Most people get mediocre results from AI because they ask it the wrong way. They type a short question, get a generic answer, and assume AI is overrated. The truth is the output is only as good as the instruction you give it.
The Prompt Machine builds a structured, detailed prompt for you based on six questions. It also shows you exactly why each part of the prompt matters, so you start to understand how good prompting works, not just copy a result.
The Task field is where the real work happens. Be as specific as possible about your situation. "Write a follow-up email" is weak. "Write a follow-up email to a landlord who has not responded about a maintenance issue for 8 days, keeping it professional but firm" is strong. The machine does the rest.
The Prompt Machine builds these in automatically. But understanding why they work means you can apply them yourself, even when you are typing a quick message.
Telling the AI who to be changes everything about how it responds. A generic ask gets a generic answer. A role gives the AI a frame of reference for vocabulary, tone and expertise level.
If you do not tell the AI what format you want, it will choose one for you. Sometimes it guesses right. Often it does not. Telling it you want an email, a list, a script or a paragraph removes the guesswork.
Tone is subjective and AI will default to something safe and slightly corporate unless you tell it otherwise. Naming the tone you need keeps the output sounding like you, not like a press release.
Constraints are rules about what to avoid or what limits to stay within. They prevent the AI from going off in the wrong direction. Length limits, topic restrictions and style exclusions all count as constraints.
If the first response is not quite right, you do not need to start again. Reply in the same conversation with a quick refinement instruction. The AI remembers everything above and adjusts accordingly.
The more the AI knows about your specific situation, the more relevant the output. Background details, audience information and relevant facts all sharpen the result significantly.
Who should the AI be? A specialist role frames the vocabulary, expertise level and perspective of every response.
This is the most important field. Be specific about your situation, not just your goal. The more context you give here, the sharper the prompt.
Knowing the audience changes the vocabulary, assumed knowledge level and emotional tone of the output.
If you do not name the tone, the AI defaults to safe and corporate. Name it precisely and the output sounds like it came from a human.
Constraints prevent the AI from going off-script. Telling it exactly what format and length to produce removes the guesswork.
What should the AI avoid? What must it include? Constraints are the guardrails that stop the output going somewhere unhelpful.
Once the AI responds, you do not need to start again. Reply in the same chat window with one of these to adjust it:
