Write a chain-of-thought style prompt as a named process rather than a vague request to reason.
The move: give the model a workbench, not a slogan. "Think step by step" often helps, but it does not tell the model what kind of steps matter. A release diagnosis needs evidence, hypotheses, and checks. A policy recommendation needs criteria, trade-offs, and risk. A customer summary needs extraction, grouping, and synthesis. Named process steps solve that. They work like a lightweight operating procedure for reasoning. PDCA gives a simple pattern: plan the work, do the analysis, check the result, act on what you learned. For prompting, try this: Plan Restate the goal, decision, and constraints. Inspect Extract the relevant…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in