AI agents to handle 8 job types, leaving 5 to humans

The article addresses a common worry: that AI will replace jobs. It argues instead that work is decomposed into tasks, and agents take over a portion, not the entire occupation. This is supported by Anthropic's data showing that as of November 2025, 52% of Claude.ai conversations assisted human work, while 45% delegated tasks entirely—less than half. The World Economic Forum projects a net gain of 78 million jobs by 2030, with 170 million created and 92 million lost, while 39% of skills change or become obsolete.

For non-technical managers, the practical takeaway is the 4-axis test: can the procedure be written down, can errors be reversed, can the output be verified, and is the impact internal? Tasks passing all four are safe to delegate first. The article also warns about risks: Gartner predicts over 40% of agentic AI projects will be canceled by end of 2027 due to cost and unclear benefits. It recommends limiting permissions and keeping human approval steps.

Ultimately, the article suggests that as delegation increases, the value shifts from doing tasks quickly to designing delegation and verifying outputs. It advises starting with one task that meets all criteria and expanding from there, while recording what you do with the saved time to show contributions in evaluations.