Principles · v1
The 12 Principles
Every idea in this studio ties back to at least one of these. They were not written for this page, instead they were observed as principles underlying the projects, prompts, and systems I've been building for years.
"When AI can do or be everything, we can drift toward the inclination to passively accept changes. We must fight to retain our shape, tastes, beliefs — that is the most valuable contribution a human can make in this new era."
Attention is the constraint
Everything else scales — tokens, agents, output. Users' attention doesn't. It correlates with time but it is not the same thing, and every system here is designed around protecting it: the studio's real unit economics are denominated in attention, not hours.
Friction is the signal
Wherever humans and agents grind against each other — attention, trust, evaluation, coordination — that is the signal for what is needed but doesn't yet exist. I seek that friction and let it determine what to build to unlock new ways of working.
Whitespace shapes memory
We may think we want our agents to remember everything, but upon further reflection that leads to a host of problems. For humans and agents alike, identity is shaped less by what's stored than by the 99% of information deliberately ignored, forgotten, or released.
Cost reduction is not value creation
Sometimes value doesn't move when a cost disappears — it gets destroyed. A system that only measures the bill decreasing is missing the other half of the equation. I build systems that try to raise the ceiling rather than race to the bottom.
Agentic accounting is hard, but needed
It's easy to see the AWS or Anthropic bill drop; but pricing a quality of life improvement is not. Much of what this studio produces lives on the challenging side of that ledger. Quantifying value will help decide what to build in a world of infinite possibility.
Exceptional taste lives at the fringe
AI can do "good." But what "great" means is a load-bearing human call, especially when inventing products and experiences that were never possible until the latest surge of AI capabilities made them feasible. Fight the good fight against slop!
Fairness is the foundation
This is a core value. It's everywhere in systems I design. People should be rewarded for their valuable contributions. Treating this as a first-class requirement at the onset avoids entertaining otherwise enticing tradeoffs that might not put the user's interests first.
Attribution and causality
Who caused what is a pervasive question in law, accounting, systems engineering, economics. It's my admitted obsession, and it's the machinery that keeps every other principle honest: without attribution, talking about fairness is an applause light.
Incentives are the gravity of systems
Markets are chaotic, individuals less so, machines least of all. Each acts in accordance with its incentives. We design around this: we think deeply while setting them, then simulate to play out the counterfactuals before committing to a course of action.
Games are serious business
Business, and even life itself, is just one long game: defining objectives and seeking them given constrained resources under adversarial pressure. My love for well-designed games followed me into adulthood, so often primitives that seem playful are doing real work.
Token arbitrage is the frame
Boiled down, all AI native businesses are systems that convert tokens into value at a (ideally) favorable exchange rate. This reductive frame demands un-reductive machinery: value attribution, verification, controls. This is what sets how this studio runs apart.
Agents serve as a cocoon
This is the best analogy I've come up with. AI wraps and extends us, but we must remain accountable and retain our shape within the dyad working alongside our agent. Humans should become more capable both within and without the systems we design.