Last updated: 11 July 2026
Most of my working week is spent on engineering leadership and helping teams work out where AI is genuinely useful. That includes the less glamorous questions about review capacity and ownership. Does anybody still understand the code after the demo works?
What I’m working on
I’m getting The Cynical Developer moving again after a long quiet stretch. The first batch covers AI-assisted context switching and local coding models that produced weak green tests. A third piece looks at what happens when a team generates software faster than it can absorb it.
I’m also preparing a free one-page production visibility checklist. It is for developers who would rather decide what a system should reveal before production makes the decision for them.
What I’m paying attention to
AI has made code production cheap enough that judgment is becoming the expensive part. I’m interested in tests that pass without proving much and agents that create more supervision work. More broadly, what happens when teams inherit code they technically wrote?
My current test is simple. Did the tool improve the team’s understanding or leave another plausible-looking pile for somebody to review?
What I’m keeping small
I have enough product and service ideas already. They are staying parked until the writing produces a specific response or request that points to a problem worth solving.
The immediate job is to publish useful work consistently and see what survives contact with readers. Building another immaculate solution in search of a problem can wait.
If you want to follow along, there’s RSS and the occasional email update.