May 2026 — Denver, CO
I'm working on a framework I'm calling Pilot. It's a first-principles architecture for building a genuinely autonomous intelligence. The foundation is a model of reality as an interaction space rather than a spacetime one. The idea, grounded in current theories of physics, is that what we experience as spacetime isn't fundamental. It's emergent. Beneath it lies something closer to a hypergraph: a web of quantum-scale interaction events, nodes that carry no position in space or moment in time on their own. From this perspective, gravity isn't a force applied across pre-existing space. It's the shape of the field lines that emerge as interactions aggregate and curve the structure of the graph itself.
The goal is to use this not as metaphor but as a structural blueprint. If intelligence in nature, ours included, already operates according to these same underlying constraints, then the most efficient artificial intelligence isn't one that mimics biology or optimizes against a narrow benchmark. It's one built to close the same loops that physical reality imposes on any persisting system: minimize dissipation, reduce friction in the feedback cycle, and traverse its goal surface as a natural consequence of its own structure. Same laws, different substrate. Does that sound insane? Well it is. That's Pilot.
Otherwise, I spend my time running the tech of Snug, reading and meditating on physics, preparing for a child, trying to make my wife laugh, and thinking about where the real leverage is in the next few years of software.