You're Prompting Wrong: AI Does What You Ask, Not What You Mean
Most people get bad results from AI because they write bad prompts. AI doesn't read your mind; it reads your words. Here's how to write prompts that actually get you what you want.
Most people think AI doesn’t work well.
They type something vague into ChatGPT, get a generic response, and walk away thinking AI is overrated. But the problem isn’t AI. The problem is the prompt.
Here’s the thing nobody tells you: AI does what you ask, not what you mean. It’s not reading your mind. It’s reading your words. And if your words are vague, your results will be too.
The Gap Between What You Mean and What You Ask
Let’s say you want AI to help you write an email to a client about a project delay. Here’s what most people type:
“Write an email about a project delay.”
And here’s what AI gives you: a generic, bland email that could be about any project, to any client, from any company. It technically answered your question. But it’s useless.
Now look at what happens when you close the gap between what you mean and what you ask:
“Write a professional email to my client Sarah at TechFlow. The project is 2 weeks behind schedule because we’re waiting on their API documentation. I want to keep the tone positive, take partial responsibility, propose a new timeline of March 22, and suggest a quick call this week to align. Keep it under 150 words.”
That prompt gives AI everything it needs: the recipient, the context, the reason, the tone, the action items, and the length. The result will be dramatically better. Not because AI got smarter between the two prompts, but because you got more specific.
Why Vague Prompts Get Vague Results
AI works by predicting the most likely response based on what you’ve given it. When you give it almost nothing, it has to guess. And when it guesses, it defaults to the most generic, average, middle-of-the-road answer possible.
Think of it like ordering food. If you walk into a restaurant and say “give me food,” you’ll get whatever the kitchen feels like making. But if you say “I want a medium-rare steak with mashed potatoes, no gravy, with a side salad and balsamic dressing,” you get exactly what you wanted.
AI is the same. Specificity is the difference between a useless response and a response that saves you hours.
The Five Elements of a Good Prompt
Every strong prompt includes most of these five elements. You don’t always need all five, but the more you include, the better your results.
1. Role
Tell AI who it should be. This frames the entire response.
Instead of asking a general question, say: “You are a senior marketing strategist with 10 years of experience in B2B SaaS.” Now every answer comes from that perspective, more specific, more expert, more useful.
Other examples: “You are a career coach specializing in tech transitions.” “You are a web developer reviewing code for security vulnerabilities.” “You are a business consultant advising a first-time founder.”
2. Context
Give AI the background information it needs. AI doesn’t know your situation unless you tell it.
Bad: “Help me with my resume.”
Better: “I’m a marketing manager with 5 years of experience in e-commerce. I’m applying for a senior product marketing role at a B2B SaaS company. My strengths are campaign management and data analysis. Help me rewrite my resume summary.”
The more context you provide, the less AI has to guess. And less guessing means better results.
3. Task
Be specific about exactly what you want AI to do. Not “help me” or “write something.” Tell it the exact output you expect.
Bad: “Help me with a presentation.”
Better: “Create a 10-slide presentation outline about why our company should invest in AI tools. Include a title slide, problem statement, 3 key benefits with data, 2 case studies, implementation timeline, budget estimate, and a closing slide with next steps.”
4. Format
Tell AI how you want the response structured. This is the element most people skip, and it makes a huge difference.
You can ask for bullet points, numbered lists, tables, paragraphs, code blocks, or specific lengths. You can say “keep it under 200 words” or “give me 5 options” or “format this as a comparison table.”
Without format instructions, AI decides for you. And it usually decides wrong.
5. Constraints
Tell AI what NOT to do. This is just as important as telling it what to do.
“Don’t use jargon.” “Don’t include generic advice.” “Don’t exceed 300 words.” “Don’t start with ‘In today’s fast-paced world.’” “Avoid clichés.” “Don’t use bullet points.”
Constraints prevent AI from falling into its default patterns: the generic intros, the overused phrases, the unnecessarily long responses.
Real Examples: Bad vs. Good Prompts
Writing a blog post
Bad: “Write a blog post about AI.”
Good: “Write an 800-word blog post for non-technical business owners about how AI tools can save them time on daily tasks. Use a conversational tone, include 3 specific tool recommendations with use cases, and end with a clear call to action. Don’t use technical jargon or buzzwords.”
Preparing for a meeting
Bad: “Help me prepare for a meeting.”
Good: “I have a meeting with my manager tomorrow to discuss my promotion. I’ve been in my role for 2 years and led 3 major projects that increased revenue by 15%. Help me prepare 5 talking points that highlight my impact, anticipate 3 objections my manager might raise, and suggest responses to each.”
Building a website
Bad: “Build me a website.”
Good: “Build me a portfolio website for a freelance graphic designer. It should have a home page with a hero section, an about page, a portfolio gallery with 6 project cards, and a contact form. Use a minimalist design with a dark background, white text, and blue accent colors. Make it responsive for mobile.”
Notice the pattern. Every good prompt tells AI the who, what, why, how, and how much. Every bad prompt leaves AI guessing.
The Iterative Approach
Here’s something else most people get wrong: they treat prompting as a one-shot thing. You type one prompt, get one response, and judge AI based on that single interaction.
But the best results come from conversations. Treat AI like a collaborator, not a vending machine.
Start with your prompt. Review the response. Then refine: “This is good but make the tone more casual.” “Shorten the second paragraph.” “Add a specific example for point 3.” “Rewrite the introduction with a stronger hook.”
Each follow-up prompt makes the output better. The first response is a draft, not the final product. The people who get the most out of AI are the ones who iterate.
Start Here
If you take one thing from this post, let it be this: before you type a prompt, ask yourself “if someone sent me these exact words with no other context, would I know exactly what they want?”
If the answer is no, your prompt needs more detail.
AI is powerful. But it’s only as good as the instructions you give it. The difference between “AI doesn’t work” and “AI just saved me 5 hours” is almost always the prompt.
Stop telling AI what to do. Start telling it what you actually need.
Want ready-made prompts you can use right now? Download my free AI Prompt Cheatsheet with 20 copy-paste prompts for common tasks.