North star: “The problem is real and worth money.”
The idea itself, tested by hand.
What you hold is the tool of discovery — every loop's input, never its output. The name tells you what you're hunting; your hands tell you where you're standing.
“The problem is real and worth money.”
None — only conviction. Nothing is validated yet.
None — manual delivery. The founder is the product, and that is the discipline. The cheapest evidence is bought with zero code, and it is infinitely reversible.
Earns: Problem validated. All five exit bars →
Which is why it is premature before you know the decision is right — and the most expensive possible medium for writing down a hypothesis.
Most products have none worth the name — their risks are all on the customer side. The few that do have exactly one killer assumption, which can almost always be tested alone, cheaply, before anything is built.
A conversation costs an hour, a manual delivery costs a week, a prototype costs a month, a build costs your runway. The craft of this loop is buying the same learning from the cheapest counter.
Your first prototype-of-thought is free: an AI assistant will pressure-test the idea, draft the pitch and simulate the skeptic before breakfast. The ground itself also moves now, so this loop gains one modern decision — the rising-tide test — before any build conversation.
The assistant cannot deliver your outcome to a paying customer by hand — and that, not the drafting, is this loop's evidence.
AI as your thinking partner and skeptic.
The release-note news cycle — the rising-tide decision replaces it with three numbers.
Guide in progress — in the book, and in the newsletter when it ships.
Guide in progress — in the book, and in the newsletter when it ships.
Decisions explicitly deferred while you're in this loop. Not wrong questions — early ones. Written down, and set down.
Free, ungated plain-text files you hand to Claude or ChatGPT. All skills →