Back to beliefsThesis

Factory Thinking

Invest in capability infrastructure, not features


Factory Thinking

Build the factory, not just the product.

When Volkswagen builds a new vehicle line, they don't just design the car. They invest in the factory: welding robots, assembly line ergonomics, quality stations, andon cords for worker autonomy. The goal isn't "build this car" but "build the capability to build great cars continuously."

Digital work requires the same thinking. Instead of buying features, invest in infrastructure that compounds:

Industrial Factory Digital Factory
Welding robots AI pipelines, custom software
Assembly line ergonomics Workflow design, tool UX
Andon cords Human-in-the-loop controls
QA stations Taste validation, data hygiene
Worker training "We can just build things" culture

The Management Question

Not "what AI features should we buy?" but:

"How do we invest in tools, techniques, and talent to build the best [X] possible?"

What Factory Investment Looks Like

  • Data foundations: Clean, accessible, well-structured data
  • Intelligence infrastructure: Pipelines that deploy judgment at scale
  • Communication bandwidth: Tools that increase information flow between people
  • Decision velocity: Systems that enable faster, data-driven choices
  • Build culture: The capability and permission to "just build things"

Implication

Features depreciate. Factories compound.

Companies that buy AI features will need to keep buying. Companies that build AI factories will produce differentiated experiences continuously.

Contrarian To

"Let's start with a chatbot pilot and see how it goes."

Pilots don't build capability. Factories do. The investment should be in infrastructure that makes the next thing cheaper, not in a feature that stands alone.


Related