AI just removed fifteen hours from a middle manager's calendar. Her organisation called it a win. She called it something else.

This issue is about the quiet career crisis hiding inside every successful AI rollout — what the research actually says about middle management and AI (and what it carefully avoids saying), what the managers who are thriving are doing differently, and a tool you can use before your next development conversation.

Five reports. My honest take on each. One question that will stay with you.

Read time: 4 minutes.

She was one of the best people on the program.

Twelve years at the company. She knew which team lead would actually deliver and which one would promise and then go quiet. She knew when the numbers in the weekly status update didn't match what she'd heard in the corridor. She knew who to call at 4pm on a Friday when something was about to break before Monday morning.

Three months after the AI-powered project management tool went live, her calendar had fifteen fewer hours in it.

Which sounds like a gift.

It wasn't.

The status updates now ran automatically. Risk flags were generated by the tool. Dependency tracking lived in a dashboard that refreshed in real time. Her manager — a VP who had been pushing for the new tooling for two quarters — looked genuinely pleased at the next steering committee. "Look how much more visibility we have," he said.

I caught up with her after the meeting.

"I don't know what I'm for anymore," she told me. "I mean that literally. I know what I do. I just don't know if any of it matters now."

That is not a technology problem. That is a career problem that arrived disguised as a software rollout. And it is happening right now, in programs across every enterprise that has deployed AI in the last eighteen months.

Let me tell you what the research says — and then I'll tell you what I actually think.

What the research says:

The WEF Future of Jobs 2025 report identified coordination and monitoring roles — the core of middle management — as among the most exposed to AI displacement in the next five years. McKinsey's State of AI 2025 found that AI is automating tasks fastest in the layer between strategy and execution: synthesizing information upward, tracking delivery, chasing dependencies. HBR published a piece arguing that companies now need "agent managers" — humans who supervise AI agents rather than human teams. BCG's most recent work on AI at scale found organisations actively reducing management layers as AI takes over the connective tissue. MIT Sloan and Deloitte's joint research on the future of the manager found that the managers who thrive in AI-augmented environments are the ones who shifted from information relay to judgment and interpretation — fast.

What I actually think:

The "augmentation not replacement" framing in these reports is more optimistic than the reality will be for people who don't move. Not because the research is wrong, but because enterprises don't tell people clearly enough what to do next. The AI tool goes live. The coordination work disappears. And the person whose job was that coordination is left waiting for someone to tell them what their new role is.

Nobody is coming to tell them. That conversation is not in the implementation plan.

The middle managers I have watched handle this well did one thing differently: they stopped being the people who moved information and became the people who interpreted it. They asked harder questions in the room. They built judgment that the tool couldn't replicate. They got closer to the decisions that needed a human in the loop — and made sure everyone knew that was where they lived.

That is a real opportunity. But it doesn't arrive on its own. You have to go get it.

Here are the five reports worth your time. I'll tell you what they get right, what they're cautious about, and what they won't say plainly.

Understand what's happening:

1/ WEF, Future of Jobs Report 2025 🔗 https://www.weforum.org/publications/the-future-of-jobs-report-2025/

What it says: Middle management coordination roles are in the top tier of AI-exposed jobs. The report forecasts significant task displacement by 2030 across monitoring, reporting, and information synthesis functions.

My take: The WEF is always careful not to say the quiet part loud. "Exposed to change" is doing a lot of work in this report. Read it as: these roles are shrinking unless the people in them change faster than the technology does.

2/ McKinsey, The State of AI 2025 🔗 https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

What it says: AI adoption is accelerating fastest in knowledge work — specifically coordination, synthesis, and reporting tasks. Companies are seeing real productivity gains in these areas.

My take: McKinsey presents this as good news for organisations. It is. The individual story is more complicated. When your organisation's productivity goes up because AI is doing what you used to do, you need a new answer to "what do I contribute?" before someone else asks you that question.

Make sense of the debate:

3/ HBR, Companies Need Agent Managers 🔗 https://hbr.org/2025/01/companies-need-agent-managers

What it says: The next generation of management isn't managing people who do tasks — it's managing AI agents that do tasks. That requires a different skill set: understanding what the agent is optimising for, where its blind spots are, when to override it, and how to explain its decisions to a room full of people who don't trust it yet.

My take: This is the most useful reframe I've read. "Agent manager" is a real role emerging inside enterprise AI programs right now, even if it doesn't have that title yet. Someone has to own the AI's output and stand behind it. That person is usually the most experienced operator in the room. Which is exactly where middle managers should be positioning themselves.

4/ BCG, AI at Work: From Potential to Profit 🔗 https://www.bcg.com/publications/2024/from-potential-to-profit-with-genai

What it says: Companies that extract real value from AI are restructuring roles faster — including flattening management layers and redistributing decision rights. The organisations that aren't restructuring are just running AI on top of old org charts. That doesn't work for long.

My take: BCG is polite about this. What they're describing is that slow-moving enterprises are accumulating a structural debt. The middle management layer that was designed for a world where information was slow to travel is now sitting on top of infrastructure that moves instantly. Something has to give.

Act on it:

5/ MIT Sloan / Deloitte, Winning with AI: Leaders Speak Out 🔗 https://sloanreview.mit.edu/projects/winning-with-ai/

What it says: The managers doing best in AI-augmented organisations share one trait: they focus relentlessly on judgment-intensive work — the decisions, relationships, and interpretations that AI cannot make reliably. They've stopped competing with the tool on its own terms.

My take: This is the only report that answers the right question: not "what will AI do to managers" but "what are the managers who are thriving actually doing?" Read this one before the others.

Run this before your next development conversation — or if you manage a team, run it with each person individually.

The Middle Manager AI Readiness Check:

"If an AI tool took over [status reporting / dependency tracking / upward synthesis — pick the 2-3 things you spend the most time on], what would you do with those hours?

Name one judgment call in your current role that requires knowing the people, reading the room, or deciding with incomplete information. Not a task — a judgment call.

Who in the organisation currently knows that you're the one who makes that call?

If the answer to the last question is 'no one outside my immediate team' — that is the gap to close before the next performance cycle."

For program directors running AI programs: do this exercise with your middle managers before the AI tool goes live. Not after. You want them walking toward their new contribution, not discovering mid-rollout that nobody defined it.

The middle manager who knows why she's in the room — the real reason, not the job title — will survive any AI rollout you throw at her.

Can you name that reason for the middle managers on your program? Not their responsibilities. The specific judgment, relationship, or call that would be noticeably worse if they weren't there.

If you have to think about it for more than ten seconds, so do they.