Notes from the work.
Short, honest pieces on what is really happening when a team ships with AI agents. No hype, no listicles. The same things we would say on a call.
The spec is the floor, not the senior engineer
A good spec lifts the floor under what an agent writes. It also makes the output look more trustworthy than it is, and that is where spec-driven teams ship their quiet bugs.
8 June 2026Confidence is earned, not granted
When writing code costs almost nothing, the scarce thing is being sure it does what you think. That confidence is a property of the system around the agent, not a favour you grant it because the tests went green.
5 June 2026The repo decides the speed, not the prompt
You do not make the agent faster with a better prompt. You make it faster with what is already in the repo before it writes a line.
5 June 2026Code is cheap. Judgment is the constraint.
Code is now the cheapest part of building software. Most teams are still organised as if it is the expensive part.
4 June 2026Comprehension debt
The code passes every test, ships clean, and nobody on the team can tell you how it works. That is a kind of debt most teams are not tracking.
3 June 2026The stall
Why throughput plateaus after the tools go on, and what the plateau is actually telling you.
20 May 2026Instrument before you blame
Before deciding AI is helping or hurting, measure what is actually happening in your loop.
12 May 2026Operator, not consultant
Why the useful help with AI coding tools comes from someone who still ships, not someone who only advises.
4 May 2026New pieces land every week or two.