Three Workforces, Three Scorecards
The next operating model has to compare humans alone, humans with AI, and governed AI capacity without pretending they are the same.
Most workforce planning still assumes one unit of capacity: a person.
That assumption is breaking.
In an AI-enabled organization, leaders will manage at least three kinds of capacity: humans alone, humans with AI, and governed AI agents. Each can create value. Each creates different risk. Each needs a different scorecard.
Humans alone remain essential where judgment, trust, leadership, emotional context, negotiation, and accountability matter most. Their scorecard should emphasize decision quality, relationship strength, leadership impact, client trust, and the ability to navigate ambiguity.
Humans with AI are different. The question is not whether they are “using AI.” The question is whether AI improves the work. Their scorecard should measure cycle time, quality, leverage, review discipline, and the employee’s ability to turn AI output into sound decisions. A human with AI should not be evaluated only on speed. Speed without discernment is just accelerated rework.
Governed AI agents require a third scorecard. They should be measured as managed capacity: completion rate, error rate, cost per task, escalation accuracy, exception handling, auditability, and risk exposure. The organization should know where agents outperform human workflows, where they require supervision, and where they should not operate at all.
The hard part is not building three scorecards. The hard part is refusing to collapse them into one.
If leaders compare humans and agents only on cost, they will underprice trust. If they compare AI-augmented employees only on output volume, they will miss the quality of judgment. If they ignore agents because they are “just tools,” they will miss the fact that work has already moved.
The better question is: what kind of capacity does this work require?
Some work needs human judgment. Some needs human judgment amplified by AI. Some can be delegated to governed AI capacity. A modern workforce operating system should make those distinctions visible, measurable, and manageable.