Before and After: Turning Vague Prompts into Better Instructions

Why vague prompts create uneven results

A request such as ‘make this more professional’ leaves the audience, tone, length, and definition of professional unstated. The model has to guess, so the result may be polished but still unusable.

A better instruction makes the important decisions visible. It does not need to be long; it needs to be specific about the outcome and the boundaries.

Example 1: rewriting copy

Before: ‘Make this product description better.’

After: ‘Rewrite this product description for first-time buyers. Keep it under 90 words, explain the primary benefit in the first sentence, use plain English, and do not add claims that are not present in the source.’

The revision names the reader, sets a measurable limit, defines the order of information, and adds a truthfulness boundary.

Example 2: research synthesis

Before: ‘Summarize these reports.’

After: ‘Summarize the supplied reports in a table with columns for finding, supporting source, confidence, and open question. Use only the supplied material; if sources disagree, show the disagreement instead of resolving it silently.’

The revision makes the evidence boundary and the desired review format explicit.

How AURE helps with the rewrite

AURE can use a short request as a starting point and suggest a more structured draft with role, task, context, constraints, and format. Review the suggestions before sending the prompt: an enhancer can expose missing decisions, but it cannot know facts that you have not supplied.