Maybe the Middle Manager Isn't Disappearing. Maybe the Job Is Finally Becoming Human.

AI will flatten organizations.

Managers—especially middle managers—will disappear.

With intelligent agents tracking projects, summarizing meetings, analyzing performance, coordinating schedules, preparing reports and moving information throughout an organization, companies simply won’t need as many people managing other people.

There is undoubtedly some truth in that.

But I wonder if we’re asking the wrong question.

Instead of asking how many managers AI will eliminate, perhaps we should ask:

A remarkable amount of what we call management isn’t really managing people.

It’s managing information.

And AI is extraordinarily good at that.

Think about the typical day of a middle manager.

There are calendars to coordinate.

Meetings to schedule.

Meetings to attend.

Meeting notes to summarize.

Status reports to prepare.

Dashboards to update.

Projects to track.

Performance metrics to compile.

Emails to answer.

Information from senior leadership to translate downward.

Information from teams to summarize upward.

And then there are the meetings about the meetings.

That includes helping a talented employee who has lost confidence.

Coaching someone through an assignment they have never handled before.

Resolving a conflict between two valuable team members.

Recognizing that a high performer is burning out before the high performer recognizes it.

Helping a new employee understand how an organization really works.

Giving someone difficult feedback without destroying their confidence.

Recognizing potential.

Encouraging experimentation.

Creating psychological safety.

Helping people find meaning in their work.

These are fundamentally human activities.

Yet in many organizations, they have been squeezed into whatever time remains after the administrative machinery of management has been fed.

Agentic AI changes this equation dramatically.

Imagine the manager whose AI agents can continuously track project status without requiring another status meeting.

They can summarize the important points from team discussions.

They can identify approaching deadlines.

They can surface unusual patterns in workflow.

They can prepare the first draft of a performance summary.

They can coordinate schedules.

They can find information buried inside company systems.

They can remind people about commitments.

They can prepare briefing materials before a meeting rather than requiring the manager to spend an hour assembling them.

None of this sounds revolutionary individually.

Collectively, however, it could be.

And that raises a much more interesting question:

There are two very different answers.

Some organizations will see AI efficiencies and immediately reach for the org chart.

If one manager can oversee twice as many people, why not double the span of control?

If AI can handle much of the coordination work, why not eliminate an entire management layer?

And certainly, some management jobs probably will disappear.

Organizations accumulate bureaucracy. Titles proliferate. Layers get created that eventually become difficult to justify. AI may expose some of that.

But there is another possibility.

The manager stops being primarily an information router.

The manager becomes a coach.

A mentor.

A translator of ambiguity.

A developer of talent.

A resolver of conflict.

A builder of trust.

A person who notices things that don’t appear on dashboards.

And perhaps most importantly, someone who helps another human being become better at what they do.

There is a fascinating contradiction emerging around AI.

Consider the young professional entering an AI-supported workplace.

AI can provide enormous amounts of information.

It can explain concepts.

It can draft documents.

It can analyze data.

It can recommend courses of action.

But it cannot assume organizational responsibility for helping that young person develop judgment.

Someone still needs to say: “Technically, that’s correct. But here’s why I wouldn’t do it.”

Someone needs to explain why an angry client’s email doesn’t necessarily require an angry response.

Someone needs to recognize when a young employee is following the AI recommendation too literally.

Someone needs to explain organizational context.

Someone needs to ask the question the machine didn’t ask.

Someone needs to help that employee distinguish between getting an answer and exercising judgment.

That is management.

Or at least, it should be.

This could fundamentally change how we measure management performance.

For decades, we have often rewarded managers for outputs that are easy to measure:

Did the project ship?

Did the team hit its numbers?

Did costs remain within budget?

Did everyone complete the required processes?

Those things still matter.

But AI may make many of those administrative outcomes easier to achieve.

The differentiator may increasingly become something harder to quantify:

Did they become more capable?

Did their judgment improve?

Did they learn new skills?

Did strong performers stay?

Did struggling performers get the coaching they needed?

Did people become more confident making difficult decisions?

Did the team become capable of operating with greater autonomy?

Did people leave your team better prepared for their next opportunity than when they arrived?

Those aren’t soft outcomes.

In an economy in which human judgment, creativity, discernment and adaptability become increasingly valuable, they may be among the most important outcomes a manager produces.

There is an obvious danger here.

AI gives a manager ten hours back each week.

Leadership sees ten available hours.

So leadership fills them.

More direct reports.

More projects.

More meetings.

More responsibility.

More output.

We’ve seen this pattern before.

Technology promises to reduce work, and organizations discover new ways to fill the space technology creates.

But if we do that with management, we may squander one of AI’s greatest opportunities.

That requires leadership teams to make a conscious decision about what they believe managers are actually for.

Imagine a company making this explicit.

A manager’s job description might say:

Your responsibility is not merely to supervise the output of your team.

Your responsibility is to increase the productive capacity of your team.

AI handles much of the tracking.

AI handles much of the summarizing.

AI handles much of the scheduling.

AI handles much of the reporting.

The manager handles the messy human stuff.

The conversation that can’t be reduced to a prompt.

The disagreement that doesn’t have a clean answer.

The employee whose performance problem is actually a confidence problem.

The talented person who needs someone to say, You’re ready.

The person going through something difficult who needs a little grace.

The employee with an unconventional idea who needs someone willing to listen.

The emerging leader who needs someone to challenge them.

The team that needs someone to remind them why their work matters.

That’s not administrative overhead. That’s leadership infrastructure.

I don’t doubt that AI will eliminate some management positions.

It probably should.

There are organizational layers whose primary purpose is moving information from one level of a company to another. AI will increasingly make that unnecessary.

But eliminating bureaucratic management isn’t the same thing as eliminating managers.

Machines are becoming extraordinarily capable at the first.

Humans may finally have more time for the second.

So perhaps the question for CEOs, boards and workplace leaders isn’t:

How many managers can AI help us eliminate?

Maybe it’s:

That’s a much more interesting conversation.

And it may tell us a great deal about whether the AI workplace ultimately produces better work—or simply more of it.

Our best days lie ahead.

Note: image in this post rendered on ChatGPT based on our prompts.

I’m Dan Smolen. As host and executive producer of What’s Your Work Fit? I help you make your work and workplace decisions result in better and more satisfying professional experiences and outcomes. I am also a Founding Member of The Future of Work Alliance.

Join What's Your Work Fit? on LinkedIn

Join What's Your Work Fit? on YouTube