Most prompting advice is about volume. Write longer prompts. Add context. Be more specific. These things help. They are also not the thing.
The real variable isn’t length. It’s form. And the form most people reach for — a question — is one of the least effective ways to get genuine thinking from an AI.
Questions invite answers. Answers don’t require reasoning — they require selection.
If you’ve trained yourself to notice the difference between those two things, you already know which one you want.
What Questions Actually Do
When you ask “what should I do about X?” you’ve opened a very large door. The AI can walk through it any number of ways — most of them reasonable, few of them useful. You’ve set a destination without setting any of the conditions, which means the system gets to invent its own assumptions about what matters.
This is fine for quick lookups. For creative work, strategy, or any problem that actually requires thought, it produces the equivalent of a shrug dressed up in confident language.
The issue isn’t that the AI is wrong. It’s that the question wasn’t sharp enough to make it right.
The Thing That Actually Works
Give it a situation.
A situation has edges. It has something that’s true, something that’s not possible, and something that has to be true anyway. When you put those three things together, you’ve created a problem that requires actual reasoning to solve — not just selection from a large menu of plausible answers.
“Write me a tagline for our product” is a question. You’ll get five options. They’ll be fine.
“The product is for people who distrust marketing, the budget is minimal, and the founder isn’t charismatic” is a situation. You’ll get something that has to be true to the conditions — because the conditions are specific enough to rule out the safe middle.
The difference between those two is not effort. It’s understanding that what you’re after isn’t an answer. It’s reasoning that happens to produce something useful.
The Constraint as the Invitation
The best work comes from the most specific conditions. This doesn’t mean limitations make you better in some motivational sense. It means constraints force a kind of honesty. When everything is possible, the easiest path is the most traveled one. When something is ruled out, you have to find the path that’s actually yours.
The more you constrain the problem — the more you specify what’s true, what’s not allowed, what matters above everything else — the more the AI has to work with something real instead of working with possibility space.
Possibility space is enormous. It contains everything that might be said. The specific situation is small. It contains the things that need to be said.
A Briefing, Not a Question
The best prompt isn’t a question. It’s a briefing. It tells the system what’s true, what’s at stake, and what success looks like in a way that rules out the obvious answers before the conversation even starts.
You already know how to do this. It’s how you talk to anyone you actually respect — you don’t ask open questions and wait for suggestions. You share the situation and invite the thinking.
Dante is built to work from briefings. Bring the situation. Bring what you know and what you don’t. The conversation that follows will be better for it.