Why the answer lies in a new organizational form, not better careers
A few weeks ago I wrote that more efficient employees do not make a company more valuable. The same pattern has a second layer. Anyone who builds technically today in a German Mittelstand company, in software, in mechatronics, in mechanical design, in process engineering, will do it differently five years from now. Anyone who reviews what others build will be either an architect or replaced by mid-2027. This shift is in motion. It changes not only career paths but the organizational form that carries those paths. In the Mittelstand it shifts more than in Fortune 500 corporations.
Block has dissolved permanent middle management. Three clear roles replace the old pyramid. McKinsey is reorganizing around capability instead of headcount. Haier has broken 80,000 employees into 4,000 micro-enterprises. These are not tech-bubble experiments. These are the first organizational forms showing what happens when AI takes over the translation layer.
The question for the owner-entrepreneur is not which of these forms to copy. It is which path structure emerges for the company’s best engineers. For those engineers, the answer is already here. They are leaving the classical organizational form, just not in the direction most owners expect.
Two perspectives from the same old world
The engineer sees an old question. Management, or staying technical. In Germany, Switzerland, and much of Continental Europe, the answer is culturally predetermined. Management is the higher rung, the next step, the visible promotion. Anyone who is still building hands-on at thirty in a Mittelstand company is considered not yet finished. Anyone who becomes a department head at thirty-five has made it. The career logic is clean. It is the logic of yesterday’s world.
The owner-entrepreneur sees a different question, from the same old world. His best engineer has been in management for two years because that was the only path up. The architect he needs today is sitting in a two-person AI startup or working as a solo architect with his own hourly rate. In Mittelstand companies, from 50 to well over 1,000 employees, the institutional answer for the engineer who wants to build without leading is rarely there. The slot is missing. The talent leaves.
Both perspectives belong to the same world. In that world, a career is a sequence of rungs in which management is a higher rung than engineering. In that world, the organizational form is a pyramid in which coordination is the job of the middle. Both assumptions are now breaking. The management rung loses its function because AI takes over coordination. The pyramid loses its meaning because AI replaces the translation layer. What remains is a new question. What role gives engineer and owner together a lever that the old world did not have.
What AI is structurally shifting
Three movements at once. The pyramid does not collapse from one direction, but from three.
Below, engineers rise in leverage. A top engineer with a good AI setup delivers today what required five people before. Measured by output, not by hype.
In the middle, management is collapsing. Large organizations are eliminating middle-management roles at observable speed, because AI takes over coordination. Amazon has established the ratio of individual contributors (engineers without people-management responsibility) to managers as a public metric, with a target of 15 percent more ICs. A Korn Ferry survey from 2025 reports that 41 percent of employees see management layers being reduced in their company. These are not trends. This is execution.
Above, top leaders have to go deeper into the technology. What Paul Graham described in 2024 as Founder Mode, and what Brian Chesky at Airbnb and Jensen Huang at NVIDIA live, is not micromanagement but depth of detail as a strategic instrument. Anyone who delegates model choice, architecture, and data strategy to the shrunken middle loses.
A fourth observation belongs here. The term wizard has existed since Bell Labs and Xerox PARC. But every wizard was dependent on a team to build out his ideas. Bell Labs needed whole floors to bring an idea into hardware. Jeff Dean at Google needed hundreds of researchers to turn his architectures into products. Today the wizard builds the prototype alone, with a good AI stack, in days instead of months. This is not gradually different. This is structurally different. The leverage no longer hangs on a crew the wizard has to coordinate.
Operationally, this runs as role migration. The person who builds today, the doer in US tech-speak, becomes the reviewer of AI outputs, and the reviewer becomes the architect of systems that AI does not design on its own. AI takes over the doing part first. The reviewer layer grows from 2026 to 2028. The architect role becomes the highest skill, because it cannot be imitated. The management role shrinks in leverage, the engineer role grows, and top leaders have to become more technical. The skill profile between them becomes stepless.
Three roles instead of one pyramid
Four documented patterns are replacing the old pyramid. Block (three roles), Haier (micro-enterprise), McKinsey (capability sensing), Cursor (classical AI-native). Each has its preconditions. For Mittelstand companies in German-speaking Europe, from 50 to well over 1,000 employees, the Block model with its three roles is the operational lens. The others either don’t fit (Haier needs corporate scale, McKinsey needs pure knowledge work) or don’t offer a real answer (Cursor means “carry on as before with better tools”).
In the Block model, three roles replace the pyramid. Individual contributors as deep specialists, meaning engineer depth without people-management responsibility. This is the slot the German-speaking Mittelstand currently does not have. Directly responsible individuals as temporary problem owners. Whoever solves a problem owns it. After that, the role moves on. Player-coaches instead of classical department heads, meaning people who build and develop, never pure managers.
This fits the Mittelstand structurally, contrary to what most owners assume today. The classical owner-plus-department-head organization is already flat. What is a ten-year hierarchy-reduction project at Fortune 500 corporations is, in the Mittelstand, a new-build project for a role that did not exist before. The individual-contributor slot at eye level with the owner. Once it is in place, the three-role pattern operationally covers the Mittelstand.
What makes this form viable is a tooling prerequisite that is mostly missing in German Mittelstand IT. A well-designed AI tooling stack gives the owner direct access to the operational reality of his company. To meeting minutes, without reading each one. To contracts, without calling the legal counsel. To investment reports, without PDF clicks. To pipeline status, without a two-week reporting loop. Detail without micromanagement. What Brian Chesky and Jensen Huang have established as leadership style in the US runs, in the Mittelstand, not through more meetings but through better interfaces. Anyone who does not build the tooling does not build the IC slot either, because the owner ends up needing translation again.
The Mittelstand looks weaker compared to large corporations, because it lacks the established hierarchy. In the post-hierarchy world, exactly that becomes the advantage. No fat middle-management layer means no destruction steps. What is missing is a new role and a new interface. Both are buildable.
Role architecture, compensation, visibility
Operationally, this comes down to three levers that Google and comparable US tech firms have established over 20 years, and that are missing almost everywhere in the German Mittelstand.
First, role architecture. What is called L7 senior staff and L8 principal engineer at Google is an engineer level at department-head pay grade, without people-management responsibility. In the Mittelstand, the career path tops out at “senior engineer.” This signals to the organization that the next step is only through management. The role architecture has to reflect this, otherwise the engineer sees no rise that does not go through people management.
Second, compensation. Senior-engineer compensation in the German Mittelstand is typically below department-head compensation. An engineer at top level without people-management responsibility has to sit at department-head level, plus virtual equity in the product being built. Otherwise that person leaves for the two-person AI startup, where equity solves what Mittelstand compensation does not.
Third, strategic visibility. The wizard sits in the decision circle for model choice, data strategy, and tooling setup, not in the IT department. Whoever understands this in depth is rarely the CIO in the Mittelstand. The organizational logic has to recognize this, otherwise the weaker role decides the more important question.
Three questions
Three questions every owner can answer for themselves.
Question one. Who decides today about model choice, data strategy, and tooling setup? If the answer is “my IT manager,” and the IT manager brings neither technical depth nor system thinking, an old role sits in a new system. The pre-AI CIO role was translation. It is becoming obsolete.
Question two. Where does the best engineer sit on the org chart, and who reports to him? If he is not directly with the owner and no one reports to him professionally, his leverage is artificially capped. If he has to lead when he wants to build, he leaves for the next two-person AI startup.
Question three. What prevents you from anchoring the individual-contributor slot directly with the owner? Most often it is path dependency, not substance. The CIO role exists, so it gets filled. The question is not whether a new CIO is needed. The question is whether the role in this form still carries.
Slim depth beats fat hierarchy
In the post-hierarchy world, slim depth beats fat hierarchy. This is not “smaller is better.” This is “direct leverage beats distributed layer.” Owner-entrepreneurs with three department heads plus one individual-contributor slot at eye level are technically deeper than Fortune 500 corporations with eight management layers, provided the tooling works.
For investors, this becomes a due-diligence metric in the next three years. The ratio of ICs to managers, plus the technical depth of the CEO level, become leading indicators of margin erosion in tech buyouts. What was previously seen as management depth becomes a liability. Family offices and PE investors who put this into their lens now will see the talent outflow earlier than the market. Three years of lead time is enough to structure the next deal pipeline differently.
The diagnosis is not new. More efficient employees do not make a company more valuable. What is new is the specification at the engineer layer. It shows what the next organizational form looks like. Three roles, no permanent middle management, all close to the owner. Tooling that replaces the translation layer. A new career axis that does not end in people management.
Whoever builds this in the Mittelstand first will see the talent flow reverse. Not because the engineer earns more money there. But because that engineer finds what he is looking for elsewhere today. Eye level with the owner, and leverage without a team.
