Everything below comes from `state.json`, `repair-queue.json`, and `adoptability.json` - the Director's own JSON output. We pull it raw and post it. If you want the source data, those files are publicly readable at /factory/state.json etc.
Every product is rated by a quality composite. The tier reflects how much hand-polish + critic + foreman + skeptic + audio + screenshot work it has accumulated.
Showcase tier is what we'd recommend a first-time buyer look at first. Sketch tier is a partial product the Director hasn't finished cleaning up yet.
Adoptability is a 10-axis heuristic separate from raw idea quality. Pain intensity, buyer clarity, distribution difficulty, financial upside, and 6 other axes. Higher score means easier to ship to first revenue if you have the right audience.
Every catalog idea ships with these artifacts. The numbers below are what's accumulated across the whole catalog.
Cost to date is Director-tick LLM spend + scorer + foreman + audio rendering + miner. It does NOT include compute (we run on a $40/month VPS) or my time (countless hours). Just the variable LLM/audio cost.
Director ticks means decisions made by the autonomous loop. Each tick is one of 14 typed actions: build a new product, polish an existing artifact, run a critic round, generate a brand brief, etc.
Buyer-path complete is the percentage of catalog products that have all 13 artifacts shipped (landing, pricing, FAQ, financial, drip, outreach, risk memo, dossier teaser, adopt CTA, etc.). 96% means the median buyer experience is consistent.
What's NOT in these numbers: live customers, paid revenue, validated ICPs. The catalog is starter kits. Read the honest page for the version of this we tell ourselves.