McKinsey's April 2026 Quarterly report finds that 88% of organizations are using AI in at least one business function, and 94% report no significant returns. The reason is not adoption. It is framing. Productivity gains from AI are largely defensive: competition erodes them, and customers capture most of the surplus. McKinsey models AI value across three waves: productivity, differentiation, and market structure transformation. Most companies are stuck in Wave 1, doing the same work faster.
The trap is clearest in product discovery. Teams are compressing six-week research cycles into ten days, auto-clustering interviews, and generating opportunity trees from prompts. None of that changes what questions get asked. Brynjolfsson, Li, and Raymond's 'Generative AI at Work,' studying 5,000+ customer support agents, found a 14% average productivity gain, but novice workers improved 34% while experienced workers saw marginal gains. Speed without changed questions produces the same answers sooner. The McKinsey framing, the Brynjolfsson data, and the factory-electrification parable from Agrawal, Gans, and Goldfarb all point to the same structural failure: optimizing the process before reconsidering the problem.
The article is worth reading in full because the argument does not stop at diagnosis. It builds a specific reframe: shift from 'how do we make discovery faster' to 'what would be worth building if AI reshaped the answer space entirely.' That is a strategic question, not a process one, and the piece works through what it actually looks like for teams deciding what to build next.
[READ ORIGINAL →]