McKinsey survey finds AI agents used widely, scaled by only 10 percent
TL;DR
On September 14, 2026, AlgeriaTech summarized McKinsey’s State of AI 2026 survey, reporting that 62 percent of organizations are experimenting with AI agents but under 10 percent have scaled them in any given business function. The piece notes that only about 6 percent of companies attribute significant, broad EBIT impact to AI, despite 88 percent using AI regularly and 72 percent using generative AI.
About this summary
This article aggregates reporting from 1 news source. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Read next
Race to AGI Analysis
McKinsey’s numbers, as relayed here, are a useful correction to the hype around AI agents. If nearly nine in ten organizations say they use AI and almost three quarters use generative models, but fewer than one in ten can scale agents in any function and only 6 percent see large EBIT contributions, we are still early in turning raw model capability into durable business value. The constraint is not access to models. It is workflow redesign, evaluation, governance and cost control.
From an AGI perspective, this suggests a widening gap between a thin layer of highly disciplined adopters and a long tail of companies burning money on poorly integrated tooling. The former group will learn rapidly how to orchestrate agents across complex processes, building tacit know how that matters more than raw parameter counts. The latter risk souring boards and regulators on AI altogether if promised gains fail to show up.
For the frontier labs, these findings are a reminder that competitive advantage may come less from squeezing a few more points out of benchmarks and more from helping customers navigate organizational change. An AGI grade system dropped into a company that cannot redesign workflows or manage agent behavior will underperform badly, blunting its real world impact even if the science is impressive.
