There is a straight line from the first Lego brick to a codebase.

Not a neat line, obviously. More like a child’s line. Slightly crooked. Unreasonably ambitious. Heading directly toward something structurally unsound and wonderful.

I know this because I’ve been building like that for as long as I can remember.

As a kid, I had thousands of Lego pieces. They came in through the usual channels: birthdays, Easter, Christmas, the occasional parental lapse in judgment. Naturally, every new set was built once according to instruction. Because of course it was. You had to understand the official version first. Respect the system. Learn the intended shape of the thing.

And then, just as naturally, it had to be destroyed.

Not out of disrespect, but out of curiosity. Out of creative necessity. The pristine police station, spaceship, pirate fortress, whatever it was, had served its purpose. It had delivered its parts into the republic. The bricks were no longer loyal to the box art. They had been assimilated into larger, stranger, more important plans being run by a child brain with absolutely no regard for scope management.

That was one of the first true flow states of my life.

Hours disappeared. The world fell away. There was only structure, tension, possibility. A pile of pieces and the intoxicating sense that reality was, at least in some small radius around me, negotiable.

That instinct went beyond Lego.

When I was three, I apparently deconstructed a vacuum cleaner. “Deconstructed” is a generous word for what was, from the vacuum’s perspective, a catastrophic event. My father, an electrician and engineer, had to put it back together. I’m sure this was inconvenient for him. But in retrospect I like to think he recognized the species of problem. Some children play with toys. Some want to know what the toy is hiding.

Or, more precisely: how the machine works, where the seams are, and whether it could be made to do something else.

That urge led me, among many other things, toward engineering. Which felt less like a career choice than a formalization of a preexisting condition.

Engineering, at its best, is organized agency.

It is the refusal to stand in front of a system and treat it as fixed just because somebody else assembled it first. It is the belief that environments can be understood, modified, redesigned. That constraints are real, but not sacred. That the world is, in fact, made of parts. And if it is made of parts, then it can be learned. If it can be learned, it can be shaped. And if it can be shaped, then maybe you are not merely living inside fate. Maybe, to some degree, you get to build it.

That idea never really left me. Only the medium changed.

I learned actual coding with QBasic, then Pascal, then C in the early 90s. In the early 2000s I moved into web development. From there into design, creative direction, strategy. On paper, that can look like a sequence of pivots. From the inside, it feels much simpler: I’ve always been building.

Sometimes with bricks. Sometimes with code. Sometimes with language, systems, teams, narratives, interfaces, brands, and operating models. But always with the same basic instinct: take the thing apart, understand the pieces, imagine a better architecture, build again.

Which is why the current AI moment feels less alien to me than it seems to feel to some people.

A lot of the current discourse around coding agents carries either panic or cosplay. Either software is over, or everyone is suddenly a ten-person product team with a prompt window and a dream. Both are a little silly. The more interesting truth is simpler: the cost of building has collapsed again, and that changes who gets to play.

That matters.

Because agents, at their best, do not eliminate the need for human taste, judgment, or ambition. They amplify them. They give people with agency more surface area. More reach. More iterations. More ways to move from idea to artifact without needing an entire institutional machine just to test whether the idea has legs.

In other words: more people get to play with Legos again.

Just with different bricks.

This weekend, while working on our latest release, I had that thought more than once. There I was, once again in a room, happily immersed for hours, arranging parts into systems: AI agents, code fragments, Python classes, components, prompts, event flows, schemas, states. Same feeling. Same quiet electricity. Same ridiculous optimism that if I keep moving the pieces around long enough, something elegant might emerge.

And maybe that is one of the most beautiful things about this moment.

For all the noise around AI, one of its gifts is that it returns building to people who were previously kept at the edge of the workshop. Not everyone will use that gift well. Many will build nonsense. Some are building haunted demoware held together by vibes and unsecured API keys. That, too, is part of the tradition.

But some people are using these new tools the way children use bricks: seriously, playfully, obsessively, with taste and nerve and unreasonable hope. They will build because they can. Then build because they must. Then wake up one day and realize that the real joy never was the finished object. It was the agency.

The chance to shape your environment a little more deliberately.

The chance to shape yourself with it.

We never really stop being the child on the floor surrounded by parts. The lucky ones just find better workshops.

And better toys.

—

Jo Wedenigg is the founder of Apes on fire, where he builds human x AI collaboration systems for creative, strategic, and transformation work. He is the creator of Ape Space and focuses on turning AI into a partner for advanced thinking.

—