What we've learned, written down.
Short, honest notes from the work — AI, the field, and where the two meet. No hype, no case-study theatre. Just the things we wish more organizations knew before they start.
All field notes
Why most AI pilots die at the deployment site.
The demo works. The pilot gets praised. Then it reaches a real site — where the power flickers, the signal drops, and the phones are five years old — and quietly stops being used. Here's what actually kills them, and how to build so it doesn't happen to yours.
Read the note →Sovereign AI, in plain terms.
“The data can't leave the country.” What that actually means for your architecture — on-device, on-site, or in-country cloud — and how to pick without overpaying.
AI is new. Knowing what works in the field isn't.
Everyone's AI is only a few years old. What separates useful work from hype is the years of judgement about what survives contact with reality — and how to borrow it.
What “offline AI” actually costs — and when it's worth it.
Running AI where there's no internet is a real capability with a real price. A plain look at what drives the cost, and the honest test for whether your mission needs it yet.
Prefer to just talk it through?
A short call with people who've done this in the field beats any blog post. If any of this maps to your work, let's compare notes.