🧩 Pro course · 12 lessons · Go deeper

Prompting Patterns

Six repeatable patterns that work in any AI tool — decompose big jobs, make it interview you, compare options, and steer without starting over.

about 69 minutes total, one short lesson at a timefor office workers, managers, teams

Your first win: Turn vague AI requests into repeatable prompt patterns with criteria, constraints, and checks.

Every lesson below has a free preview. Pro unlocks the full lessons with live AI practice, quizzes, the coach, and spaced reviews.

1. One job per prompt

Big bundled requests produce mush; a pipeline of small single-job prompts produces quality

Telling the model to ask you questions first beats guessing what context it needs

Asking for three distinct versions with trade-offs beats accepting the first answer

Stating hard limits — length, banned content, must-keep facts — prevents the failures you cannot un-see

Targeted change requests beat re-explaining, and a fresh chat beats a polluted one

A winning prompt with fill-in-the-blank slots is a tool your whole team can reuse

Define what good looks like before you ask the model to produce anything

Ask for a plan, execute one step, then verify against criteria before continuing

Turn messy input into structured output, then use that structure in the next prompt

Ask a second pass to find risks, assumptions, weak claims, and missing perspectives

Format is part of the task — specify columns, keys, or checklist items exactly

Diagnose weak prompts with a checklist: job, context, criteria, format, scope