Executives say AI will help them do more with the teams they have. The early labor-market signal points to something sharper: fewer opportunities for the youngest workers in AI-exposed jobs, and fewer rungs in the middle of the org chart. At the same time, a subset of companies using AI at scale report outsize performance and push a shift from static hierarchies to dynamic, project-centered work.
The first cuts are showing up in headcount
A new working paper that links employer payroll data to task exposure offers a clean signal. Early-career workers, ages 22 to 25, in the occupations most exposed to AI suffered a relative 16% decline in employment, while more experienced workers in the same roles and workers in less exposed fields did not see comparable drops (Brynjolfsson, Chandar and Chen, 2025). The adjustment is happening through employment, not wages or compensation, which stayed relatively stable. Where AI tends to automate tasks, employment fell. Where it tends to augment human work, employment held up better. The authors show these patterns are robust to alternative explanations and firm types, but the paper remains a working paper, not yet peer reviewed.
That distinction between automation and augmentation is the crux for leaders. If a job is largely decomposable into tasks that current AI can do alone, headcount will take the hit. If AI can meaningfully assist humans without replacing them, the headcount story looks different. The policy lever inside firms is the design of roles and workflows. Push tasks toward augmentation, and you keep more people in the loop. Let automation run unchecked in entry-level work, and you erase your own talent pipeline first.
Employers plan to recalibrate, not just retrench
Surveys of leaders show expectations that match the tension. The World Economic Forum projects that 22% of jobs will be disrupted by 2030, with 170 million roles created and 92 million displaced, a net gain of 78 million. It also reports that nearly 40% of workplace skills will change by 2030, 63% of employers cite skills gaps as the top barrier to business transformation, 77% plan to upskill their workforce in response to AI, 41% plan to reduce headcount as AI automates tasks, and about 59 of every 100 workers will need some form of reskilling or upskilling. Roughly half of employers expect to reorient the business toward AI-related opportunities (World Economic Forum, 2025). These are employer expectations, not observations, although the report is enriched with labor-market and learning-platform data.
The practical reading is straightforward. Companies are preparing for a double move: invest in skills while also shrinking roles that are mostly automatable. If the Stanford evidence is a guide, the employment adjustment will arrive before retraining catches up, and it will be most acute for early-career workers in exposed occupations (Brynjolfsson, Chandar and Chen, 2025). That argues for targeted on-ramps that build human complementarity quickly, not generic training catalogues that ignore task design.
The great flattening has started
AI is not only reshaping frontline roles. It is also changing what managers do and how many layers a company needs. A global survey of nearly 10,000 business and HR leaders finds managers spend close to 40% of their time on firefighting and administrative work, and only 13% on developing their people. More than a third, 36%, feel unprepared for the people side of the job (Deloitte, 2025). In the United States, employers were advertising 42% fewer middle-management roles at the end of 2024 than in spring 2022, and Gartner, cited by Deloitte, expects that by 2026 about 20% of organizations will use AI to flatten their structures (Deloitte, 2025). This is survey and job-posting evidence, not causal proof, but the direction is clear.
Where does that leave the shape of the firm? Microsoft reports that 82% of leaders plan to use digital agents to expand their workforce’s capacity in the next 12 to 18 months, and that 46% already use agents to automate entire workflows or processes. In companies it labels Frontier Firms, which deploy AI organization-wide and use agents actively, 71% of employees say the company is thriving, compared with 39% globally. The report argues for replacing the static org chart with a “Work Chart,” a dynamic map of outcomes, teams, and agents (Microsoft WorkLab, 2025). The data mix survey responses and Microsoft 365 telemetry, and the source has a commercial interest in AI, so leaders should treat the framing with that context.
Taken together, these signals describe a firm that is thinner in the middle, clearer on outcomes, and more fluid at the edges. Layers that route approvals or collect status are a poor use of human attention if agents can assemble information and run standard processes. What remains for managers is judgment, coaching, escalation, and the stewardship of norms and priorities. The Deloitte figures show how far most managers are from that ideal today (Deloitte, 2025).
How to build a firm that benefits from AI
Start by making outcomes legible. If you want to reduce layers without losing coordination, you need tight definitions of deliverables, service levels, and decision rights. Microsoft’s Work Chart idea is one way to visualize this: map work to goals, teams, and the agents that support them, then adjust assignments as priorities change (Microsoft WorkLab, 2025). The tool is less important than the discipline. People cannot self-manage if they cannot see what good looks like.
Protect and redesign early-career work. The Stanford evidence implies that the entry point into exposed professions is where employment risk concentrates (Brynjolfsson, Chandar and Chen, 2025). Do not automate the whole ladder away. Carve out apprentice-grade tasks that pair humans with AI for learning and feedback. Treat AI as a co-pilot to speed skill formation, not as a wall that hides the craft until mid-career.
Repurpose the manager role around coaching and escalation. If managers currently spend 40% of their time on administrative churn and only 13% on people development, they are misallocated for an AI world where agents can handle standard reporting and scheduling (Deloitte, 2025). Rewrite job descriptions and metrics to reward coaching, cross-team problem solving, and rapid decision-making. Expect fewer layers and more span of control, with clearer playbooks and better tooling.
Invest where augmentation beats automation. Use a task-level audit to separate work where AI can replace people from work where it can elevate them. Push the boundary of augmentation by redesigning processes to keep humans in the loop for exceptions, synthesis, and stakeholder context. Where automation is the right call, plan the redeployment before you pull the headcount. The World Economic Forum’s projection that 59% of workers will need reskilling or upskilling gives you the order of magnitude, even if it is based on employer expectation rather than observed outcomes (World Economic Forum, 2025).
The close-in payoff is retention of the skills you still need while you thin the layers you do not. The longer-term payoff is an organization that runs on clear outcomes, transparent work, and teams that can re-form fast. AI will flatten by default. The firms that benefit will be those that choose what stays standing: judgment, trust, and a manager’s core craft of helping people do the best work of their careers.
Sources
- Brynjolfsson, Chandar and Chen (2025). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of AI. Early-career workers in AI-exposed jobs saw a 16% relative employment decline, and adjustment occurs via employment rather than wages, with losses concentrated in automating tasks. https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/
- World Economic Forum (2025). The Future of Jobs Report 2025. Employer expectations point to 22% job disruption by 2030, net job growth of 78 million, large skills shifts, widespread upskilling plans, and some headcount cuts tied to AI. https://www.weforum.org/publications/the-future-of-jobs-report-2025/
- Deloitte (2025). 2025 Global Human Capital Trends. Managers spend heavy time on admin and little on development; mid-management postings are down 42% in the US and Gartner projects 20% of firms will flatten structures with AI by 2026. 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. Leaders report expanding capacity with agents, significant workflow automation, and stronger performance signals in Frontier Firms; introduces the Work Chart concept. https://www.microsoft.com/en-us/worklab/work-trend-index/2025-the-year-the-frontier-firm-is-born