Not Every Workflow Needs AI
The useful question is not where AI can help. It is where AI creates enough leverage to earn a durable place in the workflow.
Essays from systems I helped build, the teams around them, and the human judgment that becomes more important as implementation gets easier.
The useful question is not where AI can help. It is where AI creates enough leverage to earn a durable place in the workflow.
AI has made software feel playful and alive again. That same energy can quietly turn into burnout when engineers never stop using the muscles that make the moment exciting.
The old picture of engineering culture was a quiet room with headphones on and nobody talking. Ours looked nothing like that.
We started using AI in engineering for the obvious reason. The interesting part was discovering that the playbook ported directly to operations.
A year ago, I would have described an engineer as someone who writes code. I would not say that now.
A year ago, the most ambitious thing AI did for our engineers was tab completion. A year later, it ran overnight bug triage and prepared pull requests.
A newer version of the site is ready.