Give The AI A Job, Not Just A Task
If you ask, “Write an email,” the AI has to guess what kind of email, who it is from, who will read it, and what result you want. That guessing is why many answers sound bland. The fix is to assign a role, which simply means telling the AI what kind of helper to act like.
Instead of “Write an email to my landlord,” try: “Act as a calm tenant writing to a landlord. Help me request a repair without sounding angry. The goal is to get a clear next step by Friday.” The role does not need to be fancy. “Act as a patient tutor,” “act as a practical travel planner,” or “act as an editor for a busy executive” is enough to steer the answer.
Name The Reader And The Situation
Once the AI has a job, tell it who the answer is for. This matters because a good answer for a beginner is not the same as a good answer for an expert, and a note to a close friend is not the same as a note to a client. Without audience details, the AI often writes to a vague average person.
Give a short profile: “This is for my manager, who likes short updates,” or “This is for my mom, who is nervous about medical terms,” or “I am new to budgeting and need plain English.” Add the real situation too. For planning a trip, include the month, budget, pace, dislikes, and must-haves. For research, say whether you need a quick overview, a decision, or questions to ask next.
Show What Good Looks Like
After role and audience, examples are the fastest way to reduce generic output. An example can be a sentence you like, a format you want, or a rough version of your own words. You are not asking the AI to copy it exactly. You are giving it a pattern.
For example: “Use this style: short paragraphs, direct, friendly, no salesy language.” Or: “Here is a good update: ‘We finished the first draft, found two open questions, and need your decision on pricing by Thursday.’ Write mine in that style.” Bad examples can help too: “Do not write something like ‘I hope this message finds you well.’ Start with the reason for the email.”
Use A Narrow Prompt, Then Improve It
The biggest mistake is trying to get the perfect answer from one broad prompt. Better results usually come from a short back-and-forth. First, ask for a draft using role, reader, situation, and example. Then ask for a revision with one clear change: “Make it warmer,” “make it half as long,” “add three likely objections,” or “rewrite it for someone who has no background in this topic.”
You can also ask the AI to anticipate what comes next. If you are preparing for a meeting, ask: “What questions might they ask after reading this?” If you are learning something, ask: “What would a beginner probably misunderstand?” This turns the AI from a text generator into a practice partner. The central habit is simple: stop asking for “an answer” and start describing the fit you need.
Key takeaways
- Assign a role so the AI knows what kind of helper to be.
- Name the audience, their knowledge level, and what they care about.
- Give one example of the tone, format, or quality you want.
- Use negative examples to block phrases or styles you dislike.
- Revise in small steps instead of trying to perfect one giant prompt.
- Ask the AI to predict follow-up questions, objections, or misunderstandings.