AI agents ship many small changes a day. Running every test on every change is slow and costly. How we pick the tests that matter, and the safety nets.
Blog
From the workshop
How we build with AI, and the checks around it.
-
-
Inside our AI software factory
How we build software with AI agents: twelve stages, a merge gate, and the decisions that stay with people. We call it the AI Dark Factory.
-
Where a person still decides
AI agents write much of our code, and one of our products will soon draft orders with AI. Here is where a person still makes the call, and why.
-
Tests first, code second: keeping AI code honest
AI writes code that looks right. Here is how we check that it is: tests written first by someone else, real entry points, held-back data and traced bugs.
Have a question we haven’t written about?
Write or call, and we will answer it plainly.