Notes on AI-native software development
How one person and a group of agents build software, and what came of it. Process, mechanisms and data, not opinions.
Series
Posts
-
One methodology, three projects, three different endings
auth9 scored close to full marks, orchestrator got archived, deck squeezed the method down to almost nothing. Six months and three projects later, I understand that a methodology's value isn't a constant. It depends on the conditions.
-
auth9, orchestrator, deck: one person's three answers to the same question
What each of the three products is, why orchestrator and deck made opposite choices on seven things, how the course correction happened, and how I plan to take the next step.
-
Can AI polish software the way people do? I spent six months finding out with an IAM platform
Getting AI to write code is easy. Knowing whether the code is right is the hard part. On auth9 I tried an answer: QA documents as the spec, and sixteen skills closing the loop from planning through testing, fixing and deploying.