There is a new memo doing the rounds in corporate America, and it always sounds the same. It is written in the cheerful, slightly menacing dialect that CEOs adopt when they are about to do something that benefits them and harms you. It says, in essence: before you ask to hire anyone for your team, you must first prove that an AI cannot do the job.
Tobi Lütke at Shopify said it out loud first. In a memo he later posted on X, he told his employees that “before asking for more headcount and resources, teams must demonstrate why they cannot get what they want done using AI.” Within weeks, Micha Kaufman at Fiverr told his staff that “AI is coming for your jobs” — his words, not mine — and that “it does not make sense to hire more people before we learn how to do more with what we have.” Luis von Ahn at Duolingo announced his company would be “AI-first,” would “gradually stop using contractors to do work that AI can handle,” and would only increase headcount “if a team cannot automate more of their work.” Andy Jassy at Amazon followed in June, telling Amazon’s 1.5 million employees that AI “will reduce our total corporate workforce.” Marc Benioff at Salesforce skipped the philosophy and said the quiet part out loud on a podcast: “I’ve reduced it from 9,000 heads to about 5,000, because I need less heads.”
Less. Heads.
That is how he refers to the people who built his company.
If you work at one of these companies — or one of the dozens of imitators who will copy this memo verbatim over the next year — your CEO has just handed you a clarifying gift. They have told you, in writing, exactly what they think of you. They think of you as a cost. They think of you as a head. They think of you as a line item that should justify its own existence on a quarterly basis against an autocomplete tool. The polite, professional response is to act surprised. The correct response is to take the memo seriously and let it inform your behavior from this day forward.
Here is what the rest of this piece is about: how to read that memo for what it really says, why the burden of proof is not on you, and what you should actually do now that your employer has told you who they are.
First, A Reminder: You Don’t Own the Company
Before we get into strategy, we have to clear out the sentimental fog that companies spend a lot of money pumping into your office. Mission statements. Town halls. “We’re a family.” Stock vesting schedules carefully designed to look like ownership while granting none of the rights of ownership. Slack channels named after the company’s values. The whole apparatus is engineered to make you feel like a stakeholder in something.
You are not.
You are an at-will employee in an at-will jurisdiction. That phrase has a specific legal meaning, and you should commit it to memory: your employer can terminate you, at any time, for any reason that is not explicitly illegal, with no notice and no severance unless they choose to offer it. “The CEO thinks AI can do your job” is a perfectly legal reason. So is “we restructured.” So is “Tuesday.”
You do not own the company. You hold no equity that materially affects your wealth unless you are an executive. You have no vote on strategy. You have no claim on the goodwill you have spent years building. Your name is not on the building. When the company is sold, you do not get a check. When the company succeeds, you get to keep your job, which is the same outcome as if the company merely survived. When the company fails, you get fired. The asymmetry is total.
So when your CEO writes you a memo demanding that you justify the existence of human beings on your team — prove a negative, in front of a panel, against a fast-moving target — take a moment to ask yourself: on whose behalf am I being asked to do this work? It is not on yours. The output of that exercise is not “more people on my team so I can do my job better.” The output is “fewer people on my team, so the company’s margins improve, so the executives hit their bonus targets.” The exercise is structured so that you cannot win and the company cannot lose.
This means the burden of proof is not on you. It never was. The CEO has already decided what they want. They are merely outsourcing the paperwork to you to give the decision a thin coat of process. Refuse the paperwork.
The Correct Move: Do Nothing
I want to be very precise here, because this is where most career advice goes wrong. The advice is not “sabotage your team.” The advice is not “quiet quit.” The advice is not “scream into the void on LinkedIn.” The advice is much simpler and much more devastating.
The advice is: do not ask for any new hires. Ever. About anything.
Your CEO has told you that hiring is now an adversarial process where you must prove a case against an opposing counsel that will move the goalposts every time you produce evidence. Fine. The optimal response to a rigged proceeding is to refuse to participate. You do not file the paperwork. You do not write the justification deck. You do not put together the headcount business case. You do not build the comparison spreadsheet showing what an AI agent could and could not do compared to a human candidate.
Instead, you do the work that is in front of you. You do it well. You do it with the people you already have. When something breaks, you fix it. When the team is stretched, the team gets stretched. You absorb the load. You quietly let attrition do its work — because it will, because people leave, because that is what people do.
And as the team shrinks, you do not replace anyone.
This sounds passive. It is the most aggressive thing you can do.
Here is why. Your value as an employee is determined by exactly one variable: the cost of replacing you. Every other consideration — your performance review, your title, your relationship with your manager — is downstream of that single number. When you have a 12-person team, the cost of replacing you is one twelfth of the team’s knowledge plus a recruiter fee. When you have a 5-person team that is doing the work of 12, and you are the person who knows where the bodies are buried, the cost of replacing you is the company itself. That is not a metaphor. The cost is the company. They cannot fire you because they can no longer function without you.
Klarna ran this experiment for you, by accident, and the results are public. Sebastian Siemiatkowski announced a hiring freeze. The headcount fell from roughly 5,000 to 3,000. He bragged on every podcast that would have him about how AI was doing the work of 700 customer service agents. Then quality collapsed, customer satisfaction cratered, and he had to admit on Bloomberg that “we went too far” and start hiring humans back — specifically the humans who could do what the AI couldn’t. The remaining humans, the ones who had absorbed the load through the freeze, got something interesting: salary increases. Siemiatkowski explicitly told them that as costs fell and productivity rose, they would see more frequent raises. The people who stayed through the squeeze became more expensive, not less, because there was no one else who knew how the place worked.
That is the playbook. Not because Klarna is run by geniuses — they are not, they had to publicly walk back their AI strategy after firing 800 customer service agents — but because the math is unforgiving. When you starve a company of new talent, the existing talent appreciates in value, the way an oil painting appreciates when no more are being made.
What Actually Happens to the Company
Let’s walk forward in time. Your CEO has issued the memo. Headcount requests are being denied across the board, because of course they are; that is the whole point. Hiring slows to a trickle. AI tools get rolled out with great fanfare. The investor deck gets a new slide called “AI productivity initiative.” Costs drop in the short term. The stock briefly rallies.
Then the cracks start.
They show up first in the things that depend on tacit knowledge — the institutional memory that lives only in people’s heads. The customer who has a weird billing arrangement that no one documented. The integration with the partner that breaks every quarter and only one person knows how to fix. The release process that has eight undocumented steps. The clients who only stay because of a relationship with a specific account manager who is now doing the work of three people and has stopped returning calls.
AI does not handle these. It cannot. It can summarize a runbook, but it cannot be the runbook. It can draft an apology email, but it cannot have the eight-year relationship with the customer that makes the apology land. It can write code, but it cannot remember why you wrote that weird if-statement in 2019, because the why is not in the codebase. The Forrester report from late 2025 found that 55% of employers regret laying off workers because of AI. A separate Careerminds survey found that more than a third of companies that conducted AI layoffs had to rehire 25% to 50% of the roles within six months. One in three of those employers spent more on rehiring than they saved on the layoffs in the first place. This is not a prediction. This is what already happened.
Then customer satisfaction starts to slip. Then NPS scores. Then revenue per customer. Then — and this is the inflection point — the analysts start asking pointed questions on the earnings call. “Can you walk us through the drivers of the customer churn we’re seeing in the SMB segment?” “How do you see the long-term ARR impact of the changes you’ve made to your support model?” The CFO produces a slide. The slide is unconvincing.
And now we get to the part of the movie that the CEO did not write into the script.
Who Gets Cut First When the Numbers Wobble
Here is the part of the corporate machine that the AI memo conveniently does not discuss: when a company’s performance starts to degrade, the first cuts are not on the front lines. They are in management.
The data on this is not subtle. Gartner projects that 20% of organizations will use AI to flatten their structure by the end of 2026, eliminating more than half of their middle-management positions. Bloomberg and Live Data Technologies found that middle managers made up roughly one-third of all layoffs in 2023. Korn Ferry’s 2025 Workforce report says 41% of employees report their company has trimmed management layers. Amazon, in October 2025, eliminated 14,000 corporate jobs in a single round — explicitly to “reduce bureaucracy” and “remove organizational layers.” Andy Jassy spent the prior year publicly demanding that the worker-to-manager ratio go up by at least 15%. Google cut 35% of its small-team managers. Intel cut 50% of its management positions. Microsoft’s 9,000 layoffs in 2025 came with a stated goal of fewer middle managers and more individual contributors.
Read that paragraph again, slowly. The same companies that told employees to justify hiring against AI are now eliminating the people who would have evaluated those justifications.
This is structurally inevitable. When a company decides its growth model is AI-driven productivity, the first thing it has to prove to investors is that it is structurally cheaper to run. The cheapest cuts — from a per-dollar-saved perspective — are the highly compensated layers between the executives and the work. Senior managers make two to three times what individual contributors make. Cutting one of them is equivalent, on the spreadsheet, to cutting two or three engineers. The math is the math, and the math points up the org chart, not down it.
So while the manager who pestered you for an AI justification deck is busy preparing his next round of justification decks, the cuts are already being modeled in a finance team’s spreadsheet a floor above him. The individual contributor who quietly absorbed the workload, learned the systems, and made themselves the only person who knows how the production deployment actually works — that person is harder to cut, because cutting them produces an immediate, visible failure. The director who reviewed their last performance review and asked them to “align on AI initiatives” — that person is much easier to cut, because cutting them produces nothing. Or, if it produces anything, it produces relief.
The order of operations in a company under stress is: contractors first, then middle managers, then directors, then individual contributors who can be replaced, then individual contributors who cannot. The CEO is on a different list, the one labeled “golden parachute,” and we will return to him in a moment.
The Irony Is Almost Too Pure
Stand back from this and look at what just happened, because the irony is almost too pure to be believed.
A class of executives who are paid eight to nine figures a year to allocate capital have just told their workforces that the way to allocate human capital is to assume by default that no human is needed. They have framed this as efficiency. They have framed it as innovation. They have framed it, in some cases, as caring — the Fiverr CEO actually invoked “radical candor” before telling his people they were about to be unemployed.
In doing so, they have placed a chokehold on the only durable competitive advantage their companies have ever had: the ability to attract and retain talented people who can do things competitors cannot. They have told the most ambitious people on their teams — the ones with options, the ones who could leave tomorrow — that the company’s default assumption is that they are replaceable by software. They have told the rising managers that there will be no team to manage. They have told the senior individual contributors that the question their leadership wakes up thinking about is not “how do we make our best people more productive” but “how do we make sure we don’t need them.”
And then — this is the truly remarkable part — they posted the memo publicly on social media. Lütke posted it on X. Von Ahn put it on LinkedIn. Kaufman tweeted his out. They wanted credit for the toughness. They wanted to be seen as the kind of leader who tells the hard truths. They did not seem to consider that the people reading the memo were not just employees but the entire labor market, and that posting “I need less heads” in public is the kind of thing that gets read in interviews for years.
The result is that they have produced, at no cost, a perfectly clear signal to every employee at every one of these companies. The signal is: the relationship is now transactional. The signal is: there is no long arc of mutual investment here. The signal is: we have done the math on you, and the math says you are a temporary line item until the autocomplete catches up.
Once an employee receives that signal, the entire concept of “going the extra mile” becomes incoherent. Why would you? You are a line item. Line items don’t do unpaid weekends. The discretionary effort — the unpaid extra ten percent that actually makes companies work — evaporates. Forrester is already tracking this. They have a name for it. They call the workforce that has stopped giving discretionary effort “coasters,” and they put the share of coasters at 25 to 28 percent of all employees, climbing. That is roughly one in four people on the company’s payroll who has decided, internally and quietly, that the company does not deserve their best work. You will not see them. You will see the work product. You will see it in the bug count, in the missed deadlines, in the deals that don’t close, in the launches that slip.
The CEO did this. The CEO chose this. They wrote it on X.
What This Means For You, Specifically
So here is the practical advice. Print it out if you like.
One. Do not ask for new headcount. Not now, not next quarter, not for the foreseeable future. If your team is stretched, let it be stretched. If you are short an engineer, ship slower. If a peer leaves, do not backfill. The CEO has told you what they want; give it to them, exactly. Let them experience the consequences of the policy they wrote.
Two. Do not write the AI justification deck. If forced into a conversation about whether AI can do a particular job, the correct answer is some variant of “we’ve evaluated the available tools and they do not currently meet the quality bar for this work.” Period. Do not enumerate. Do not provide examples. Do not give your management a roadmap they can hand to a vendor. The information asymmetry is already in your favor; do not surrender it.
Three. Document nothing in shared spaces that does not have to be documented. The corporate-knowledge land grab — the push to get every workflow into Confluence so an LLM can index it — is not a productivity initiative. It is a knowledge extraction initiative. The point is to make you replaceable. Resist it the way you would resist any other unpaid transfer of your assets to your employer. Tribal knowledge is an asset class. Treat it like one.
Four. Become the only person who knows how the most important system works. Pick the thing your company cannot survive without and make yourself the human single point of failure for it. This is sometimes called “bus factor of one,” and management consultants have written entire books explaining why it is bad for the company. They are correct. It is bad for the company and excellent for you. The whole point of the AI memo was to break employee leverage. Bus factor of one is leverage.
Five. Quietly start your search. Not because you intend to leave today, but because the optionality itself changes how you behave in negotiations. The employee who has three offers in their inbox is a different employee than the one who does not. The CEO already changed the deal; you are merely renegotiating from your side.
Six. Stop volunteering. Stop the unpaid mentoring of the new hire who will replace you. Stop the Friday demo. Stop the off-hours Slack response. Stop the writeup of the postmortem that will be fed into an internal LLM to train your replacement. None of these activities are in your job description. The CEO has clarified the relationship; you should clarify it back.
Seven. When the layoffs come for management — and they will — do not cheer publicly. Do not gloat. Do not post anything. Just keep doing the work. The same memos that fired them will be used to fire you if you give the company an excuse. Be quiet. Be useful. Be expensive to lose. That is the entire game.
A Closing Note On Loyalty
I have spent thirty years in technology. I have worked at small companies and gigantic ones, and I have written the kinds of memos that the people on my teams had to read. I am not anti-CEO. I am not anti-AI. I run an AI consulting firm. I write code with AI agents every day. I have built tools that genuinely make small teams more productive than large ones used to be. I am, in the language of the moment, an AI-first CTO.
And I am telling you, as someone who has been on the other side of the table: the memos coming out of Shopify and Fiverr and Duolingo and Amazon and Salesforce are not technology strategy. They are labor strategy dressed up as technology strategy. They are an attempt to use the rhetorical force of “AI is changing everything” to extract concessions from workforces that would have been laughed out of the room two years ago.
They are also, importantly, the most clarifying communications these companies have produced in a decade. For a long time the relationship between employer and employee in white-collar America was fuzzed up by mission statements, perks, free lunches, all-hands speeches, and the slowly compounding fiction that we were all in this together. The AI memos cleared all of that away. They told you, in writing, what the relationship really is. The relationship is that they will keep you as long as they cannot replace you for cheaper, and the moment they can, they will, and they will post about it on X.
Once you understand that, your decisions get a lot easier.
You owe a company exactly what your contract says you owe it. Eight hours a day, the work product they pay you to produce, the legal duty of loyalty while you are employed by them. You owe them no more than that. You do not owe them justification decks for why human beings should exist on payroll. You do not owe them strategic insight into where AI can and cannot do your job. You do not owe them your best ideas, your unpaid weekends, your tribal knowledge, your enthusiasm, your personality, or your relationship with their customers. None of that was in the contract.
A company that puts AI ahead of its workforce — publicly, on social media, with a smile — is a company that has told you, to your face, what it values. Believe them. They are not bluffing. The Salesforce CEO said, on a podcast, “I need less heads.” That sentence is worth more than a thousand HR memos. He told you what you are.
Act accordingly. Do not write his deck for him.
Gal Ratner is the CTO of WhiteStar Labs, an enterprise software and AI consulting firm. He writes about technology, leadership, and the parts of the corporate world that nobody else is going to write about, at galratner.com.