How we think.
Five rules. Every project we open is held to all of them.
Pick problems we can point at on the chain.
If we cannot say “here is the failure mode, on Solana, today,” we do not start. A real problem is the only honest brief.
Sit with the problem longer than is reasonable.
The pull is to start building on day one. We don’t. We stay with the problem until its real shape shows. That’s where the leverage is, not in the typing.
Carry the idea all the way to deployment.
Research, infrastructure, product, operations: one team, no handoff. The judgment calls hide where two layers meet, and you can’t make them from one floor up.
Reliability is the brief, not the polish.
The chain runs continuously. Anything we put in the path behaves the same way: measured, retryable, explainable after the fact.
Usage is the only signal we trust.
We plan against evidence, not enthusiasm. Paying users keep a project alive; their absence ends it. Roadmaps built on hope are how labs quietly fail.
Useful, or it doesn't ship.
Everything else is a draft. We prove the behaviour against real traffic, keep what holds, and delete the rest without ceremony.
Replay.
The five rules, made concrete: a failure you can point at on the chain, time on the problem before the code, one team across research, infrastructure, and product, reliability as the brief, and paying teams on the other side of it.
see replay