Agentic AI is moving from experiments to production, but the human and operational side is lagging. The Register reports that Gartner expects deployment to jump from 17% to 60% of businesses over the next two years. Yet a WalkMe study cited in the piece finds 40% of digital transformation expenditure underperforms, mainly because employees are not adopting the tools effectively. The average worker loses 7.9 hours a week to friction, including re-entering data and reworking AI prompts.
Visibility is a core problem. Executives believe their organizations use about 35 applications; the real count is 661, according to the report. For AI tools, the estimate is 21 versus 80. Shadow AI is widespread: 45% of workers admitted using unapproved AI tools in the previous 30 days, and 36% of those used confidential company data. That makes governance and workflow optimization difficult.
ROI is also murky. Many financial cases assume productivity gains or headcount cuts, but the real benefits may lie in better work quality and human augmentation. Token consumption and costs are often unknown. WalkMe's Ofir Hatsor says the gap between perceived and actual value could take three to seven years to close, and may never close at some companies.