Signs of “the Great Flattening” are emerging across corporate America. As companies increasingly deploy AI agents to handle workflow orchestration, task coordination, reporting, and information sharing, they are beginning to rethink one of the most enduring features of modern organizations: layers of middle management. The traditional corporate pyramid—where information flows up and down through multiple tiers of management—is starting to compress as AI systems take over much of the coordination work that once justified those layers.

Around 41% of employees say their companies trimmed management layers last year, according to Korn Ferry’s survey of 15,000 professionals worldwide. And recent restructuring efforts at companies including Meta, Citigroup, CrowdStrike, and GitLab have fueled debate about whether AI could accelerate a broader shift toward flatter organizations.

Cloudflare CEO Matthew Prince recently offered one of the clearest articulations of this view. After cutting roughly 20% of the company’s workforce while posting record revenue, Prince said: “The vast majority of those we laid off last week were measurers,” he wrote. He defined “measurers” as those in middle management, finance, legal, internal auditing, and revenue recognition.

Prince said the company kept what he called the “builders,” such as engineers. That claim stands in contrast to the belief that software engineers are among the most vulnerable to AI, as the technology can code quite well, especially following the release of Anthropic’s Claude Code. He said “sellers” are also relatively safe from automation. Prince added that the layoffs weren’t about reducing headcount, but about shifting the nature of work. The company has a record number of open positions, according to Prince, in “areas that drive growth.”

Bret Greenstein, chief AI officer at consulting firm West Monroe, argues that the role of management itself is changing. In an AI-enabled organization, managers will increasingly be expected to contribute measurable business outcomes rather than simply serve as conduits of information.

“Thanks to AI at everyone’s fingertips, CEOs know things as fast as anyone in the team,” Greenstein said. “You don’t really need a translator.”

In addition, a significant portion of management work has traditionally involved gathering information, communicating updates, scheduling meetings, tracking progress, and keeping teams aligned. AI agents can now perform many of those functions continuously and at scale.

“A mid-level manager can spend something like a third of the week in meetings, most of it keeping people in sync,” said Andy Williamson, CEO of ONLC Training. “That’s exactly the work the software can handle now.”

But Max Martina, president of Cambridge Leadership Associates, emphasized that this doesn’t mean management layers will fully disappear. Instead, AI will supplement management work and support decision-making.

“The activity of management will be supported with new tools, greater efficiency, and an opportunity to move from task-focused execution to leadership behavior,” he said. “It will open doors to the domain of real leadership, not management.”

The Great Flattening is far from mainstream—for now

Yet the Great Flattening remains, for now, far from mainstream. Williamson notes that the most significant organizational changes are still concentrated among technology-forward companies with sophisticated digital infrastructure. Most organizations are still in the early stages of AI adoption, and many lack the systems needed to automate coordination at scale.

Even so, experts increasingly believe the era of the “safe middle” is ending.

“The companies flattening their org charts are not just cutting costs,” said Mark Vena, CEO and principal analyst at SmartTech Research. “They are admitting that a lot of management became workflow babysitting, and AI agents are very good babysitters.”

At the same time, remaining managers are increasingly expected to become effective supervisors of AI systems. The emerging skill is not simply knowing how to write prompts, Williamson said, but understanding how to direct multiple AI agents toward the right work, evaluate their outputs, and integrate those results into business decisions.

“The real skill is pointing a handful of agents at the right work and checking what they send back,” he said.

The human side of change

But none of this change is easy, no matter where you sit on the corporate org chart. “There is a lot of fear of possible job loss and changing responsibilities in general,” Williamson said. “Leading people through that gets more valuable, not less.”

That points to one of the paradoxes of the AI era: The more work becomes automated, the more valuable distinctly human leadership skills become. Organizations still need people who can build trust, navigate uncertainty, resolve conflicts, mentor employees, and help teams adapt to change.

Managers themselves may struggle with the transition. “If you haven’t produced deliverables in a long time, the idea that now you should do it is scary,” said Greenstein. “Managers may also feel their identity is threatened if their value was defined by how many people reported to them rather than how much impact they had on the business.”

Greenstein argues that companies should resist viewing AI purely as a cost-cutting tool. Instead, they should focus first on automating routine work so employees can spend more time on higher-value activities.

“What I tell companies is to use AI to automate the routine, low-risk work off people’s plates to free up time,” he said. “Then build systems that help people and AI work together to solve problems neither could solve alone.”

Martina agreed, saying that AI will ultimately supplement management work and support decision-making. “I don’t see management layers disappearing, rather, I see management layers increasingly efficient and capable of building new capacity in their teams more quickly,” he said.

Leadership, not management

What emerges from a flatter organization is not necessarily less leadership, but a different kind of leadership, Martina argues.

For decades, many managers built careers by coordinating work, approving decisions, and serving as information brokers between executives and frontline employees. As AI takes over more of those responsibilities, the value shifts toward skills that are harder to automate: exercising judgment, navigating ambiguity, building trust, developing talent, and creating alignment around a shared vision.

“The activity of management will be supported with new tools and greater efficiency,” Martina said. “What remains is leadership.”

Yet flattening organizations creates a new problem that few companies have solved. Historically, many professionals—from lawyers and accountants to engineers and analysts—developed expertise through entry-level work before advancing into management and senior leadership roles. If AI automates a significant portion of that early-career work, companies may find themselves with a shrinking pipeline of future experts.

“The biggest risk isn’t necessarily fewer middle managers today,” Martina said. “It’s what happens 10 years from now if fewer people are getting the experience they need to become senior leaders and experts.”

Ultimately, the managers who succeed will be the ones who are hungry to learn, said Greenstein, noting that AI agent tools like Codex and Claude Code have hit tipping points in the past couple of months, moving from software developer use cases to ones suitable for any knowledge worker.

“The next three months are going to be wild as leaders figure out how to adopt it, or they are pressured by people who have learned how,” he said. “The same way that webmasters who could code in HTML became CMOs, people who use agents to do work are the leaders of tomorrow.”