Paste a prompt and watch what the AI is forced to guess. Every blank you leave, audience, goal, tone, length, format, the model fills in for you, differently each time. This is why vague prompts disappoint.
When you type “write me a landing page,” you know exactly what you mean, who it’s for, what it should achieve, how long, what tone. The model knows none of that. So it guesses, and it guesses differently every time. That’s why the output feels generic and why two runs of the same prompt give different results. This tool makes the invisible guessing visible: every decision you didn’t make, the model made for you.
Leave a decision blank and the model picks one, and you don’t control which. Pin them down and the output gets sharp and repeatable. This runs entirely in your browser; nothing is uploaded.
No. It detects which of the six decisions your prompt specifies and shows the ones it leaves open, using pattern matching, not a model. The “guesses” are illustrations of how differently a model could interpret the gap.
Make the blanks explicit. The Prompt Builder walks you through each decision, and the before/after gallery shows the difference it makes.
No. Detection is entirely client-side, so your prompt stays on your device.
Fix the gaps with the Prompt Builder, see the prompt’s structure with Prompt X-Ray, or start the Learn Prompting course.