I spent five years designing nuclear reactors in the US Navy before I became a product design consultant. In the Navy, "the plant doesn't care about your intentions" isn't a saying — it's an operating principle. Systems either work or they don't. Procedures either cover the failure mode or they don't. Assumptions either hold or they don't, and when they don't, you find out fast.
That background shapes how I build with AI.
Most of what gets published about AI tools is about the upside. What prompts to use, what tasks to automate, how to 10x your productivity. I'm more interested in the failure modes — why hooks silently fail, how context windows drain, what happens when an agent operates on stale assumptions, and how to design for the cases where the model confidently gives you a wrong answer.
Arcanium is the product line I built from those failure modes.
The products are practical reference materials for people building real systems with Claude and Claude Code. Not tutorials. Not beginner guides. Field guides — the kind of reference you keep open while you work, built around the distinctions that actually matter.
The Weekly Forge ships every week. One idea. Field notes from whatever I'm building. No hype. Subscribe free on Substack.
I'm located deep in the mountains of northwest Montana. I do AI coaching and consulting for enterprise teams through IBM. I hold a master's degree in nuclear engineering from Missouri S&T and qualified at the Bettis Atomic Power Laboratory.
Questions or feedback: help@arcanium.us
Arcanium — Forge Smarter Systems