TechLuddite

About

I'm a technical services director at a managed service provider. Nearly thirty years of that career has been hardware, networks, servers, and eventually the security decisions. Programming was the one thing I deliberately left to other people.

The AI shift made that position impractical. My assumption is that you can't make sound security and service decisions about tooling you've never used hands-on, and the use cases, supply chains, and data flows underneath managed services are all changing at once. These projects are how I'm closing that gap: built in public, streamed live, and honest about what went wrong along the way.

Several of these apps started as one-shot generations in an AI studio and were then audited line by line. Most shipped with a postmortem in the repo, because the bugs weren't random. They fell into recurring shapes, and the worst of them were apps reporting results they hadn't actually measured. The house rules here came directly from those postmortems.

None of this is an endorsement of how AI is currently being deployed. It's an attempt to understand it properly, which I believe is the only responsible position for someone making these calls.