Leadership: augment, don't just cut

Executives pushing AI into the workplace are increasingly being told to redesign jobs around judgment and let software agents handle the repeatable work, not just cut headcount. (mitsloan.mit.edu) That advice tracks with the rise of “agentic” artificial intelligence: software that can plan, use tools and complete multistep tasks with limited supervision, rather than just answer prompts like a chatbot. Amazon Web Services said in June 2025 that autonomous agents can compile research, manage applications and complete work “in tandem with – or on behalf of humans.” (aws.amazon.com) McKinsey said in late 2025 that the “agentic organization” pairs humans and AI agents side by side, with leaders rethinking workflows instead of dropping software into old org charts. MIT Sloan reported in February 2026 that agents are already being deployed at scale to perceive, reason and act in digital environments. (mckinsey.com) (mitsloan.mit.edu) The labor backdrop is why the leadership debate has shifted from pure efficiency to role design and retraining. The World Economic Forum said in January 2025 that employers expect 22% of today’s jobs to be disrupted by 2030, with 170 million roles created and 92 million displaced. (weforum.org) The International Labour Organization said in its 2025 update that generative AI is more likely to transform jobs than eliminate them outright, with clerical work facing the highest exposure. The Organisation for Economic Co-operation and Development says AI will complement humans in some tasks, replace them in others and create new work in the process. (ilo.org) (oecd.ai) That is where the “augment, don’t just cut” argument lands: move people toward strategy, exception handling and accountability, while agents take on documentation, routing, drafting and other repetitive tasks. A 2025 management paper in the *Journal of Organization Design* said operational decisions are more likely to be automated, while higher-stakes tactical and strategic decisions are more likely to be shared between humans and AI. (link.springer.com) The same argument usually comes with a second demand: measure the transition. McKinsey said agents can automate complex processes, but companies need a plan for where autonomy starts, where humans approve work and how performance is tracked. (mckinsey.com) Governance is the other half of the playbook. KPMG said in an October 2025 report that scaling AI agents is “not just a tech conversation” and requires leadership alignment, execution plans and preparation for the workforce. (assets.kpmg.com) That means oversight for bias, security, privacy and errors, especially when agents act on their own inside finance, hiring, customer service or software systems. Augment Code, which sells coding agents, said in May 2025 that trust depends on giving agents enough context to follow architecture, policies and best practices rather than generate plausible but faulty work. (augmentcode.com) The thread running through the latest leadership advice is narrower than the hype: automate the busywork, keep humans on the hook for judgment, and treat retraining and governance as operating requirements, not public-relations add-ons. (weforum.org)