External signal·McKinsey & Company·Sep 1, 2025·Alexander Sukharevsky, Alexis Krivkovich, Arne Gast, Arsen Storozhev, Dana Maor, Deepak Mahadevan, Lari Hamalainen, and Sandra Durth·15 min read
The agentic organization: Contours of the next paradigm for the AI era
“Organization charts (based on traditional hierarchical delegation) will pivot toward agentic networks or work charts (based on exchanging tasks and outcomes).”
Summary
McKinsey's People & Organizational Performance Practice argues AI is driving the largest organizational paradigm shift since the industrial and digital revolutions, toward what it calls the "agentic organization" — humans working alongside virtual and physical AI agents at near-zero marginal cost. Drawing on work with early adopters, it lays out five pillars (business model, operating model, governance, workforce, and technology and data) and 15 transformation themes. Concrete examples include a global bank's "agent factory" running know-your-customer processes with ten agent squads, and another bank cutting legacy-system modernization time and effort by up to 50 percent with humans "above the loop." It claims a human team of two to five people can already supervise an agent factory of 50 to 100 specialized agents on an end-to-end process such as onboarding a customer or closing the books, and that the traditional org chart will give way to a flatter "work chart." It estimates 89 percent of organizations still operate an industrial-age model, 9 percent a digital one, and only 1 percent as a decentralized network.
Predictions for the future of work
The piece forecasts that work will be redesigned "AI-first," with humans moving from executing tasks to steering end-to-end outcomes, and operating models flattening into networks of small, outcome-aligned "agentic teams." It predicts three emerging human archetypes — M-shaped supervisors (AI-fluent generalists who orchestrate agents), T-shaped deep specialists (who handle exceptions and safeguard quality), and AI-empowered frontline workers (focused on interpersonal tasks) — alongside executives needing CIO-level technology fluency and reskilling at scale. Citing METR, it notes the length of tasks AI can reliably complete has doubled roughly every four months since 2024 and could reach four days of unsupervised work by 2027. The timeframe is near-to-medium term, with human oversight capacity flagged as the binding constraint on how fast agentic adoption can scale ("governance itself a potential bottleneck to productivity").
Originally published by McKinsey & Company · Sep 1, 2025
Read the original at McKinsey & Company