Context, Role, Evidence, Output — A framework built for answers that need to be grounded in specific evidence.
CREO is the right choice when you're asking the AI to reason from specific facts, data, or source material rather than general knowledge — research summaries, analysis of a document you're pasting in, or any task where you want the answer traceable back to what you gave it.
The situation and what the evidence is being used to decide or explain.
The expertise or perspective the AI should apply when interpreting the evidence.
The specific facts, data, or material the answer must be grounded in.
The exact form the answer should take.
Bad prompt:
“What do you think about our survey results?”
Structured with CREO:
Context: We're deciding whether to prioritize a mobile app or a browser extension next quarter. Role: You are a product strategist evaluating user research. Evidence: 68% of respondents said they use our product primarily on desktop; 22% asked for mobile access; 10% mentioned browser workflow friction. Output: A one-paragraph recommendation citing which numbers drove the conclusion.
A one-paragraph recommendation that names a clear priority, explicitly ties the conclusion back to the specific percentages given rather than general assumptions about mobile usage.
CREO is especially strong with Claude for long-context, evidence-grounded analysis, and works well with ChatGPT when the evidence is pasted directly into the same conversation.