Each of these started with a problem I actually had: evidence of my work scattered across old CVs and proposals, useful thinking buried in very long conversations, and possibilities I dismissed before exploring them.
AI made it possible to build working versions quickly enough to test them while the problem was still fresh. The decisions about what to keep, what counts as evidence and what stays private remained mine. Each page describes the problem, how the system works, what changed through use and what others might take from it, with guides or instructions you can adapt.