One dense doc. The six core patterns that separate reliable AI output from coin-flips — role, few-shot, chain-of-thought, spec-first, verification, red-team — each with a tiny worked example, how to stack them, and the mistakes that quietly wreck your results. $4.99. Keep it open while you work.
Instant download · Stripe secure checkout · Stuck or unhappy? Reply to your receipt — a real person reads it.
The whole field compresses to one idea: models predict what comes next, so your job is to make the good answer the obvious next thing to say. Every pattern here is a concrete way to do that.
What's inside
The six patterns — and more
RoleSet who's answering and what they optimize for — narrows the whole response toward that expertise.
Few-shotShow 2-3 examples so the model copies the pattern instead of guessing your format.
Chain-of-thoughtReasoning first, conclusion last — so the answer is built on the thinking, not rationalized after.
Spec-firstDefine "done" up front: format, length, include, exclude. The highest-leverage habit on the sheet.
VerificationMake the model re-check its own output against a rule before you trust it.
Red-teamAssign the skeptic role so it surfaces the failure modes agreeableness hides.
Combining patternsHow a strong real-world prompt stacks role + spec + reasoning + verification — with a worked example.
10 common mistakes & the fixA table of what quietly kills output quality — and the one-line correction for each.
The 30-second checklistSix boxes to run before you send anything that matters.
Instant download · Stripe secure checkout · Works with any AI tool — Claude, ChatGPT, Gemini.
The one promise
No jargon-for-jargon's-sake, no 40-page course padded to feel worth it. One tight reference you'll actually keep open — understand these six patterns and you're past 90% of people typing into an AI box.