Latest

Your Codebase Was Always This Bad
AI didn't create the fragility. It walked into it.
A team turned on agent mode. Three weeks later, incidents spiked. They were ready to blame the tool. Then their senior traced the bug back to 2019.
Latest

AI didn't create the fragility. It walked into it.
A team turned on agent mode. Three weeks later, incidents spiked. They were ready to blame the tool. Then their senior traced the bug back to 2019.
Previous
The tool isn't the problem. The abdication is.
For 23 years I've watched developers copy code they didn't understand. First from Stack Overflow, now from AI. The pattern never changed. We just changed who we blame.
Why documentation doesn't matter if your AI can't read it
I was watching the head of architecture demo his team's AI setup when I realized we'd been solving the wrong problem. The style guide wasn't missing. It was invisible.
On blame culture, accountability, and why the author never mattered
Everyone has worked with a Steve. Now AI is Steve. The lesson isn't about the tool ... it's about what we tolerate.
The difference isn't the tool. It's what brain you're pulling from.
For the first month I treated Copilot like a junior engineer with ambition and no guardrails. It created variables named R and T, broke tests it didn't care about, and left me cleaning up for two hours.
The skill that made you valuable is becoming table stakes. The new value is orchestration.
A senior engineer posted on Reddit recently about how it's almost like he's in grief. More than a decade of craft. The things he loved. And now it's all changing.
It started with you ... and it'll keep going until you own it
AI is amplifying the technical debt you were already permitting. The slop in your codebase isn't a tool problem ... it's a permission problem.
The conversation you stop having
We stopped talking about DevOps. Now everything uses it. AI is slowly crossing that same threshold.
The vending machine problem
We treat AI like a vending machine. Prompt in, code out. When it fails, we prompt harder. That's nonsensical.
Why large PRs get rubber-stamped while small ones get torn apart
I built a dashboard to track PR review times. What I found broke my assumptions about code review culture. Large PRs were getting approved almost as fast as small ones ... and that says everything about where we're actually paying attention.