The Manager is Dead! Long Live Management!

The fundamental thesis underpinning Refactor Management is the idea that engineering management is going to change profoundly as agentic AI becomes prevalent.

This isn’t just a simple skill change where managers use AI to build presentations and summarize Jira activity — I’m talking about a fundamental shift in how people think about engineering management.

By “people” I mean your CEO and CTO.

By “shift,” I mean the role of engineering manager may be completely eliminated.

… And by “may” I mean “is already happening.”

This is a little distressing for me because, dangit, I really enjoy engineering management. I love being a professional enabler and cheerleader. I love seeing my team ship great features. Helping people grow and thrive in their profession is profoundly rewarding for me, and is worth all the hard knocks and political BS that comes with the role.

Nevertheless, I think the writing is on the wall.

Here’s why.

Fact 1: Software development increasingly requires AI.

You don’t require AI to write code; that’s silly. I mean the day to day professional software developer must use AI to keep up. The adoption varies by company and team, but the trajectory is clear: engineers who effectively leverage AI are dramatically outperforming those who don’t.

This trend will accelerate. Back in May, Anthropic released Claude Code. It is shockingly good, if you know when, where, and how to use it. I’ve seen top performing engineers triple their output — with high quality results — because they’ve figured out how to use the tools.

But here’s the thing: not everyone figures it out. Using AI effectively isn’t like using a search engine. It requires specific skills that separate the engineers who get 3x productivity from those who get frustrated and give up.

Fact 2: People who are good with AI have good management skills.

In order to get good results with AI, you have to be specific, thoughtful, and engaged. You can’t just tell an AI what to do, you have to coach it through discovering, analyzing, and debating different ways to solve problems. You have to understand its limits, you have to question its conclusions, and you still have to follow good development practices and collaborate with your teammates to avoid foot-gunning your codebase.

The skills required to orchestrate work for agentic AI has a strong, well-researched overlap with the day to day management of a software development team. A team of researchers from Harvard found a very strong performance correlation between people who were good at managing humans, and people who were good at managing AIs.

How strong? ρ=0.81.

Go ahead and read it: https://www.nber.org/system/files/working_papers/w33662/w33662.pdf

The implication is clear: engineers who excel with AI are developing management skills whether they realize it or not. They’re learning to delegate, provide context, review work, and iterate on solutions. They’re practicing the core competencies of management every day.

But — and this is crucial — not every engineer will develop these skills. Just like not every engineer becomes a good manager today, not every engineer will become effective at AI collaboration. The difference is that those who do develop these skills won’t need a separate manager for many traditional management tasks.

Fact 3: Managers are expensive overhead.

When I interview with companies, I tell them straight out: my job is to make myself redundant.

Teaching people how to independently collaborate, resolve conflicts, and make decisions is the goal of any good manager. When everyone’s taking care of their business, the job of a manager gets pretty boring — sure, there’s a spicy crisis every now and then, but what else? You’re costing your company over a thousand bucks a day to sit in a chair telling an AI to build a presentation containing a summary that another AI generated.

Claude MAX costs $200 per month, and I guarantee that’s a higher return on investment in the hands of a talented engineer who’s learned to manage AI effectively.

I’m watching this play out in real time. Companies are already eliminating EM roles, betting that senior engineers with strong AI skills can handle the coordination and technical leadership. They’re not wrong — at least for certain team sizes and contexts.

Add it all up, and the conclusion is compelling …

The Manager is Dead.

Given the necessity of AI adoption in the workplace, the development of management skills required to use AI effectively, and the dwindling responsibilities of a dedicated management position — I think the vector is clear: engineering management as a distinct role is dying.

This doesn’t mean it happens overnight or uniformly. Some organizations will always need dedicated people managers for large teams, complex cross-functional work, or high-stakes human situations. But the era of one manager per 5-8 engineers? That math is changing rapidly.

Long Live Management!

The role of front line engineering management might be dying, but the skills required to be a good manager are becoming more critical than ever. Coaching, conflict resolution, decision making, and strong communication skills are now table stakes for any senior engineer who wants to remain relevant.

Here’s what the Harvard research tells us: management skills and AI orchestration skills are nearly the same thing. The engineers who thrive in this new world will be those who can switch fluidly between coaching an AI through a complex refactoring and coaching a teammate through a difficult technical decision.

People still need connection, advocacy, and human leadership. The emotional intelligence, career development, and psychological safety that great managers provide isn’t something AI can replace. People still need management.

They just might not need a manager.

The future likely holds fewer people with “manager” in their title, but more people practicing management every day. The most successful engineers will be player-coaches — building systems while building people, orchestrating AI while orchestrating teams.

For those of us who love management, this isn’t really an ending. It’s a refactoring.