The AI Skill Companies Are Missing Is One Operations Managers Already Have

Nearly nine in ten organisations now use AI somewhere in the business. Very few are getting real value from it. The difference is not the technology. It is whether anyone redesigned the work around it — and that is an operations skill.

The 21% problem

McKinsey's latest State of AI research puts a number on something operations people have sensed for a while. AI adoption is close to universal. Redesigning the work around AI is not. Most organisations have added AI on top of processes designed for a world without it.

That gap shows up in the results. McKinsey identifies a small group of AI high performers — roughly 6% of organisations — that attribute meaningful bottom-line impact to AI. High performers are nearly three times as likely to have fundamentally redesigned individual workflows. In McKinsey's analysis of 31 organisational practices, intentional workflow redesign made one of the strongest contributions to real business impact.

The differentiator is process work

Not model selection. Not prompt libraries. Not which vendor won the pilot. The practice that most separates organisations getting value from AI is the deliberate redesign of how work flows: what gets automated, what gets escalated, where a human checks the output, and how exceptions are handled.

There is a name for the person who maps processes, finds the handoffs that break, decides where controls sit, and redesigns the flow. That person is an operations manager. This has been the operations toolkit for decades. The AI era has not made it obsolete. It has made it the scarcest input to AI success.

Where the humans belong

One more data point deserves attention: defined human-in-the-loop validation. High performers know exactly when and how AI output gets checked by a person. Most organisations do not.

Deciding where the human belongs in a process is judgment work. It requires knowing which errors are recoverable, which steps carry regulatory weight, and where a wrong output causes downstream damage. In finance operations, our editorial line applies: AI drafts the narrative, never the ledger. In HR operations: AI assists, never decides the hire. Drawing those lines, function by function, is exactly the work the data says organisations are failing to do.

What to do with this

• Reframe your experience. Process mapping, exception design, and controls are AI-enablement skills now. Describe them that way — in conversations, in reviews, and on your profile.
• Volunteer for the redesign, not just the rollout. When your organisation deploys an AI tool, the valuable seat is not user number 200. It is the person deciding how the workflow changes around it.
• Learn enough AI to hold the conversation. You do not need to build models. You need to understand what AI does reliably, where it fails, and how to design checks around both. Our function-by-function guides are built for exactly this.

The encouraging conclusion: organisations do not have too few AI tools. They have too few people who can rebuild work around them. Operations professionals are closer to that role than anyone else in the building.