Fire 50% of your team

You should fire 50% of your team and replace them with AI.

I’ve been turning that sentence over in my head for a few months now. Testing it. Arguing with it. Running the actual numbers at Reboxed and sitting with what the spreadsheet said versus what my gut said versus what feels right about the people in the business. And I want to spend the rest of this piece figuring out whether I actually mean it - because the honest answer is: I’m not sure. And I think that uncertainty is more useful to you than a clean take.

So let’s start with the bit most people are quietly walking past.

AI didn’t create the fat. It turned the lights on.

Every growing business carries dead weight. Not dead weight as in bad people - I want to be clear about that. Dead weight as in roles that exist because they used to be necessary, and nobody has had the courage or the occasion to question them since.

Think about what’s actually changed in the last two years. The task that took three days now takes forty minutes. The report that needed a contractor needs an afternoon with the right prompt. The content calendar that justified a dedicated hire now needs a decent brief and a good eye for editing. The analysis that required a specialist can be scaffolded in an hour.

This was always coming. The automation was always on its way. AI just dramatically accelerated the timeline and made the reckoning something you can no longer defer to next quarter.

So when I say “fire 50% of your team,” what I’m actually saying is: AI has exposed how much of what we pay people to do was provisional. Placeholder work. Tasks we outsourced to humans because we hadn’t built the tools to do it any other way yet, and the tools just arrived.

That is not comfortable to say. It is not comfortable to sit with if you have people on your payroll you genuinely like and respect. I have those people. This isn’t abstract.

But it is true. And pretending it isn’t true doesn’t make it less true - it just makes you slower than the next person who decided to look at it honestly.

The case for actually doing it

Let me be clear about who this argument is for. It’s not a universal prescription. It’s highly specific to a particular type of founder.

If you’re bootstrapped, tight on runway, doing real revenue but not comfortable revenue - AI is not a productivity tool for you. It is a structural rethink of what your business can produce per person. That’s a different category of thing entirely.

A two-person team with serious AI leverage can now do what used to require eight. Not approximately. Not with corners cut and quality suffering. Actually. The content gets made. The analysis gets done. The customer responses go out. The code gets written. The first draft of everything exists before you’ve finished your coffee.

At Reboxed, I’ve collapsed work that would have required external resource into my own working week. Not by working harder or sleeping less - by being genuinely more leveraged. I’ve shut down subscription costs and automation workflows that used to require hands-on management. I’ve built things solo that two years ago would have needed a developer, a designer, and a project manager working in sequence.

For small founders, the alternative to restructuring around AI isn’t “keep your people and thrive.” Often it’s “keep your people and fold in eighteen months, at which point nobody has a job.” When you frame it that way, the moral weight shifts. Quite a lot, actually.

I’ve run the numbers at Reboxed. I’ve sat with what those numbers imply. The tension between the spreadsheet and the person is real, and I’m not going to pretend it resolves cleanly. It doesn’t.

The case against

Here’s what doesn’t show up in the model.

The person who notices something is off - not in a deliverable, not in a metric, but in you. Who catches you about to repeat a mistake you made two years ago because they were there when you made it. Who asks the question in the team meeting that you were too close to ask yourself. Who tells you the truth when everyone else in the room is nodding, including the AI.

An AI will never have a bad feeling about a direction you’re heading. It will never push back because something doesn’t sit right in its gut. It has no skin in the game, no genuine stake in whether this all works out, no accumulated context built from years of watching you make decisions under pressure.

There is a version of the AI-optimised business that produces more, faster, cheaper - and is somehow also less able to surprise itself. Less honest internally. A room where there’s one voice and one very agreeable mirror reflecting it back.

That failure mode is real. It just takes longer to show up on a spreadsheet than “reduced headcount cost” does.

The best operators I’ve worked with weren’t the most productive people in the room. They were the most honest. The ones who could see what was actually happening rather than what the plan said was happening. That quality is human, it’s built over time, and it cannot be replicated by a model that’s trained to be helpful.

Where the moral line actually sits

This is the part I keep coming back to, and I think it’s the real crux of the whole thing.

The ethics of this question are not the same for a five-person startup as they are for a 40,000-person corporation. And we need to stop collapsing those two things into the same conversation.

When a small founder restructures around AI, it is a survival decision. It is personal, bounded, made in the context of a business that might genuinely not exist otherwise. The founder has skin in the game. The decision is theirs to live with, on every level. I don’t think we should moralise heavily at founders making hard bets with their own companies.

When a company with healthy margins and a well-compensated board systematically eliminates thousands of roles because the quarterly number needs to move and AI makes the case neatly - that is a different act. Same language. Different motivation. Not survival. Extraction.

We should have different moral vocabularies for those two things, and we don’t. They get folded into the same “future of work” conversation when one is a founder trying not to drown and the other is a boardroom deciding that human beings are a cost line that the new tooling makes avoidable.

That conflation is doing a lot of heavy lifting for a lot of large corporations right now. And I think we should be more explicit about calling it out.

What I actually believe

I believe AI has genuinely changed what’s possible for small teams - not incrementally, not as a nice upgrade, but as a structural shift in what two or three people can actually produce. Pretending otherwise is expensive naivety.

I believe most businesses have accumulated work that was always provisional. AI making that visible is uncomfortable but necessary. Better now than later.

I believe the answer is not “keep everyone regardless” and not “replace everyone you can.” It’s harder than either of those, and it’s specific to your situation, your people, and what you’re actually trying to build.

I believe that large corporations using AI to hit margin targets while executive compensation stays pristine should be named clearly for what it is. Not disruption. Not innovation. Power. And it’s worth being precise about the distinction.

And I believe - genuinely - that there’s a version of this moment that shifts the playing field toward smaller, leaner, more leveraged teams. Where “we can afford more headcount” stops being a moat. Where a three-person operation with serious AI leverage actually competes with businesses ten times their size.

That’s the version worth building toward. That’s the reason I find this era exciting rather than just threatening.

So: should you fire 50% of your team and replace them with AI?

If you’re a bootstrapped founder who’s been honest about what those roles are actually producing, and the honest answer is uncomfortable - probably yes. Maybe yes. At least look at it.

If you’re a large company with comfortable margins looking for a palatable story to tell shareholders about a restructuring you were going to do anyway - I think you know. And I think somewhere in the building, so does everyone else.

The take was always a provocation. The discomfort underneath it is real.

What do you think? I don’t have this resolved. Push back, agree, tell me I’ve got it completely wrong. That’s the whole point.