What the dots mean.
Every job in the catalog carries a complexity tier: one, two, or three dots. They tell you how much judgment the machine can do on its own, and how much refinement you should expect to do before you ship the result.
Deterministic
Binary success criteria. The machine either reads your receipt correctly or it doesn't. Low variance, near-zero judgment required from the buyer.
Near-perfect. There is a right answer and the machine either finds it or it doesn't. If it misses, you'll know immediately and we'll refund.
Structured-generative
Standard form, variable content. The structure is well known (blog post, summary, changelog). Quality is reliable; flavor may need light editing.
Reliable structure. The shape is known (intro, sections, summary). Content will be accurate to your input. Plan to light-edit for voice and tone before shipping.
Generative-nuanced
First-draft structural depth. The machine gives you scaffolding and 80% of the prose. Expect to refine voice, tighten claims, and verify specifics before using in public.
First-draft structural depth. The machine gives you the scaffolding, the headings, and ~80% of the prose. Expect to tighten claims, sharpen voice, and verify specifics before you put your name on it.
Why we do this: a $4.99 receipt OCR and a $99.99 grant proposal draft are both “AI output,” but expecting the same level of polish from both will leave you disappointed or surprised. The dots tell you which one you're buying before you pay.