MIT Sloan and BCG Surveyed 2,102 Executives Worldwide. 45% Expect Agentic AI to Flatten Middle Management. Most Haven't Redesigned a Single Process.

October 14, 2026 · Josh Agusti, Founder, Aiven Consulting Group

MIT Sloan Management Review and Boston Consulting Group released the ninth annual installment of their global AI and business strategy study in November 2025, titled The Emerging Agentic Enterprise: How Leaders Must Navigate a New Age of AI. The research draws on a survey of 2,102 executives across 21 industries and 116 countries, along with interviews with senior leaders at organizations furthest along in agentic AI adoption.

The headline finding is about the org chart, not the technology. Among the leaders whose organizations have gone furthest with agentic AI, 45 percent expect a reduction in middle-management layers. Forty-three percent expect to become more open to hiring generalists rather than specialists, and 29 percent expect fewer entry-level roles. Seventy-six percent of executives surveyed said they now view agentic AI more as a co-worker than a tool. By the researchers' count, 35 percent of companies have already begun using agentic AI and another 44 percent plan to deploy it soon, putting the practice within reach of nearly four in five organizations.

The Shape of the Org Chart Is Already Changing in Leaders' Minds

Agentic AI differs from earlier waves of automation because it touches process, roles, decision rights, and accountability at the same time, rather than one function at a time. That is the report's central argument, and it explains why the middle-management finding lands differently than a typical productivity statistic. A chatbot that drafts emails faster does not change who is accountable for a decision. An agent that can plan a project, assign tasks, and flag exceptions without a human in the loop starts to resemble the actual job description of a first-line manager, which is why leaders surveyed are picturing fewer of those positions rather than just faster ones.

What the study does not find is an organization that has already finished this transition. The leaders expecting a flatter structure are, by the researchers' own framing, anticipating a shift rather than reporting one they have completed. Most of the companies surveyed have not yet redesigned the processes their agents are meant to run; they have added agents to workflows built for a different org chart and are watching what happens. Deciding what an operating model should look like once agents take on coordination work is a strategy exercise, not a deployment task, and it is the kind of decision Aiven Consulting Group's AI strategy and operating model work is built to help leaders make deliberately rather than by accident.

Generalists In, Entry-Level Out: Where the Pressure Actually Lands

The generalist and entry-level findings point to a second-order effect that gets less attention than the headline number. If agents absorb the coordination and routine-judgment work that used to be a manager's daily job, the people left in the middle need to be comfortable moving across functions rather than staying deep in one lane, which is consistent with the 43 percent of leaders who said they expect to be more open to hiring generalists. Other 2026 workforce research has sketched a similar shape from a different angle, describing knowledge work organizing into a thinner middle flanked by stronger junior and senior tiers, with someone in the organization responsible for an orchestration layer that did not exist a few years ago.

None of this is destiny. It is a set of expectations held by executives who are, by their own admission, acting ahead of a finished plan. The 29 percent who expect fewer entry-level roles are describing a risk to the talent pipeline that feeds every other tier of the organization, and that risk is easiest to manage before the restructuring happens rather than after.

The Governance Question Nobody Asked Yet

Middle managers do more than assign work. They absorb ambiguity, catch exceptions before they become incidents, and serve as an informal accountability layer that most companies never wrote down because a person was always there to fill it. Removing layers of that function without replacing what it did is where agentic AI's organizational risk actually sits, separate from whatever happens to headcount. The 76 percent of executives who now describe agents as co-workers rather than tools are, whether they intend to or not, describing something that needs the same clarity of ownership a human co-worker would get. Organizations that have not yet written down who is accountable when an agent that used to report to a manager now reports to no one in particular can start with a structured AI governance framework rather than discovering the gap after an agent's decision goes unreviewed.

Decide the Shape Before It Gets Decided For You

The organizations the MIT Sloan and BCG researchers describe as furthest along did not arrive at a flatter structure by letting agents accumulate responsibility until the org chart caught up. They are the ones treating the shape of the organization as a decision to make on purpose, with a view of what each agent is doing, who still owns the judgment calls, and what happens to the people whose jobs used to sit in between. Leaders who want an honest picture of where their own organization stands before committing to a restructuring bet can start with Aiven Consulting Group's AI readiness assessment, which measures the actual footprint of agents already in use rather than the one implied by a forecast. The 45 percent expecting a flatter middle are ahead of where most organizations are today. Whether that turns out to be a well-planned transition or an accidental one will depend on which of those two groups does the planning first.

Key Takeaways

MIT Sloan Management Review and BCG's global study of 2,102 executives across 21 industries and 116 countries found that among leaders furthest along with agentic AI, 45 percent expect their organizations to see a reduction in middle-management layers.
The same study found 43 percent of these leaders expect to become more open to hiring generalists over specialists, and 29 percent expect fewer entry-level roles, meaning agentic AI is reshaping the entry points into an organization as much as its middle.
Despite expecting this restructuring, most organizations in the study have not yet redesigned the processes their agents are meant to run, leaving a gap between anticipating organizational change and actually planning for it.