There's a famous story about Elon Musk firing 75% of Twitter's engineers overnight. Everyone called it bold. Efficient. The future of work.

Six months later he was quietly calling them back.

Because it turns out those people weren't just writing code — they were the living documentation of a system too complex to exist on paper. When they left, the knowledge left with them.

I think about this every time a client asks me to "fully transform their operations with AI."

My answer is always the same: no.

The fence you shouldn't touch

There's a principle called Chesterton's Fence. If you come across a fence in the middle of a field and don't understand why it's there — don't remove it. First figure out why someone built it.

Most legacy systems in companies are Chesterton's Fences. They look inefficient. They look old. But they exist because someone, at some point, got burned without them.

My job isn't to tear down fences. It's to quietly put a small gate in one — and see if anyone uses it.

What "small" actually means

Small doesn't mean unambitious. It means surgical.

One process, one team, one pain point. We automate the thing that costs someone 2 hours every Tuesday. We add an AI layer to the document that nobody likes reading. We build one assistant that answers the question everyone keeps asking.

Then we wait. We watch. We check nothing broke. Then we do the next one.

Why this works

Because organizations are living systems. They have immune responses. Push too hard, too fast — and they reject the change. Not because people are stupid or resistant. Because they're protecting something that works.

Small deployments don't trigger the immune response. They slip in. They prove themselves. And then — slowly — they become normal.

That's how AI should enter a company. Not through the front door with a press release. Through the side door, solving a Tuesday problem.

— Łukasz, founder of KTR-Technology