A personal field guide, built in public
Honestly, this is just my notes. The AI tooling space moves fast enough that by the time something is written up properly somewhere official, it has already shifted. So I started keeping a reference for myself — what things are, how they fit together, where the hype is and where the actual utility is. At some point that turned into a small site.
It is not a finished product. It is scratchpaper that got organized. Some pages are thorough, some are thin, and the whole thing reflects where my attention happened to land. If you find it useful, that's a bonus. If something is wrong or out of date, it probably is — this is a living document, not a textbook.
The more interesting part, honestly, is how the site was built. I used AI tools and agents as a working development system: issues to define the contract, investigative passes to make the work concrete, implementation passes to make changes, then review, cleanup, and voice correction to keep the output grounded. A lot of the commit history is that loop in public: file the issue, investigate what is actually true, implement, audit, refine, repeat.
That is what the site demonstrates beyond the content itself. I think the useful question with AI-assisted engineering is not whether an agent can produce plausible output in one shot. It is whether you can design constraints, checkpoints, and feedback loops that make the result reviewable, correctable, and worth trusting.