Conway's Law in Reverse

The moment machine intelligence became scarce and metered, our industry began rebuilding the division of labour it thought it had transcended. Adam Smith's pin factory explains the new agent economy, and his forgotten fine print explains where it has to stop.

For thirty years, software's superpower was that the millionth copy cost nothing. In 2026, a frontier AI lab hung a sold-out sign on its door, and the economics of intelligence quietly changed shape.

Machines just made competence abundant, and everything we ship now arrives looking professional. Which means the most valuable thing left to do with our own hands may be the thing we do badly.

OpenAI's most productive engineering team has not written a line of code in five months. They have shipped about a million lines. They just published what they learned. Here is the executive translation.

Andrej Karpathy spent thirty mostly technical minutes on the Sequoia stage describing the executive shift of the decade. Five things from it survive translation.

AI got an order of magnitude faster. Your company got two or three times faster. The gap is not a mystery. It is a measurement problem.

For 25 years we built interfaces for humans. Now we are building them for agents. The web just flipped.

Most enterprise AI investments are compensating for limitations that are about to disappear. When they do, the scaffolding becomes the problem.

AI is confidently wrong some percentage of the time. The skill is not checking every line. It is knowing which lines to check.

Leeloo learned 5,000 years of human history in ten seconds. Your AI can't remember last Tuesday. So Milla Jovovich built an open-source fix.

Prompting did not die. It fragmented into four different skills. Most of us are still practicing only the first one.

The most productive users of AI are the ones who are best at rejecting its output.

The skill is not solving the problem. It is slicing it. And it applies to far more than code.

The most expensive sentence in 2026 is not 'build this.' It is 'build... something like this, you know what I mean.'

AI agents are fast, moderately reliable, and have no sense of direction. The skill they demand most is not technical. It is managerial.

AI changed what we can build. It also changed what we need to know. Here are the six skills that matter now.

AI is not competition. It is rocket fuel. The question is what we finally aim at.

The kitchen just opened. The question is not who gets cut. It is what we finally get to cook.

Claude Code skills are not plugins. They are not tools. They are knowledge, uploaded at the precise moment it's needed.

Wall Street thinks AI infrastructure is overbuilt and AI is powerful enough to destroy $800B in SaaS market cap. Both cannot be true. The arithmetic picks a winner.

Two weeks ago, 147,000 stars was a signal. Now the creator of OpenClaw is inside OpenAI. The agent race just changed shape.

AI didn't copy our org chart. It reinvented it. Three companies, working independently, arrived at the same management structure. That should change how you think about everything.

OpenClaw isn't the future. But the demand it exposed is.

For thirty years we've known that scripts fail under pressure and principles don't. We applied this to people. We forgot to apply it to machines. Anthropic just remembered.

The company most obsessed with what AI should not do has become the company doing the most interesting things with it.

The IDE was designed for humans who write code. If agents write the code, the IDE becomes what the light table became to Disney animators.

When you create a 'QA engineer' agent, you're not loading specialized capabilities. You're asking a generalist to forget everything except what matters. The sculptor knew this. Now the engineer must learn it.

Why your AI agents need a system, not a bigger brain

Why mastery is now the thing that makes you lost

Gemini 3 Flash is fast, cheap, and brilliant. It also doesn't know what it doesn't know.

What Dorothy Vaughan understood about the future that most people miss

Letting your AI agents think