The share of large enterprises running autonomous AI agents in production environments, rather than limited pilots, has roughly doubled over the past six months, according to a new McKinsey survey of more than 800 large-company technology executives.
Customer service automation and software engineering copilots remain the two most mature use cases, while finance and legal functions lag, reflecting both regulatory caution and the higher cost of errors in those domains.
Vendor implications
The data supports the thesis that enterprise software vendors able to demonstrate measurable agent ROI, rather than generic chat interfaces, are capturing a disproportionate share of new enterprise AI budget.