There is a memo going around. You have probably received a version of it. It says the company is now “AI-first.” It says every employee is expected to adopt the new tools, reimagine their workflows, and upskill on their own initiative. It usually ends with something about the future.
What it does not say is what the future holds for the reader.
We now have public proof of how badly this is going. This spring, the AI labs concluded that their technology was not the bottleneck — their customers were. Anthropic stood up a deployment venture backed by Blackstone, Goldman Sachs, and Hellman & Friedman. OpenAI answered within weeks with its own Deployment Company and bought a consulting firm outright to staff it, embedding engineers directly inside client organizations. Amazon has since followed. When the makers of the machine have to send missionaries into your building, it is because something inside your building has stalled. The industry calls it the implementation gap. I want to look underneath it, because the stall is only a symptom. The disease is a failure of sight.
Start with the person on the receiving end of the memo, because the human cost is where malpractice always shows first. In my book on leadership malpractice, I use the image of a tangled fishing line. You can sit with a tangle — patient, present, tracing each strand back to where it went wrong. Or you can pull, and every leader knows what pulling does. The knots tighten. Eventually, you reach for the scissors.
The retooling mandate, delivered without a shared picture of the future, is a pull. And it produces only bad endings. The employee who resists is displaced without new skills, pushed out of a shrinking chamber with nothing to carry into the next one. The employee who complies — who learns the tools, automates their own workflow, makes themselves measurably more productive — is often displaced anyway, because the retooling was never a covenant. It was a countdown. When people watch a colleague comply their way into a layoff, the lesson moves through the culture faster than any memo: there is no version of good behavior that leads to safety here. Trust, once spent that way, does not come back at any price.
But I want to name the deeper loss, the one that never appears in the severance math. Every one of those displaced people represented a capacity the enterprise chose not to imagine a use for.
Consider what has actually happened in the last three years. An almost incomprehensible amount of agency has been released into the world. Work that once required teams now requires afternoons. Analysis, drafting, prototyping, coordination — the algorithmic and heuristic layers of nearly every job have become abundant, cheap, and available to anyone who can describe what they want. In any sane accounting, this is a windfall. It is found in the capacity on a scale that no generation of leaders has ever been handed.
And what have most enterprises done with the windfall? They have planned for subtraction. Which tasks go away? Which roles. Which teams. The entire strategic conversation has been an exercise in contraction — as if the only imaginable response to abundance were to need fewer people. Meanwhile, the expansion questions sit unasked: What could we now attempt that we never could before? Which markets were too small, which problems too hard, which services too costly to pencil out under the old economics? Who among our own people has been carrying a decade of “if only we could” that never survived a budget cycle?
Your people know the answers. The ones closest to the customers, the systems, the daily friction — they are holding the map to the expansion. A firm that co-dreams of expansion with its workforce absorbs its people into growth. A firm that plans contraction from above discards them, and then wonders why the survivors work with one eye on the door.
I can already hear the objection: this is idealism. So let me offer evidence instead.
In 2021, Ingka Group — the franchisee that operates most of the world’s IKEA stores — deployed an AI assistant named Billie across its customer service channels. Within two years, the bot was resolving nearly half of all inbound inquiries. Millions of conversations a year. Millions of euros in savings. The scarcity playbook wrote the next memo itself: 8,500 call-centre roles are redundant.
Ingka never sent that memo. Its leaders did something that should not be rare and is: they sat with the tangle. They studied what the machine could not do. The other half of the calls, it turned out, were not routine at all. Customers were asking for help designing their homes — questions of taste, proportion, and judgment that they wanted another human’s eye. Buried under years of routine traffic was a demand signal no one had been free to hear. So Ingka retrained the 8,500 people who already carried deep product knowledge and hard-won customer empathy — as remote interior design consultants. That service line now generates on the order of €1.3 billion a year, and the company aims to grow remote sales to ten per cent of revenue by 2028.
Read that again. The chatbot did not eliminate 8,500 jobs. It revealed a billion-euro business that 8,500 people were already almost qualified to run. The abundance was sitting in the call logs the whole time. The only scarce ingredient was a leadership team willing to look at displaced capacity and ask what it could become. (Candor requires a footnote: in 2026, other corners of the IKEA system announced cuts of their own. The reskilling case stands on its results; the halo does not extend to the whole empire. Even here, sight is a practice, not a possession.)
And IKEA is not an outlier. It is the visible tip of a pattern. A decade ago, AT&T discovered that roughly 100,000 of its jobs were on a path to obsolescence as the network went to software. It chose a billion dollars of reskilling over a decade of layoffs and ultimately retrained more than 180,000 people — still the most-studied workforce transformation on record. JPMorgan, facing AI across nearly every function, is redeploying through natural attrition rather than redundancy rounds: roles are redesigned as people move, retire, and grow, and the annual churn of a 300,000-person institution becomes the restructuring program. No scissors required.
The ledger runs both ways, and this is where the responsibility to shareholders comes into play. The contraction playbook is not merely unkind; it is proving to be bad business. Ford has rehired and promoted more than 350 veteran engineers after automated quality systems failed to capture what those engineers knew. Among business leaders who cut jobs for AI, a majority now admit the redundancy decisions were wrong. Roughly a third of U.S. hiring managers eliminated a role for AI and later rehired for it — and a third of those spent more on restaffing than the layoffs ever saved. Deloitte’s latest global human-capital study puts a number on the pattern: companies that take a technology-first approach to AI are 1.6 times more likely to fall short of their expected returns than those that design for humans and machines together. Scarcity is not only a wound. It is a losing strategy, and the market is beginning to price it.
So why do so few leaders ask the expansion questions? Here is the uncomfortable answer: because seeing abundance is a capacity, and not everyone has it.
I have written elsewhere about scarcity’s children — leaders whose earliest programming taught them that resources are finite, that safety is an illusion, that anything gained by another is something lost by you. That programming does not disappear when you get the corner office. It sits in the nervous system, scanning for threats, and when a technology arrives that could mean almost anything, it reliably means one thing to a scarcity mind: cut costs before someone else does. This is not stupidity. It is a wound, running the show. But understanding the wound does not lower the stakes. A leader who can only see AI as a headcount instrument will run a headcount playbook, and thousands of livelihoods and an entire enterprise’s future will be shaped by the limits of one person’s sight.
Which brings me to the part of this I have been circling for months, and I will just say it.
Leadership is the work of making spaces safe for others — safe enough to tell the truth, safe enough to imagine, safe enough to retool without wondering whether the retooling is a countdown. It is a calling, the way medicine is a calling, and like any calling it demands capacities that not everyone possesses. There is no shame in that. There is only shame in occupying a role without the capacity, while others pay the price.
So here is the standard I think this moment sets. If, three years into the most significant release of productive agency in economic history, a leader still cannot find common ground with their own workforce — cannot articulate a shared future, cannot convene the co-dreaming, cannot see a single expansion worth building — then that leader has two honorable moves left. Hire some help. Or step aside for someone who can.
Hiring help is not a weakness; it may be the most self-aware act of stewardship available. The deployment companies will wire in the technology, but the harder help is human: people who can facilitate the conversation between a frightened leadership team and a distrustful workforce, and hold the space until common ground appears. Seeking that out is what a whole leader does when they meet the edge of their own capacity.
Stepping aside is harder, and it is the move our system almost never asks of the powerful. Notice the asymmetry we have normalized: when a leader cannot imagine growth, it is the employees who move on. The people with the least power in the equation absorb all of the displacement, while the person whose failure of vision caused it remains. If we are honest, in many of these enterprises, it is the leader who should be transitioning out — with the same dignity, and the same help finding their next chapter, that I believe every displaced worker deserves. I stepped away from the company I co-founded when I recognized that better stewards had emerged for the season ahead. It was among the hardest things I have done. It was also precisely the job.
And a word to boards, because this decision is ultimately yours. You are being briefed on AI strategy in every meeting now — platforms, pilots, risk registers. Most of it is noise compared to the one question that will determine everything downstream: can the person you have placed at the top see abundance? Can they stand before the workforce and describe a future worth retooling for? Somewhere in your enterprise, right now, is the equivalent of Ingka’s call logs — a demand signal buried under routine, and a workforce almost qualified to answer it. If the honest answer is that your leader cannot see it, then no vendor selection will save you, and every quarter you wait, the trust drains and the expansion goes unbuilt. The kindest thing you can do — for the employees, for the enterprise, and truthfully for the leader — is to act on that answer before the scissors come out.
Incredible amounts of agency have been released into the world. That sentence should end with an exclamation point in every boardroom in America, and instead it lands like a threat. Great leaders — the ones this moment are quietly sorting from the rest — will hear it as the invitation it is. They will gather their people, open the books, and ask the only question that matters now: What shall we build with all of this?
If you are a leader, ask it this week, out loud, with your people in the room. If you are a board member, ask whether your leader is capable of asking it. And if you are an employee living under one of those memos, I would genuinely like to hear from you — because your answer to that question is the map your leadership has been too frightened to read.
