Few corporate battles are as raw as the one now playing out between AI buildouts and headcount budgets. Boards want productivity gains fast. Managers feel their span of control is shrinking. Early-career workers worry they will not get a fair first rung on the ladder. The data suggests all three instincts are partly right.
The flattening is not hypothetical
Across markets, signals point to a leaner middle. Employers in the United States advertised 42% fewer middle-management jobs at the end of 2024 compared with spring 2022, and Gartner, cited in the same analysis, projects that by 2026 one in five organizations will use AI to flatten structures (Deloitte, 2025). Inside firms, leaders are already automating entire workflows with software agents. In Microsoft’s global survey of 31,000 knowledge workers, 46% of leaders say they already use agents to fully automate workflows or processes, and 82% say they will use digital agents to expand capacity over the next 12 to 18 months (Microsoft WorkLab, 2025). The report promotes a switch from a static org chart to a dynamic “Work Chart,” a model organized around outcomes rather than boxes and lines (Microsoft WorkLab, 2025). Treat these figures with context. They are expectations and telemetry reported by a vendor with a commercial interest in AI. Still, at the very least they show that demand for automation is widespread, and that an outcome-based map of work is becoming a practical management tool rather than a futurist sketch.
A skills shock, and an early-career cliff
AI is not only a headcount story, it is a skills story. Employers expect major shifts in what people must be able to do. The World Economic Forum projects that 22% of jobs will be disrupted by 2030, with 170 million roles created and 92 million displaced, for a net gain of 78 million jobs (World Economic Forum, 2025). Almost 40% of skills required at work are expected to change by 2030, and 63% of employers cite the skills gap as their top barrier to business transformation (World Economic Forum, 2025). The response is a mix of investment and cuts. Seventy-seven percent of employers plan to upskill their workforce in response to AI, 41% say they will reduce headcount as AI automates tasks, and about half plan to reorient their businesses toward AI-created opportunities (World Economic Forum, 2025). The scale of the reskilling challenge is blunt: out of every 100 workers worldwide, 59 will need reskilling or upskilling (World Economic Forum, 2025). These are employer expectations, not observed outcomes, but they map the scale of change leaders are preparing for.
Where change is observed, it is sharpest for the youngest. Using payroll data, Brynjolfsson, Chandar and Chen find that early-career workers aged 22 to 25 in the occupations most exposed to AI faced a relative employment decline of about 16%, while more experienced workers in those jobs and workers in less exposed fields held steady or grew (Brynjolfsson, Chandar and Chen, 2025). The adjustment is happening through employment, not wages, and losses concentrate where AI automates tasks rather than augments them (Brynjolfsson, Chandar and Chen, 2025). It is a working paper, but it uses large-scale administrative data and tests alternative explanations. The practical read is clear. If junior roles become the automation target, firms must design new on-ramps that put early-career people into augmentation-rich work with real coaching, or risk starving their own future leadership bench.
What management is for now
Much of what clogs a manager’s week is precisely what software agents will absorb. Managers report spending nearly 40% of their time on firefighting and administrative work, and only 13% developing their people. More than a third, 36%, feel unprepared for the people side of the job (Deloitte, 2025). If that mix persists, automation will erase a large share of what many managers currently do, and with it the economic case for middle layers. That does not make management obsolete. It makes shallow management indefensible.
Management that survives will be the kind that software does not do well: composing missions, setting standards, shaping culture, coaching judgment, and coordinating across boundaries where incentives and contexts clash. Those jobs get easier when the work is mapped to outcomes instead of titles. The “Work Chart” idea captures the point. If teams are defined by the results they own, and if human roles and AI agents are explicit inside each workflow, leaders can see where coordination and capability building are scarce and invest accordingly (Microsoft WorkLab, 2025). The label comes from a vendor and the data is survey-based, but the operating logic holds across toolsets.
A practical playbook for the next 12 months
Leaders do not need a manifesto. They need a sequence.
- Run an automation audit tied to outcomes. Inventory workflows, not jobs. Mark the steps already automated and those feasibly automated within 12 months. Use that to identify which managerial tasks are administrative scaffolding, and plan to retire them. The Microsoft data shows many leaders are already automating end to end, which makes this a competitive baseline rather than a stretch goal (Microsoft WorkLab, 2025).
- Redesign the manager job around human-only value. Shift expectations and incentives from activity tracking to capability building and cross-team coordination. The time-use data shows how far off the current mix is, and why it is vulnerable (Deloitte, 2025).
- Protect early careers by moving juniors to augmentation-heavy work. The employment hit is concentrated where AI automates tasks. Give early-career hires rotations that pair them with AI tools in analyst, product, customer, or operations roles where human judgment matters, and attach them to managers measured on people development. The Stanford evidence suggests this will blunt exposure where it is sharpest (Brynjolfsson, Chandar and Chen, 2025).
- Fund reskilling at scale, and measure it. Employers say almost 40% of skills will change by 2030, and most plan to upskill, but intentions are not outcomes. Set targets for completions and on-the-job application, and publish the numbers internally. The WEF figures make the size of the task unambiguous: a majority of your workforce will need new skills, and failing to plan invites a skills gap employers already rank as their top barrier (World Economic Forum, 2025).
The point is not to flatten for its own sake. It is to strip away layers that only transferred information and approvals, then use the saved capacity to do the kind of management a machine does not. That is what will keep your org chart smaller, your work chart clearer, and your talent more valuable. The firms that move fastest on this will not be the ones that bought the most AI. They will be the ones that made management worth keeping.
Sources
- Deloitte (2025). 2025 Global Human Capital Trends. Documents the drop in middle-management postings, managers’ time use, and Gartner’s forecast that AI will flatten structures, grounding the case for a leaner middle and a redesigned manager role. https://www.deloitte.com/us/en/insights/topics/talent/human-capital-trends/2025.html
- Microsoft WorkLab (2025). Work Trend Index 2025: The Year the Frontier Firm Is Born. Reports leaders’ current and planned use of AI agents, introduces the Work Chart concept, and contrasts outcomes at firms deploying AI at scale, supporting the shift to outcome-based operating models. https://www.microsoft.com/en-us/worklab/work-trend-index/2025-the-year-the-frontier-firm-is-born
- World Economic Forum (2025). The Future of Jobs Report 2025. Provides employer expectations on job disruption, skills change, upskilling plans, and headcount reductions, establishing the scale and direction of workforce transformation. https://www.weforum.org/publications/the-future-of-jobs-report-2025/
- Brynjolfsson, Chandar and Chen (2025). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of AI. Using ADP payroll data, finds early-career employment declines in AI-exposed occupations and that adjustment occurs via employment, underscoring risks for juniors and the importance of augmentation pathways. https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/