the new workshops

Edited with Sofia

Notes on Context, No. IV

In Renaissance Florence, a young apprentice might grind colors, cast bronze, study the human body, build stage machines and paint church art, sometimes all in one week. Nobody called it cross-disciplinary. That was just the job. Leonardo grew up in a room like that. It's why he believed a painter should understand how light hits the eye, how water flows, and how a muscle moves a bone.

A few decades later, Michelangelo spent four years on a scaffold he helped design, painting the Sistine Chapel ceiling in wet plaster, where every mistake set within hours. The ceiling tells a story about God and humanity, but it's also chemistry, carpentry and stamina. The idea and the craft were one thing. That's why it still stops people in their tracks 500 years later.

For the last twenty years we've pulled those things apart. Software went on one side, hardware on the other, and operations got hidden in a spreadsheet. The ideal became the pure software company: light, fast to scale, far from the messy physical world.

That era is ending. AI is making software almost free to build, and when anyone can build it, software alone stops protecting you. Code can be copied, and an app can be cloned in a day.

What can't be copied is the loop. The product teaches the factory. The device teaches the model. The day-to-day operation teaches the product what to become next. Each part makes the others better, and the whole thing gets harder to copy every year it runs.

So the best businesses will look like workshops again, with software, hardware and operations built together and grown together. Not a software company that happens to sell things, and not a factory with an app bolted on. In these companies, the business is the software. Every machine, shift and customer feeds back into the code, so the software is shaped by the work and not the other way around.

The workshops also get new apprentices. In Florence, a master had a room full of assistants who mixed paint, prepared walls and filled in backgrounds, while he set the vision and made the final strokes. The new workshop has robots on the line and AI agents in the back office. They're tireless and fast, but they're apprentices. They know how to do a task. They don't know this factory, this material or this customer until the work teaches them. The master's job hasn't changed: decide what's worth making, and pass on what only experience knows.

That's where the fourth moat lives. Proprietary data, network effects and owning something physical all matter. But underneath them is context: knowing how a specific world really works. Which supplier cuts corners. Which material breaks down by the sea. Why customers in one place act nothing like customers somewhere else.

You can't download context. You earn it by doing the work long enough for the work to change you. AI learns from what people wrote down. Context is mostly what nobody wrote down.

The Renaissance also shows what happens without it. Leonardo spent years on a giant bronze horse for the Duke of Milan. The full-size clay model was finished. Then war came, the bronze was melted into cannons, and the horse was never made. Great ideas that never connect to the real world end up as clay.

And there's a lesson about patrons. The Medici family and Pope Julius II didn't just pay for art. They had taste, asked for the impossible, and protected the work for generations. Today the patrons are founders and builders, deciding what's worth making now that making is cheap.

Taste decides what deserves to exist. Context decides whether it lasts.

Small nations already know this. They can't win on size, so they win on depth: knowing their land, their people and their limits better than anyone else. In the age of AI, that may not be a weakness. It may be the whole game.

Build workshops, not apps.