AI Transformation Has a Workforce Capability Problem.

The short answer. AI adoption is high and enterprise returns are flat, and the model is not why. McKinsey's August 2026 survey found 80% of respondents say AI has improved their individual productivity and only 37% attribute at least some EBIT impact to it, about the same as a year earlier. Deloitte's 2026 survey of more than 3,200 leaders finds the same story from the people side: insufficient worker skills are the biggest barrier to using AI in existing workflows, and 84% of companies have not redesigned jobs around what AI can now do. The companies closing the gap make that capability visible. The rest are still counting licenses.

The return on AI is stuck, and the model is not the reason. McKinsey's State of AI survey, published in August 2026, found that 80% of respondents say AI has improved their individual productivity, but just 37% attribute at least some EBIT impact to AI, about the same as a year ago. In Deloitte's 2026 survey of more than 3,200 leaders, insufficient worker skills are the biggest barrier to integrating AI into existing workflows. The tools work. The organization around them has not caught up.

RCM ThinkLabs (rcmlabs.io) is the human capability layer for AI workforce transformation, a game-theory engine for measuring decision performance. It exists because the capability that actually decides whether AI pays back, judgment under pressure, does not show up in a usage dashboard.

---

What do the companies getting a return do differently?

McKinsey's high performers are about 6% of companies. They are 2.8 times more likely to report fundamentally redesigning a workflow around AI: 55% of them, against 20% of everyone else. Deloitte sees the same gap from the other direction: 84% of companies have not redesigned jobs around AI capabilities.

Redesigning work is a people question before it is a process question. A redesigned workflow asks someone to decide where the machine stops, to check output they did not write, and to coordinate with colleagues whose work now looks different. Those are behaviors. Almost no company measures them.

Why does the usual response miss?

Deloitte also reports that the most common response is to educate the broader workforce to improve AI fluency, not redesign roles. It is the comfortable move. It produces completion rates. It does not tell a COO whether the team will catch a confident wrong answer on a Tuesday afternoon. Only 20% of leaders in the same survey rated their organization highly prepared on talent, the lowest score of any readiness dimension Deloitte measured.

What does visible capability look like?

The gap closes when capability becomes visible. That means watching how people reason in situations that feel real, every day, and scoring the decisions they make instead of the modules they finished.

We build that instrument, so here is what it actually does. In a live deployment with an advanced engineering team, 70% of participants engaged voluntarily every day, 84% of regular participants improved on a full read, and more than 15,000 decisions were scored across 1,088 sessions in the first operating cycle, each person measured against their own opening baseline.

AI transformation will be decided by the people using it, not the model underneath it. The companies pulling ahead have already worked that out. The rest are still counting licenses.

See what visible capability looks like. Read how RCM ThinkLabs builds the human capability layer for AI transformation, or see the evidence.

Sources. McKinsey & Company, “The State of AI in 2026: On the Road to ROI,” August 2026. Deloitte, “The State of AI in the Enterprise, 2026.”