The price of intelligence and the manufacturing of experience

🕒 Published on Zendoric: July 25, 2026 · 00:23
This edition of The CEO Shortlist, McKinsey's biweekly newsletter curated by Alex Panas (global leader of industries) and Becca Coggins (global leader of functional practices and growth platforms), tackles two central questions about agentic AI: whether agents are really delivering what they cost, and…
By McKinsey CEO Shortlist.
This edition of The CEO Shortlist, McKinsey's biweekly newsletter curated by Alex Panas (global leader of industries) and Becca Coggins (global leader of functional practices and growth platforms), tackles two central questions about agentic AI: whether agents are actually delivering what they cost, and what will happen to the training of new professionals as AI absorbs entry-level work.
On the first point, the newsletter cites a recent McKinsey survey in which 93% of respondents admit to having gone over their AI budget, even though the nominal price of tokens keeps falling. According to McKinsey, the real culprits are not the tokens themselves but rework and variability: roughly 60% of the cost of an agentic task goes to reviewing and correcting the first response generated, and the same task can cost up to 30 times more depending on the model used. The conclusion of the McKinsey colleagues cited in the newsletter is that the CEO's job is not to squeeze tokens, but to govern the work of machines: to allocate artificial intelligence as if it were capital and to measure agents by the business results they produce, not by technical metrics.
The newsletter's second theme is what they call "the disappearing bottom rung." Professional experience has traditionally been forged through routine tasks — basic research, first drafts — that teach judgment through repetition. That is precisely the kind of work AI is already absorbing: it multiplies the output of experienced employees but slows down novices, who cannot always spot when the machine gets it wrong. The newsletter notes that a third of employers say AI has already reduced the tasks junior staff used to learn the trade on. McKinsey's warning is clear: a company that dismantles the base of its talent pyramid will have no one to promote a decade from now. The newsletter points to a longer McKinsey article (linked as "how this pipeline can be rebuilt") exploring how to rebuild that talent-development process.
The rest of the email includes the usual elements of McKinsey's newsletter format: a reading recommendation (McKinsey's annual book collection), links to follow its McKinsey Insights app, and the standard sections for sharing/forwarding the newsletter and managing your subscription.
🔗 Related on Zendoric


