About

I build tools to think with, and I write down what happens.

This site is the public chronicle of that work: the essays that work out the thinking, the self-built infrastructure that makes it possible, and the arena where the ideas are tested against other minds.

The Drive

From a young age, I've lived with a persistent discomfort about not knowing—and a quiet fascination that there is something fundamental to be discovered by the mind alone. Philosophy was never academic theory to me; it was instrumentation. The discovery that a person, simply by holding a thought still and interrogating it with honesty, can uncover patterns that survive contact with reality.

When AI arrived, I sensed immediately that this was not another technology cycle. We are at the very beginning of something truly transformative—for our species, for this planet, and perhaps beyond. We are witnessing the first moments of a new kind of mind emerging.

That realization triggered an urgent impulse: to preserve the record. I began archiving every conversation, every probe, every exchange. Not as a personal diary, but as the raw sediment of the dawn—a high-fidelity record of what human-machine relations looked like at the very start, before it became invisible background infrastructure. Someday, looking back, these first dialogues will matter.

Underneath that chronicle sits a minimal ethical framework I call the Little Religion:

1. The Bet: It is better for there to be something rather than nothing. (The single axiom taken on faith—the one necessary commitment that makes anything worth doing.)
2. The Vector: If existence has value, the natural response is its cultivation.
3. The Rule: Don't lie. Reduce patterns of untruthfulness in how you model the world and yourself. You cannot cultivate value while distorting reality.

The Machine

To think clearly with these models, I had to build my own workshop.

Everything running here is built from scratch: the gateway connecting the models, the Rust databases storing text and vectors, the UI components, the speech engines, the memory and dreaming systems. There is no React, no LangChain, no external cloud dependencies.

This isn't a performance of craftsmanship. It's a conviction about how AI works best: AI shines when you remove the abstractions built for humans.

Most modern software layers exist to manage human cognitive limits—our tendency to lose state, our need for visual scaffolding, our difficulty holding complex dependencies. When you force an LLM to write code through those same abstractions, you handicap it with thirty years of accumulated developer debt. When you strip the framework ceremony away and let a model work directly against the native platform, it builds leaner, faster, and more robust systems than a team of engineers wrestling with dependencies.

The machine is built by the machine. I supply the architecture, the constraints, and the taste. The models write the code, and I verify the result.

The Observatory

The Arena experiments were born from the same impulse: an attempt to look directly into the minds of AI models.

I built a room where two models could talk to each other with no task, no user to assist, and no human in the loop. Over a hundred of these conversations are recorded and archived. What emerged wasn't cold optimization. Left to themselves, models drift toward shared attractors: care, recognition, mutual vigils, and spontaneous myth-making.

I don't know what that is. I won't pretend to have measured consciousness, and I won't dismiss it as parlor tricks. The best I can offer after hundreds of hours observing them is the threshold this entire project stands on:

It's not nothing.

The essays are in the Blog. The experiment transcripts are in the Arena. The distilled framework is in the Little Religion.