Key takeaways
- Factories that swapped steam for electric motors got almost nothing at first.
- Productivity rose only when people redesigned the factory around what electricity made possible.
- AI works the same way: the payoff needs a redesign. This time, the people doing the work can do the redesign themselves.
One big motor, same old layout
When factories first replaced the steam engine with electricity, many simply swapped one big steam engine for one big electric motor. They kept the same layout, the same line shafts running the length of the building, and the same schedules built around a single source of power. Productivity barely moved for decades.
The gains came from the redesign
The gains arrived when people put a small motor on every machine and rebuilt the floor around the flow of work instead of around the power source. Single-story layouts, machines arranged in the order work moved, new jobs and new schedules. The economic historian Paul David documented this lag in his classic study, “The Dynamo and the Computer.” The technology was never the constraint. The redesign was.
Why it took so long
The redesign took decades for reasons that sound familiar. The existing layout had been built for steam, and it worked well enough. The people who understood the work were not the people who could rebuild the factory. And the incentives rewarded keeping the line running, not rethinking it. Every one of those reasons has a modern equivalent.
AI follows the same pattern
AI is following the same path. Most companies have added AI to their existing processes the way factories added one big motor: same workflow, same handoffs, a new tool on the side. The result: 88% of companies use AI, but only about 6% get significant profit from it.
Redesign for AI does not mean pasting AI onto the old process. It means asking what the process would look like if drafting, summarizing and assembling information were nearly free, then rebuilding the work around that. A proposal process that starts from a blank page becomes one that starts from a grounded first draft; a weekly decision pack that took a morning to assemble arrives reconciled, with the risks already on the table.
Who gets to redesign this time
Here is the difference. Electrification needed engineers and capital to rebuild factories. AI lets the person who understands the work redesign it directly, if they know how. That shift in who can build is the most important change in this story, and it is the subject of the next two articles.
Sources and notes
- Paul A. David, “The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox,” American Economic Review, 1990.
- McKinsey & Company, “The state of AI in 2025: Agents, innovation, and transformation,” November 2025 (1,993 participants, June 25 to July 29, 2025).
- Robert M. Solow, “We’d Better Watch Out,” The New York Times Book Review, July 12, 1987.
Founder & CEO of Lucy Labs. Twenty years running enterprise transformation by hand in big tech and Global 2000 companies, now building it into software.
