In the Age of AI Agents, Will Middle Managers Really Disappear? [Kim Mun-Gyeong’s Leadership Tech] (31)

[Financial News] AI agents (AI Agent) analyze market data on their own, calculate optimal selling prices, and even draft personalized proposals for customers. AI, which once waited for instructions from people, has now entered a stage where it autonomously plans and carries out tasks. As the technology pushes deep into the realms of day-to-day work and coordination, executives are busy running the numbers. If AI takes over the work of middle managers, is there really any reason to keep so many managers around?


An op-ed by Kai-Fu Lee, CEO of 01.AI, published recently in the Financial Times (FT), directly challenges this push to shrink organizations. While AI may thin management layers by taking over information sharing and routine operational coordination, the value of managers responsible for final decisions and accountability will actually increase. If companies hastily eliminate middle managers under the banner of AI adoption, who will shoulder the highly complex judgment and coordination responsibilities they handled?
Imagine a company deploying three AI agents in its sales department, each responsible for market analysis, pricing, and drafting customer proposals. All three produce results at remarkable speed in pursuit of their assigned goals. The problem arises not when AI gives a wrong answer, but when correct answers optimized for each agent’s individual goal are brought together. A proposal to cut prices to expand market share could directly conflict with company-wide guidance to protect profitability, or an extraordinary customer benefit could place a crippling strain on the production line. AI may be able to connect day-to-day work, but deciding what to prioritize among conflicting business goals—and who will be accountable for the outcome—is an entirely different matter.
An organizational analysis published by McKinsey this year illustrates this vast gap precisely. Although 80% of respondents said AI had improved their individual productivity, only 37% said this had led to an increase in their company’s operating profits. In other words, an individual finishing work faster and a company actually making money are separate things. Bridging this gap and integrating fragmented work outputs into company-wide results—that is orchestration, something only human managers can do.
Moreover, organizational performance does not depend solely on managers making sound decisions. How employees respond to the AI transition and how much they trust their organization are also key factors. According to another McKinsey survey (July 2026), 70% of respondents said they were ready to use AI, but only 27% of leaders judged that their organizations were ready to embrace the changes needed for the AI transition. This suggests that people are already prepared for change, while the systems and culture of their organizations are failing to keep pace.
McKinsey found that, throughout the process of adopting AI, “trust in the organization” is the most important factor directly linked to corporate value creation. Simply adopting the latest AI is not enough. Organizations must provide reskilling and psychological safety so employees understand the reasons for change and can take on new roles. Before placing blind faith in technology, organizations should first ask whether employees can truly trust and follow the company through this sweeping transformation.
These technological and environmental changes are fundamentally reshaping even the criteria used to evaluate middle managers. In the past, the benchmark was the span of control, which asked how many employees a manager supervised. Now it has evolved into the span of complexity, which asks how well a manager can handle complex decisions and business risks as people and AI work together. A manager’s true value is demonstrated not by the number of employees under their supervision, but by how weighty the decisions and outcomes they can take responsibility for are.
Yet managers on the ground are already reaching their limits. According to a recent Gartner survey, 79% of chief human resources officers (CHROs) said managers in their organizations are currently experiencing severe overload. AI adoption may have reduced workers’ routine tasks, but that does not mean the exceptions managers must handle or the complexity of their decisions will simply disappear. Gartner warned that 75% of organizations that cut middle managers in the name of AI investment will end up rehiring more than half of the management positions they eliminated by 2028. It is a sobering reminder that automating a manager’s “work” and eliminating the organization’s “management function” altogether are entirely different things.
Therefore, middle managers in the age of AI must move beyond simply assigning tasks and checking on progress. They must become conductors who design the roles of people and AI in a comprehensive way, and strictly distinguish between areas machines can handle independently and those that require human intuition. They need the authority to set business priorities when departmental goals conflict and to immediately halt AI when it goes off course. Providing employees with psychological support so they do not lose their way amid the turbulence of change is also a weighty responsibility for leaders.
Managers whose only reason for being was to compile other people’s reports and send them up the approval chain deserve to become obsolete. But it is extremely dangerous for executives, fixated solely on the numbers behind organizational downsizing, to indiscriminately cut management layers on the ground. If decision-making authority evaporates and only mechanical execution remains, or if the entire burden of responsibility falls on a single CEO, that is not innovation—it is the beginning of organizational paralysis and a crippling bottleneck.
AI readily takes over human labor, but it does not take responsibility for business outcomes. What executives need to ask themselves right now is not how many middle managers they can cut. Who will critically assess the results AI produces, who will reconcile conflicting interests across departments, and who will ultimately be accountable for the results? Answering these fundamental questions before rushing to run the numbers is the starting point for genuine organizational innovation in the age of AI.
/Kim Mun-Gyeong (Adjunct Professor at Kookmin University and Vice President of the Korean Leadership Society) 



[email protected] Jung Myung-jin, Medical Specialist Reporter