Case Studies
Systems I built and ran. What was broken, what I built, and what changed.
Writer Notes That Leaked Onto Live Pages
Notes written for a person started showing up on a live blog post once a script took over publishing. I added a check that strips the kinds of note I already know about and stops the publish when it finds a note it doesn't recognise.
71 Idea Files Made Searchable, Zero Installs
My idea folders held 71 files that no search could read. In one overnight run I turned 57 of them into plain text notes, with nothing new installed and nothing paid.
Catching Status Notes That Had Gone Stale
My notes and jobs kept claiming things that were no longer true. I built a nightly check that finds where two files disagree or a promise has lapsed, and learned that most of my old checks had never really been checking anything.
Rebuilding My Firm's Website With an AI Pair
I rebuilt my consulting firm's website in four days with an AI coding partner. The new look mattered less than two other things: the claims I removed, and the faults I only found by looking at the live pages.
Putting Every Content Rule Behind a Check
My content system had written rules that nothing enforced. In one day I gave every one of them a weekly check, and the checks found problems that the rules on their own never had.
112 Scheduled Jobs, and How Many Really Ran
I had 112 scheduled jobs quietly running my operating system. When I checked, only 66 showed any proof of ever running. So I built a record that makes every job show its own work.
The Quality Gate I Said Never Fired
My AI content system had two automatic quality checks. I audited them, claimed one had never fired, and was wrong. The lesson I took away was about how I had checked, more than about what I found.
Taking Website Publishing Off a Person
Website posts were copied onto my two sites by hand, and the hand step kept failing. I moved the job to a daily script, with checks that stop a bad post before it goes live.