Essay
There is a specific kind of executive anxiety going around right now. It shows up as a race: work faster, produce more, learn every tool, prove you are not the slow part of the process. I understand the impulse. I also think it is a losing position, because speed of production is the one dimension where a person will never beat a machine.
The leaders who come out of this cycle stronger are not competing on the machine’s terms. They have moved to the part of the job that gets more valuable precisely because everything around it got cheaper.
Four things that do not automate
Critical thinking — not analysis, which is now partly automated, but the ability to ask whether the question itself is even the right one, and to notice what a very convincing answer conveniently leaves out.
Judgement under incomplete information. Every meaningful business decision gets made before the data is sufficient. Someone has to choose anyway and then own what follows. A model can quantify a risk. It cannot accept one.
Narrative. Organisations rarely move because a strategy is logically sound. They move because someone made the direction feel real, urgent and worth personal effort, and that transmission fails instantly the moment the person delivering it does not believe it themselves.
Meaning. People give discretionary effort to work that connects to something they care about, and no system manufactures that connection on a leader’s behalf — though plenty of internal comms decks have tried.
You cannot delegate the responsibility for a decision to something that cannot be held responsible.
The new management problem: mixed teams
The concrete change in the leadership job is that teams are no longer only human. A senior manager now allocates work across people and systems, and the two fail in opposite ways. People fail slowly, visibly, with warning signs along the way. Systems fail instantly, invisibly, and with total confidence right up to the moment it goes wrong.
That demands a new operating discipline: knowing which decisions require a human name attached, designing checkpoints where fluency alone is not accepted as evidence, and protecting the space where junior people still learn by struggling. Automate every difficult first draft away and you optimise this quarter while eliminating your bench for the next decade.
- Which decisions in my organization must always carry a human signature?
- Where are we accepting confident output without a competent reviewer?
- How do people still develop judgement if the hard first attempt is automated away?
Irreplaceable is not a personality trait
People misread the word constantly, hearing it as being indispensable to one company — a fragile, slightly anxious place to stand. What I mean by it is closer to the opposite: being valuable in a way the market keeps paying for, whoever you happen to be working with at the time. That is portable, and it survives a reorganisation that your job title never will.
Being difficult to replace has nothing to do with charisma or hoarding information. It comes from standing in the part of the system where human capability is genuinely required: understanding people well enough to predict them, deciding under pressure, carrying accountability, connecting work to a direction people are willing to follow on their own.
That position gets built deliberately, over years, not discovered by accident. Most careers were designed for a market that paid for output. That market is repricing toward judgement, and whoever makes the transition early will spend the next decade with an advantage the rest of the field is still pretending does not exist.
What to stop doing
Before any list, one honest audit. Take your last ten working days and split them into decisions you made and material you produced. If the second pile is heavier, you are competing with a machine on the machine’s terms, and no amount of tool fluency will rescue a job description that is mostly output.
Stop measuring your own contribution in volume. Stop treating tool fluency as a strategy — it is table stakes, and it will be commoditized within the year. Stop running a team where nobody is allowed to say the answer is wrong, because that is precisely the condition under which amplified mediocrity becomes official policy.
What remains of the leadership job is harder and more human: fewer documents to write, more decisions to own, and more of the consequence landing on a named person. For anyone who actually wanted to lead rather than administrate, that is a good market to be in.