Essay
There was a period, not long ago, when access to a capable model or a well-built system was itself a competitive edge. Fewer companies had it, fewer people knew how to use it, and simply having the capability separated you from competitors who did not. That period is closing quickly. The same underlying capability is becoming available to essentially anyone willing to pay a modest subscription, which means the capability itself stops being a source of advantage the moment everyone has it.
This is a familiar pattern in business history, but it is happening faster and more completely than previous waves of technology diffusion. Whenever a capability becomes universally accessible, the advantage does not disappear. It migrates, to whatever remains unevenly distributed. And what remains unevenly distributed, even in a world where everyone has the same tools, is judgement, taste, empathy, narrative and the ability to earn trust in a room.
Capability is converging. Judgement is not
Two companies with access to the same systems can produce work of similar technical quality. What separates them is which one knew which problem was worth solving, which output was actually good enough to ship, and which piece of feedback from a customer mattered more than the ten that did not. None of that is a capability question. It is a judgement question, and judgement does not converge just because the tools underneath it do.
Judgement is built the slow way: through experience, through being wrong and correcting for it, through pattern recognition earned over years inside a specific industry or a specific customer base. It cannot be purchased as a subscription, and it does not improve simply because the tools around it got faster. If anything, faster tools raise the cost of poor judgement, because bad decisions can now be executed and distributed at a speed that used to be impossible.
The machine can generate options faster than any team in history. It still cannot tell you which one is right for the person in front of you.
Critical thinking becomes the actual bottleneck
As generating plausible answers becomes trivial, the scarce skill becomes the ability to interrogate a plausible answer and ask whether it is actually correct, actually appropriate, and actually aligned with what the situation requires. That is a different muscle from producing the answer, and it is one that many organizations have quietly allowed to atrophy while optimizing for speed. Fluent output is not the same as sound judgement, and a workforce that has stopped practising the second because the first is now so cheap is trading a long-term asset for a short-term convenience.
Critical thinking here is not an abstract virtue. It is a commercial one. The company whose people can tell a good answer from a merely confident one will make fewer expensive mistakes, catch problems earlier and keep the trust of customers who eventually notice the difference between a genuinely considered response and an efficiently generated one.
Empathy, narrative and taste do not scale by subscription
Empathy is understanding what someone actually needs, often before they can articulate it themselves, and responding in a way that acknowledges them as a person rather than as a query. Narrative is the ability to make a set of facts mean something to someone, in a way that moves them to act. Taste is the accumulated judgement to know when something is good, not because it followed a formula, but because it feels right against a standard that took years to develop. All three of these remain stubbornly human, not because a machine cannot approximate the surface of them, but because the approximation is detectable, and people are increasingly good at detecting it.
- Would a customer describe an interaction with us as something a person clearly cared about?
- Does our work carry a point of view someone had to actually hold, or a synthesis of what already exists?
- Can our people tell the difference between a fluent answer and a right one, under pressure, without checking twice?
- If the tools disappeared tomorrow, what would still make someone choose us?
Purpose and trust decide who gets chosen
As functional differences between competitors narrow, because everyone has access to comparable machine capability, the decision of who to buy from, who to work for and who to follow increasingly rests on purpose and trust rather than on specification. People choose organizations and leaders whose intentions they believe they understand, and that belief is built entirely through human behaviour over time: consistency, honesty when it is inconvenient, and a demonstrated willingness to prioritise something beyond the immediate transaction. No system can manufacture that belief on a company’s behalf. It has to be earned by the people inside the organization, repeatedly, in view of the people deciding whether to trust them.
Leadership is the advantage that compounds
The organizations that will separate themselves are not the ones with the most advanced deployment of these tools. Most competitors will eventually have comparable access to comparable systems. The separation comes from leadership: people capable of setting a direction worth following, making a hard call under uncertainty, and holding a team together through the disruption that adopting these tools inevitably causes. That kind of leadership was never automatable, and the fact that everything around it has become faster has made its absence more visible, not less.
Do not mistake this shift for humans being handed a consolation category of skills to retreat into while machines take the rest. The work that was always the actual source of competitive advantage — the judgement, the empathy, the narrative, the trust — has simply stopped being hidden behind the cost of production. Now that production is nearly free, what remains visible is what was always doing the real work.