Here is the advice most people are getting right now: take courses, get certified in AI tools, learn to prompt, add “AI experience” to your CV. None of this is wrong, but it may address the wrong opportunity.
People are wondering: how do I stay relevant as AI gets better? The question worth asking is: what do I know how to do that AI cannot replace, and how does that combine with genuine breadth across adjacent domains to make me irreplaceable?
Those are not the same question. The first has a shallow answer. The second has a structural one. The answer to the second is the same shape it has been for thirty years: the t-shaped professional.
Tim Brown, the designer and IDEO chief executive, used the T-shape to describe what the best collaborative designers looked like. The vertical bar is depth: the area where you know more than most, where your expertise is real rather than performed. The horizontal bar is breadth: the adjacent domains where you have genuine literacy, not just passing familiarity. You can follow the argument, ask a useful question, spot the gap that the specialist inside the domain is too close to see.
Both bars matter. The T-shape is not a metaphor for being a generalist who knows a lot about everything — that is the —, not the T. It is depth plus breadth, in combination.
What AI changes is the cost of having depth without breadth. It is also what made the Renaissance Professional an argument before AI made it a requirement.
A pure specialist — someone with the vertical bar but no meaningful horizontal bar — was already at a ceiling in complex organisations. They had expertise, but they could not cross the boundary where the real value was created: the intersection between functions, between the team with the problem and the team with the solution. The specialist delivered. The T-shape translated.
That dynamic existed before AI. What AI does is accelerate it, and add a second pressure.
First: AI is now capable of performing at the technical moat of many specialists. Not perfectly, not yet universally, but well enough that the moat is shallower than it was. A developer who only writes code in one framework, a data analyst who only runs one class of analysis, a product manager who only knows one vertical — each is more exposed today than three years ago. Not because their depth is worthless. Because depth alone is not the scarce resource it was.
The second pressure is less discussed: AI does not just replicate depth. It can cross domains and surface patterns from adjacent fields faster than any specialist can read. A specialist who could once compensate for narrow depth by being at least adjacent to something broader is no longer differentiated by that alone.
What AI cannot do is integrate. Not synthesise in the sense of combining outputs — it can do that. Integrate in the sense of a person who has operated under real conditions in multiple domains, who brings the judgment of genuine cross-domain experience to the question being asked right now. That capability lives in the T-shape. It does not live in the model. It is the same argument You Can’t Build a Centaur makes from a different angle.
This series makes that argument across five posts, publishing daily 8–12 June.
Why AI Is Killing the Pure Specialist (8 June) — which specialists are most exposed, why, and how the erosion happens at the edges of depth before it reaches the core.
Rebuilding Your T (9 June) — most attempts to build breadth fail because people approach it from the outside. There is a method that works and one that looks like it works but does not.
The T-Shaped Advantage (10 June) — what the T-shape actually gives you that depth alone cannot: position at the intersections, where the valuable problems live.
The T-Shaped Organisation (11 June) — the same logic applied to teams. Organisations built around T-shaped breadth are structurally better equipped to work alongside AI than those that are not.
Integration Is the Skill AI Can’t Replicate (12 June) — the closing argument: the skill at the intersection of the two bars is not a combination of them. It is something categorically different, and the most direct answer to what actually survives.
The AI course you take, and the certification you earn, are signals. They tell your employer you are paying attention. They do not tell your employer — or you — that you have solved the structural problem.
The structural problem is that the value of any capability is a function of what surrounds it. Depth surrounded by nothing is a moat that is getting shallower. Depth surrounded by genuine breadth, in a person who has learned to work at intersections — that combination is getting more valuable, because AI makes the individual components easier to acquire.
The parts are cheap now. The person who holds them together is not.
Leave a Reply