AI Prompt Masterclass – Guide P01 of 10

Overview

This short masterclass is about the mechanics that make prompts reliable: clear roles, explicit constraints, output shape, and a review step. You will be able to reuse the same pattern across tools and models, so you spend less time fighting the model and more time editing good first drafts. Treat it as a checklist you paste above every serious prompt—especially when the stakes are high or the audience is not you.

Prompt building blocks

Copy the skeleton, fill the bracketed context, add your guardrails

1. Role + audience

“You are [role]. The reader is [audience and literacy level]. Tone: [formal|direct|supportive].”

2. Task + output shape

“Task: [one sentence]. Return: [bullets|table|email|checklist] with [max words].”

3. Context pack

“Context (facts only): [paste data]. Constraints: [brand/legal/time]. Unknowns: [say what is missing].”

4. Review pass

“List assumptions, then propose 3 risks and how to verify each before sending.”

✅ Pro tips

→Start narrow: one audience, one action, one format—then generalise after it works.

→Never paste secrets or personal data; summarise or redact in the “context pack.”

→Compare two temperatures: a strict “policy mode” and a creative “diverge mode” for headlines.

→End with: “If anything is uncertain, list questions before answering.”

This document is a free educational handout from IWAB (April 2026). Double-check any figures, legal text, or compliance claims with qualified professionals. Generative tools change; always verify in your live environment before production use.