A few well-chosen words can change the direction of a ChatGPT response. That is the appeal of Compressed Prompting Instructions, or CPIs: short labels that stand in for longer instructions you might otherwise type repeatedly.
Overview
A CPI is not an official ChatGPT command, hidden feature, or magical “power word.” ChatGPT interprets it as ordinary language and uses the surrounding conversation to determine what you mean. Familiar instructions often work immediately. Unusual ones may produce inconsistent results unless you define them first.
For example, instead of writing, “Analyze this proposal aggressively, challenge its assumptions, and identify ways it could fail,” you might simply add RED TEAM. That saves time, especially when you regularly request the same kinds of analysis, rewriting, or explanation.
Three Useful CPIs
ELI10 — Explain the subject as though speaking to an intelligent ten-year-old. This usually produces plain language, basic definitions, and familiar examples.
RED TEAM — Challenge the proposal rather than accepting it at face value. Look for weak assumptions, risks, contradictions, and failure points.
80/20 — Apply the Pareto principle by identifying the relatively small number of actions or factors likely to produce most of the value.
These work because their intended meanings are reasonably easy to infer. Other examples include NO FLUFF, STEP-BY-STEP, PLAIN ENGLISH, BOTTOM LINE, PROS/CONS, and DEVIL’S ADVOCATE.
CPIs are especially useful when revising an answer. After ChatGPT produces a lengthy explanation, you can respond with BOTTOM LINE to request the conclusion, ELI10 to simplify it, or RED TEAM to test it. Several can also be combined: PLAIN ENGLISH + 80/20 + ACTION PLAN.
Do not pile on a dozen CPIs and expect precision. Some instructions may conflict, while others are too vague to guide the response reliably. If a particular CPI matters, define it: “When I write TRUTHMODE, separate verified facts, reasonable inferences, and unsupported assumptions.”
Create Your Own CPI Reference
Copy and paste this prompt into ChatGPT:
“Create an extensive alphabetical reference list of Compressed Prompting Instructions that I can use with ChatGPT. Include broadly understood shorthand such as ELI10, RED TEAM, 80/20, PLAIN ENGLISH, and STEP-BY-STEP. For each CPI, provide a one-sentence explanation of the result it is intended to produce. Clearly state that these are informal prompting shortcuts, not official ChatGPT commands. Organize the finished list alphabetically and avoid duplicates.”
Keep the CPIs that genuinely improve your results and discard the decorative nonsense. If shorthand repeatedly causes confusion, replace it with a clear sentence. A shortcut that needs constant repair is not saving you anything.