An uncomfortable truth for middle managers about AI

We're not getting caught out

Good morning, happy Tuesday gang,

We’re standing just over a week away from announcing something we’ve been building for a long time. We’ve got a big photoshoot tomorrow, and you’ll hear from me next week 👀 22nd April. 11am BST.

But, back to the very important topic at hand, we need to discuss…

An uncomfortable truth for middle managers about AI

Jack Dorsey, founder of Twitter, published a blog post arguing that AI could take over much of what middle managers do. Track projects. Move information up and down. Keep teams aligned. Of course all the news outlets lapped it up.

And to be fair, he has somewhat of a point. A lot of what happens in the middle layer of organisations is information admin: status updates, progress reports, chasing approvals. If AI can handle that, fine. Literally nobody went into management because they wanted to be a human relay station.

But here’s the part that the article (and Dorsey himself) glosses over. That coordination work is never what management is supposed to be. It’s just what management becomes when organisations haven’t developed their people properly.

And that distinction, between managing and actually leading, is one we’ve always understood. We just never truly had to choose between the two. Until now.

Management vs The Roman Empire

In his blog post, Jack Dorsey uses the Roman army as his metaphor for management in modern organisations. He explains that layers of command existed because humans couldn’t move information fast enough due to long distances. Therefore, her argues that we don’t need as many layers now that AI can move it faster than any human ever could.

But Roman centurions weren’t just passing messages down the line. They were the person their soldiers would follow into something terrifying, they held the line when strategy collapsed at first contact, they read the field in real time and made judgment calls that no general, however well-informed, could make from a distance.

That’s not a manager. That’s a leader.

And that part can’t be automated. Not because AI can’t track a project or flag a risk, it can. But because leadership at it’s core isn’t about processing information, it’s about building people up to be courageous enough to do hard things. It’s the conversation after someone gets passed over, or the team that’s quietly burning out before it shows up in any dashboard.

A manager coordinates. A leader makes people feel like it’s worth showing up.

AI is coming for the first one. It was never going to touch the second.

How most managers operate

Think about how most managers end up in the role. They were brilliant at their job, so they got promoted, but nobody really prepared them for what came next. So they defaulted to what they knew… Doing and tracking, rather than leading.

They weren’t shown how to step into a difficult performance conversation, so we avoided it and filled the diary with status update meetings instead.

Decisions escalate upward because nobody had built the courage or clarity for people to make calls themselves. Meetings multiplied because alignment hadn’t been built through trust, so it had to be manufactured through process and tracking.

The managing expanded to fill the space where the leading should have been. And for a long time. So AI isn’t replacing middle management, it’s just filling the admin space where real leadership should have been all along.

So what does this mean, for us managers?

Middle management isn’t going anywhere. But the version of it that hides behind a full calendar and a project tracker is on borrowed time.

The managers who survive and fly in an AI-augmented organisation will be the ones who push back on prioritising the admin work in the first place.

The ones who were having the honest conversations, developing their people, and building the kind of culture that allows a flatter structure to actually function.

That’s not a management job, its a true leadership role. And it turns out it’s the same role most organisations have undervalued for years, because the admin layer was so visible and so measurable that it crowded out the harder, quieter work.

AI removing the admin layer doesn’t remove that harder leadership work, it just makes it unavoidable. So the leaders left standing will need to be genuinely, not just administratively, good at leading people.

Final thoughts

The leaders vs managers debate has been around for decades.

We’ve written the books, rann the workshops, drawn the Venn diagrams. And somewhere along the way, most organisations nodded along, but just quietly carried on rewarding the wrong things.

AI hasn’t changed what good leadership looks like. It’s just starting to remove the hiding places.

For the managers (myself included) who once upon a time built our identity around being across everything, filling in the reporting spreadsheets, knowing every update, owning every decision… I’m afraid that role is shrinking.

But the leaders who built trust, grew their people, and made themselves less necessary over time? They’re exactly what a leaner, faster, AI-augmented organisation needs more of.

We’ve always known the difference between a manager and a leader.

It’s just that until now, we got away with never having to choose.

Peace,

H

P.S. I'm being deadly serious about this announcement on the 22nd. The middle management revolution is incoming.