Why experts and executives are both running out of time
For decades, a functional equilibrium existed inside almost every major organisation. Senior experts and executives needed each other in a relationship of structured co-dependency. That equilibrium is dissolving. The information asymmetries that underpinned both power bases are being eroded simultaneously, faster than either group currently recognises, by a force neither group fully controls.
The people who understand this earliest, and move accordingly, will establish the new power positions. The people who do not will be displaced. Not by artificial intelligence. By the colleagues who saw it coming.
Part one: The equilibrium that no longer holds
Until recently, organisations ran on a clearly understood division of cognitive labour. Senior experts held power because they knew more than their executives needed to know. The gap between what the expert understood and what the executive could access was wide enough, and reliably enough maintained, to make the expert indispensable. That indispensability was the source of their organisational power, their commercial value, and their career security.
Executives held power because they could do something experts typically could not: define commercial problems in a strategic context, make judgements about the feasibility and priority of solutions, and translate organisational complexity into decisions that could be executed. The expert provided the depth. The executive provided the direction. Neither could fully function without the other.
This was not a hostile arrangement. It was genuinely symbiotic. The expert needed the executive to translate domain knowledge into commercial consequence. The executive needed the expert to access capability they could not themselves possess. The relationship worked precisely because each side held something the other could not easily replicate.
What the expert had that the executive could not get
The crucial asymmetry was not simply knowledge. It was the lived experience of expertise landing in the real world. An executive could read about a domain, attend briefings, commission reports. What they could not acquire was the accumulated pattern recognition that comes from repeated direct exposure to the consequences of decisions in that domain. What works in practice versus what works in theory. Where the models break down. What the data does not capture. What the experienced practitioner knows without being able to fully articulate why they know it.
This experiential depth was the expert’s most durable asset. And for a long time, it was secure.
What the executive had that the expert could not easily access
Equally, the executive possessed something the expert typically lacked: a track record of operating at the level where commercial problems are defined rather than solved. The ability to hold multiple competing considerations simultaneously, financial, political, strategic, operational, and make judgements under conditions of genuine uncertainty, with incomplete information and real consequences.
Experts are trained to solve defined problems. Executives are shaped by experience to define which problems are worth solving and in what sequence. That is a different cognitive skill, developed through a different kind of exposure, and it was not easily transferable in either direction.
Part two: The dissolution – what AI is actually doing
The popular narrative about AI and employment focuses on task automation. Which jobs will be replaced, which roles are safe, which sectors face the most disruption. That framing is too narrow for the purposes of this analysis.
What AI is doing at the level that matters for this thesis is collapsing information asymmetry at scale and at speed. The expert’s knowledge gap – the reliable distance between what they know and what their executive can access – is closing. Not because executives are becoming experts, but because the cost and effort of acquiring sufficient domain understanding to challenge, question, and direct experts has fallen dramatically.
This is a structural change, not a cyclical one. The gap that took decades to build is being eroded in months.
The Karpathy signal
Andrej Karpathy is one of the few people alive who has been present for all three paradigm shifts that built modern AI. He was a founding member of OpenAI. He led the Autopilot team at Tesla. During his PhD at Stanford, he co-designed and was the primary instructor for CS231n, the university’s computer vision and deep learning course, which became one of its most popular classes and has since been watched by millions of people online.
His Software 1.0/2.0/3.0 framework is the clearest map currently available of what has structurally changed.
Software 1.0 is what almost everyone still means when they say code. Explicit human-written instructions that tell a machine exactly what to do. Software 2.0, a term Karpathy coined in a 2017 essay, describes the shift to machine learning, where the logic lives not in hand-written rules but in the weights that emerge from training on data. He watched this happen in real time at Tesla, where hand-coded computer vision logic was replaced by learned models week after week.
Software 3.0 is the shift that has just arrived. The programming language is now English. You describe what you want in natural language and a large language model executes the intent. The prompt is the programme. The barrier to producing sophisticated software behaviour has collapsed from formal language fluency, a skill fewer than one percent of humans ever acquired, to the ability to think clearly about what you want and describe it precisely.
The implication Karpathy draws is the one that matters most for this analysis. The bottleneck is no longer syntax or technical knowledge. It is knowing clearly enough what problem you are actually trying to solve. That is a judgement and domain understanding problem dressed as a technology observation. The expert who cannot think at the systems level cannot frame the problem correctly for the AI to work on. The executive who does not understand the domain cannot direct the intelligence effectively. Both are exposed by the same shift from execution to direction.
What this means for the expert’s power base
The expert who has built their indispensability on knowing more is discovering that knowing more is no longer sufficient protection. The executive who used to rely on them can now access enough domain understanding, through AI tools, through accelerated learning, through better information synthesis, to ask questions the expert was not previously required to answer.
The expert’s depth is not being challenged directly. Their monopoly on access to that depth is being challenged. And it is the monopoly, not the depth itself, that generated the power.
What this means for the executive’s power base
The executive faces a different but equally serious threat. Their advantage in problem articulation, the ability to frame a question clearly enough to direct others toward a useful answer, is being eroded as AI tools make sophisticated problem framing accessible to people who previously lacked it. The expert who masters the new tools and develops genuine big picture thinking starts to encroach on territory the executive considered secure.
More pressingly, the executive who cannot understand the domain well enough to direct AI effectively is losing the ability to make astute judgements about solutions they cannot evaluate. The judgement function, their most durable asset, depends on sufficient domain understanding to know whether an answer is credible. Without that understanding, the judgement becomes dangerously thin.
Part three: The arms race – the double threat
What makes the current situation genuinely novel is that both groups face threats from two directions simultaneously. Each is threatened by AI eroding their information asymmetry. And each is threatened by the other group moving to close the capability gap that previously separated them.
The expert who develops genuine judgement capability and big picture thinking starts encroaching on executive territory. The executive who develops real domain mastery stops being dependent on their experts. Both moves are rational responses to the same underlying pressure. The result is an arms race between two groups who are each other’s most immediate competitive threat, while both face a deeper structural challenge from the technology itself.
The Cuban observation
Mark Cuban, American entrepreneur, investor, and long-term student of where economic value actually collects during technological transitions, identified the mechanism that resolves this arms race clearly.
His observation, made in the context of the 33 million small and medium businesses that will never have AI budgets or AI experts, was deceptively simple. The wealth of the electricity era did not go to the engineers who built the generators. It went to the people who walked into dark factories and showed the owners where to plug in. You do not need to build the brain. You need to build the nervous system.
Applied to the expert-executive dynamic, the principle is direct. Raw intelligence, whether artificial or human, is worthless until it meets a precisely understood commercial problem. The people who survive the transition are not those who possess the most knowledge or the most sophisticated AI access. They are those who can place their capability precisely at the intersection of the technology’s potential and the specific commercial problem that needs solving.
Wealth collects where the brain meets the business. Not where the brain is built.
This is the principle that both experts and executives need to internalise. Domain mastery alone is not enough. Astute judgement alone is not enough. The value is created at the intersection, and the intersection requires both.
The Ternus signal
In April 2026, Apple’s board appointed John Ternus as Chief Executive Officer. The decision illuminates the thesis at the highest level of corporate life.
Ternus joined Apple in 2001 straight from a mechanical engineering degree at Penn. He has been at one company for twenty-five years. His title throughout has been some variant of hardware engineering. He has never run an OS team, an AI lab, or a services business. He has no LinkedIn profile because he has never needed to job search.
The board passed over Craig Federighi, who runs software. They passed over Eddy Cue, who runs services. They passed over Johny Srouji, who actually designs the chips. They chose the hardware engineer at a four trillion dollar company that had spent two years being publicly criticised about its AI capability.
The board’s read: AI is a vertical integration problem. The competitive moat is not the model. It is the silicon that runs the model. Specifically, Apple Silicon’s ability to run inference at the edge, on device, at a power envelope no competitor can currently match. Every iPhone shipping today runs a Neural Engine that processes AI inference locally at a fraction of the energy required by cloud-based competitors. That gap is widening.
Ternus has been present at every architecture review for the hardware and chip roadmap for over a decade. He did not need to learn what was coming. He scoped it. The board chose him not despite his deep domain focus but because of it. At the moment of maximum technological uncertainty, the person with the deepest mastery of the most strategically critical domain was the one most qualified to lead.
Tim Cook’s defining advantage was supply chain mastery. John Ternus’s bet is the stack.
The expert who spent twenty-five years going deep into one domain, and who understood precisely where that domain intersected with the most consequential commercial problem his company faced, did not need to reinvent himself. The board found him.
That is domain mastery as a power strategy in practice, at the highest level, publicly visible, and directly illustrative of everything this analysis has been building toward.
But domain mastery secured the appointment. The question the next few years will answer is whether Ternus has the political dexterity to succeed now he holds the authority. Whether the man who mastered the silicon can navigate the human system he has just been handed is a different question entirely, and one the board could not verify in advance.
Part four: The convergence – where both paths lead
The arms race has a destination, and it is the same destination regardless of which side you start from.
The expert who successfully repositions acquires sufficient big picture thinking to understand the strategic context their domain operates within, genuine judgement capability developed through exposure to commercial consequence rather than purely technical problems, and the ability to define problems rather than simply solve them.
The executive who successfully repositions acquires sufficient domain mastery to direct AI effectively in the areas that matter most to their commercial decisions, the experiential depth to know whether an answer is credible, and the ability to evaluate solutions rather than relying entirely on experts to do it for them.
Both capability profiles, when complete, describe the same person. Someone who combines deep domain understanding with astute judgement and the big picture thinking to connect both to commercial consequence. The distinction between expert and executive collapses into a single category: the complete strategic operator.
The domain mastery question
This is where the thesis encounters its most important unresolved question. If the convergence requires domain mastery, how many domains must an executive master? And how deep must the expert’s capability become before it constitutes genuine astute judgement rather than informed opinion?
The naive version of the thesis suggests that both groups need to become renaissance figures, broadly capable across multiple domains and deeply skilled in strategic thinking simultaneously. That is probably not achievable at scale, and may not be necessary.
The more useful formulation is this. AI itself becomes the multi-domain capability layer. The complete strategic operator does not need to master all relevant domains to the depth of a career specialist. They need sufficient domain understanding across the domains that matter to their specific commercial context to direct AI effectively in each, to ask the right questions, evaluate the answers critically, and know when the model is wrong.
This is a lower bar than full mastery but a substantially higher bar than current generalist executive thinking. Combined with astute judgement, the ability to operate where the problem is contested, politically loaded, and resistant to clean formulation, it becomes the complete capability profile that neither current experts nor current executives typically possess.
What cannot be automated
The capability that sits at the centre of this convergence, and that neither AI nor the other group can easily replicate, is the ability to operate effectively in conditions of genuine ambiguity. Where the problem itself is contested. Where the organisational and political texture of the institution shapes what is even possible. Where the right answer is not knowable in advance and where judgement, rather than analysis, is the operative capability.
This is not a marginal or residual human function. It is the function that generates the most value and carries the highest stakes. It is also the function that is hardest to develop, least susceptible to acceleration through AI tools, and most dependent on the kind of accumulated pattern recognition that only comes from repeated direct exposure to consequential decisions.
That is the territory where the complete strategic operator lives. And it is territory that very few people, from either the expert or executive starting point, currently occupy.
Part five: The window – and what it requires
The arms race has a timeline, and it is shorter than most people currently assume.
The transition is not arriving gradually. It is arriving unevenly, which is more dangerous. Some organisations and some individuals are already through the first wave of repositioning. Others are still debating whether the threat is real. The gap between those two groups is widening every month, and it is not a gap that effort alone will close once it becomes visible to everyone.
The window for first mover advantage is open. It will not stay open indefinitely. The people who move now, while the majority are still in early problem recognition or quiet denial, establish positions that will be very difficult to displace. Not because they are more talented. Because they moved first, and position compounds.
What the transition requires is not reinvention. That framing is both inaccurate and paralyzing. The complete strategic operator is not a new kind of person. They are a senior professional who has understood what the shift actually demands, developed the capabilities the new position requires, and had the political dexterity to carry that repositioning into consequence inside the systems they operate within.
Very few people currently occupy that position. The ones who do are not broadcasting it. They are simply harder to displace than they used to be, more valuable than their peers recognise, and already operating in territory that others have not yet identified as contested.
The question is not whether this transition is happening. It is whether you are moving before or after the people around you.
Part six: The emerging vocabulary – why language matters here
When a structural shift moves faster than most people can perceive it, precise language becomes a strategic asset. Naming something accurately is not merely semantic. It creates the cognitive handle that allows someone to see what they could not previously see, act on what they could not previously articulate, and move with purpose while others are still trying to describe what is happening to them.
Three terms have emerged from this analysis as the right conceptual handles.
‘Domain mastery’ describes the depth of understanding, combining technical knowledge with lived experiential pattern recognition, that makes an expert’s capability genuinely irreplaceable rather than merely expensive. It is distinguished from subject matter expertise by its emphasis on the experiential layer: the accumulated knowledge of how the domain behaves in real conditions, under commercial pressure, with real consequences. Domain mastery is what the executive cannot yet get from AI, and what the expert must preserve and extend as their primary strategic asset.
‘Astute judgement’ is the capability to read a complex situation accurately enough, seeing the interests, agendas, and pressures that others either miss or choose to ignore, and to determine the right response, at the right moment, with a clear understanding of what it will cost. It is distinguished from analysis by its operability under conditions that resist clean formulation. Where analysis requires sufficient information to reach a conclusion, astute judgement operates precisely where information is incomplete, interests are competing, and the situation will not wait. This is not intuition. It is disciplined perception combined with the willingness to act on what you see, even when what you see is uncomfortable.
‘Political dexterity’ is the implementation capability, the ability to move human systems toward outcomes under conditions of competing interests, imperfect information, and shifting power. It is what converts domain mastery and astute judgement into actual results inside real organisations. Without it, the complete strategic operator remains a capable individual who cannot translate capability into consequence. It is the third element, and in complex organisations, frequently the decisive one.
Conclusion: The complete strategic operator
The thesis resolves to a single destination.
The expert and executive roles, as they have existed for decades, are converging under the pressure of a technological shift that is simultaneously eroding both their information asymmetries and dissolving the co-dependent relationship that made both groups valuable to each other.
The destination is a new category of senior professional, the complete strategic operator, who combines domain mastery with astute judgement and the political dexterity to translate both into consequence inside complex organisations. This person does not need to know everything. They need to know enough, deeply enough, in the domains that matter to their specific context, and they need the judgement and political capability to operate where AI and analysis cannot take them.
Very few people currently occupy this position. The transition window is open. The arms race has begun.
The question for every senior expert and executive is simple: which side of this transition do you intend to be on, and how much time do you believe you have?
See also: A New World Demanding More
Colin Gautrey, May 2026