CodeFreeIQ
โ† All cheat sheets
Cheat sheet ยท 5 min read

5 things to stop putting in your prompts

Most 'bad AI answers' aren't a model problem. They're a prompt problem, and usually one of over-asking.

People think better prompts mean longer, more elaborate prompts. Often it's the opposite: the fastest way to a sharper answer is to remove the clutter that's confusing the AI. Here are five things to cut, and what to do instead.

What's inside

1

1 and 2: Padding, and ten questions at once

Long preambles ('I hope you can help me, I'm not very technical, please could you possibly...') don't improve answers, they bury the actual request. And stacking five unrelated questions into one prompt forces the AI to split its attention. Ask the real thing, clearly, and ask one thing at a time when it matters.

What this means for you: clarity beats politeness. Say what you want in the first line. You can always be warm; just don't make the AI dig for the ask.
2

3 and 4: Vagueness, and no example of 'good'

'Make this better' gives the AI nothing to aim at. Better than what? For whom? The other silent killer is not showing an example of what good looks like. If you have a piece you love, paste it and say 'match this tone'. The AI is brilliant at pattern-matching, but only if you give it a pattern.

Instead of

'Write me a good LinkedIn post about AI.'

Try

'Write a LinkedIn post for non-technical business owners, punchy, first person, like this example: [paste].'

What this means for you: specificity is the whole game. A vague prompt gets a vague answer, every time.
3

5: Treating the first answer as final

The last mistake isn't in the prompt, it's in what you do with the reply. The first draft is a starting point, not the finished thing. The real magic is in the follow-up: 'cut that in half', 'make it more direct', 'you missed X'. AI is a conversation, not a vending machine.

What this means for you: the people who get great output aren't better at the first prompt. They're better at the second, third and fourth.