Built with Claude
Every project on this site was built in collaboration with AI. Here's what I've learned.
The Confession
I'm a litigation attorney. I've been one for twenty years. I am not a programmer. I have never taken a computer science course. I cannot, off the top of my head, tell you the difference between a framework and a library (though Claude has explained it to me several times).
And yet, in the past two years, I've built:
- KubeTrakA family chore tracker
- Rad LibsA retro word game with 108 stories
- The LadderA legal trivia game with 1,000+ questions
- White SteelA 3D Catskills resort tycoon game
- Context CabinetA macOS developer tool for Claude Code
- FableforkA choose-your-own-adventure engine
- BDK ComparoA macOS folder comparison utility
- This websiteIncluding the Geocities easter egg
Every one of them was built with Claude. Not as an autocomplete, not as a search engine — as a collaborator, in the full sense of the word.
What I've Learned
Here's what two years of this has taught me: the bottleneck is almost never the AI. It's whether the human can say what they actually want, look at what they get back, and explain specifically what's wrong with it. Most people aren't practiced at that. Litigators, it turns out, are freakishly good at it.
Twenty years of drafting arguments, cross-examining witnesses, and negotiating settlements gave me a very particular skill set: I know how to be specific about what I want, skeptical about what I receive, and relentless about closing the gap between the two.
When I tell Claude "the rotation logic in KubeTrak should be fair even when someone misses a day," I'm running the same process I use when I tell opposing counsel "this settlement term is ambiguous and here are four scenarios it doesn't cover." Different room, same work.
Why This Matters
If a litigation partner with no coding background can build these things in his spare time, imagine what a dedicated team could build for the legal profession — with someone who understands both sides of the equation.
I beta-tested AI legal tools back in 2023, when the models weren't ready yet. They are now. The legal profession is at an inflection point, and I'm convinced the tools that define the next era will be built by teams that include people who've actually practiced law — not just people who've studied it from the outside.
These projects are how I know. Not from a white paper. From doing it.