Giving every employee an AI tool changes access to capability. It does not automatically change how work moves. Old handoffs, unclear decision rights, duplicated reviews, and weak data can absorb the speed gained at an individual task.
The decision behind the headline
McKinsey’s research on people, agents, and robots argues that the larger opportunity comes from redesigning complete workflows rather than automating isolated tasks. The report estimates technical potential, not guaranteed job loss or economic value. Its useful contribution is the unit of analysis: tasks sit inside workflows, roles, controls, and relationships.
What the evidence supports
A workflow redesign should answer:
- which steps are routine enough for an agent or automation;
- where human judgment, context, trust, or accountability remains essential;
- who owns exceptions and when the system must escalate;
- what skills shift from execution toward framing, verification, and coordination;
- how performance is measured for the combined human-AI system.
McKinsey reports that more than 70 percent of skills sought by employers appear in both automatable and non-automatable work. That suggests many skills will be applied differently rather than simply disappear.
Social verdict
Ethan Mollick wrote on X that work will change around AI, but the direction is not predetermined: firms can use it only to cut roles or to expand what people can do. A Reddit discussion of the McKinsey report similarly focused on workflow redesign as the path from tool deployment to measurable value. Both are interpretation, not proof of a specific outcome.
Current verdict
Start with one end-to-end workflow and map its decisions, evidence, handoffs, exceptions, and owner. Move routine preparation to AI only where the next human can inspect it quickly. Redesign the role around the judgment and coordination that remain, then measure the workflow outcome rather than counting tool usage.