When we talk about AI, one of the most common terms is “efficiency.”
A task that once took an hour now takes ten minutes; work that needed three people can be done by one person with AI.
Those are real changes.
But I increasingly think efficiency is only the first layer of change AI brings.
The more significant change is this:
AI is changing the assumptions on which many traditional ways of working depend.
Why was work divided into so many roles? Why did it need so many documents, meetings, approvals and handoffs?
Often, that was not because the work inherently had to happen that way. It was because one person's capabilities were limited.
Each person could do only part of it, so we divided the work. Division created a need for collaboration. Collaboration created a need for documents, processes, standards, meetings and intermediate outputs to move information from one person to another.
Many familiar workflows therefore solve one underlying problem:
How can people with different capabilities and information accomplish something together?
Collaboration has a cost. An idea must be understood by someone else and executed by another person; each transfer can lose information. Time goes into communication, explanation, confirmation, waiting and rework.
AI is beginning to change this.
It lets one person cross more of the capability boundaries that used to separate roles. Someone previously skilled in one field can now use AI to do work that once needed several roles.
This does not mean everyone becomes an expert in every field. It means:
Fewer handoffs between people may be necessary to finish a task.
When handoffs shrink, the processes built around them also change.
Previously, we might spend a long time describing something: writing documents, preparing proposals, organizing requirements and passing them through layers before someone else builds it.
When the cost of building something to inspect becomes low enough, the working method can change completely.
Work may no longer require a complete description before execution. Instead, we can:
Build something concrete enough, then refine it through real feedback.
This is a substantial shift. When execution was expensive, we wanted to think as clearly as possible before acting. As execution becomes cheaper, we may increasingly act to help ourselves think more clearly.
That means more than greater efficiency. The once-clear boundary between thinking and execution starts to blur.
A common problem in enterprise AI adoption is assuming the existing organization and workflow are correct, then asking:
How can we add AI to every step?
If AI has changed the conditions that made those processes necessary, the real opportunity may be to redesign them.
Some steps can be combined. Some intermediate outputs can disappear. Boundaries between roles can soften. Work that previously required several people may become something one person owns directly through to the result.
Instead of repeatedly asking “How can AI make our current work faster?”, it may be more useful to ask:
“If we designed this work today, would we still do it the old way?”
AI's importance may go beyond accelerating existing work. It gives us a first opportunity to redesign ways of working we once took for granted.
Thanks for reading.