The factory electrification analogy
This historical parallel illustrates why AI transformation takes decades, not months, and why early implementations underwhelm.
Published: July 8, 2025 — Original post
The history
Phase 1: Before electricity (centralized mechanical power) Factories ran on water or steam. A central engine turned a shaft; machines connected via belts. Machine placement was dictated by power requirements, not workflow efficiency.
Phase 2: Early electrification ("woo-hoo, wait...") First wave: swap steam engines for electric motors. Same layout, same processes. Minimal gain. New tech, old workflows.
Phase 3: Decentralized motors Electric motors attached to individual machines. Central shafts removed. But factories still kept the same layouts because everything was built around existing arrangements.
Phase 4: Real transformation Someone realized: "Now that machines are independent, we can move them." Arranged by task order, not power requirements. Birth of the assembly line. This is when productivity took off.
The application to AI
Today, AI is being plugged into existing workflows built around human constraints. This is Phase 2—new tech, old workflows. The real transformation comes when AI removes constraints and we redesign workflows around what AI makes possible. This takes time, happens incrementally, and is only obvious in hindsight.
The key insight
Boring infrastructure was essential throughout factory electrification. Wires, outlets, voltage regulators, circuit breakers—none of it was exciting, all of it was necessary. The same is true for AI: stable, reliable infrastructure (secure workspaces, identity management, governance frameworks) enables transformation more than cutting-edge AI capabilities.
Using this framework
When people feel underwhelmed by current AI impact, or when enterprises are chasing the next model release instead of building infrastructure: we're in Phase 2. The payoff comes in Phase 4, and it requires the boring infrastructure work that nobody wants to fund today.