You don't need a hundred research prompts. You need five you trust and actually reuse.
Keep a negative list
Ask for the reasoning, then the result. When the model shows its steps you can see exactly where to push back.
- Ask for three angles, then cut to the one worth shipping
- Save the prompt that worked so the next person reuses it
- Compare two outputs instead of trusting one
Generate options, not answers
Ask for the reasoning, then the result. When the model shows its steps you can see exactly where to push back.
Cut before you polish
Treat the first output as a draft, not a verdict. The prompt does the typing; you keep the judgment about what ships.
- Ask for three angles, then cut to the one worth shipping
- Save the prompt that worked so the next person reuses it
- Lower the reading level, then check what broke
Write down what you never want as explicitly as what you do. The exclusions often do more work than the instructions.