yours. A companion series makes the identical argument for engineers working inside a codebase; this one is for everyone running the business around them.

What to do

Draw the line honestly through your own week. On one side, the recombination: the pricing draft, the three org-chart options, the campaign variants, the second and third and fourth version of a memo whose shape you already understand. Hand it over, at volume, and stop treating the ability to produce it as a scarce and precious skill. On the other side, the constraints: the market you are actually bound by, the adversary who gets a move after you commit, the promises you cannot take back, and your own definition of what good means. That side is yours, and the rest of this series lives there.

For now the correction is enough to change how you spend Monday. Stop asking whether your strategy is new. Almost none of it is, and that was always true; the machine simply made it undeniable by producing the recombination faster than anyone can pretend otherwise. Ask instead what constraint your strategy has to survive, because the patterns are old, the recombination is nearly free, and the constraint is the business.