If I cannot explain what I mean by make it more creative, a longer instruction will not resolve the uncertainty. I judge an AI brief by the clarity of its decisions rather than its length.
Name the actual task
Headline alternatives, research questions and a complete article are different assignments. I specify the output and where it will be used. The reader’s familiarity with the subject also determines the appropriate starting point.
Explaining a fictional product to a first-time user differs from describing a change to an existing user. Asking for compelling copy in both cases hides that distinction.
Establish the evidence boundary
Which information is verified, provisional or excluded? When supplying source material, I make clear that unsupported achievements, personal experiences and client outcomes must not be invented. Missing information should produce a question or visible gap rather than a confident guess.
Names and sensitive details may be unnecessary for the task. Reducing them can simplify the brief while limiting what is shared. Preserving context does not require handing over an entire archive.
Demonstrate the voice
Warm, professional and strong can mean different things. A short example and an expression to avoid can be more useful. A voice that distinguishes observation from interpretation and avoids guaranteed outcomes provides a concrete boundary.
The example should not become a sentence pattern repeated in every draft. Consistent voice differs from identical structure. A question, comparison or scenario can each provide an appropriate opening.
Define the review
Source accuracy, repetition, relevance to the reader’s question and natural expression should have review criteria. This is where human editorial judgement enters the process.
A brief does not guarantee a finished article. It makes clear what the editor should accept, question or complete independently.