The biggest commercial impact of AI might not be productivity; it might be workforce design.
Too many companies are still thinking about AI as a copilot, a tool to help people write faster, summarise faster, code faster, search faster and create faster. That is useful, but it is also too small.
Anthropic published a piece recently called "When AI builds itself", which is well worth a read.
The point that stood out to me was not just the productivity uplift. It's the shift in the type of work AI is starting to handle.
Anthropic describes Claude moving from executing tightly defined tasks, to handling more open-ended problems where humans supply the goal, but no longer need to supply the full method. That is a big shift. Humans still set the direction, but AI is increasingly helping identify the path, execute the work, evaluate the result and suggest the next best step.
That is much bigger than "write this email faster" or "summarise this document".
Because once AI can help decide the next step, not just complete the current task, the operating model starts to change.
In the same article, Anthropic describes a world where a 100-person company can increasingly do the work of a 1,000-person company, because each employee sits on top of a pyramid of agents.
I don't think it is as far away as it sounds.
Jack Dorsey is making a similar workforce transformation argument at Block. Block published a piece called "From Hierarchy to Intelligence", where they are using AI to challenge the traditional org structure itself.
This org hierarchy has existed because humans were the coordination mechanism. Managers moved information, maintained context, aligned priorities and coordinated work between teams.
But if AI can take on more of that coordination work, then the traditional management layer starts to look very different. Not gone, just transformed into something different. That is a much bigger idea than "everyone gets a copilot".
The question is not "How do we make each person 10% more productive?" It's "How would we redesign the company if every capable person had a digital team underneath them?"
That question changes hiring plans, management structures, job design, operating rhythms, decision rights and workflow ownership. The shape of the company itself.
This is where I think many companies will get AI wrong. They will buy AI tools and leave the operating model untouched. Same org chart, same approvals, same meetings, same workflows, same job descriptions, same handoffs. Just with a few AI licences layered on top.
That will create some benefit, but it misses the bigger opportunity.
The companies that win with AI will not just be the ones that adopt the tools. They will be the ones that redesign the organisation around the new leverage.