What we built, what broke, and what we would do again.
5 posts
Testing AI generated code gets harder once an agent writes it fast. What a Harvard study shows about review, and how to test agent code before it merges.
How to scope a software project so a fixed price holds: map the documents, steps and exclusions first, and say where fixed price is the wrong tool.
A retry storm turns a dependency's bad hour into a longer outage. What GitHub's August report shows, and how a retry budget and one owner stop it.
What an AI native service business is, why investors want firms like yours, and how an owner can capture the margin first by starting with one document.
Human in the loop AI works best when the person confirms at the point where work is recorded, not when they watch the agent. Here is where to put the gate.
