What sets apart those who truly benefit from AI from those who just use it for nonsense comes down to prompts. It’s not magic, it’s knowing how to ask for things.
Look at the difference. A vague prompt like “write something about marketing“ gives you back a generic text that’s completely useless. On the other hand, if you tell it: “Act as a marketing director with 10 years of experience. Create a strategy for a sustainable fashion online store targeting women aged 25 to 40. Include priority channels, monthly budget per channel, and KPIs to measure. Format: executive document with bullet points”, the AI delivers a detailed and useful plan you can use right away. The key is giving it a role, a complete context, and being very specific about what you want it to do and how.
? For programming and working with code
This is where AI goes from being a curiosity to a real work tool. It’s not just about it writing functions for you, but about it thinking like an engineer.
- To have it optimize your code: Instead of “fix this”, tell it: “Optimize this Python code to be more efficient in execution time and memory usage”. It will suggest changing algorithms or data structures.
- To make it more readable: “Refactor this code to improve its readability without changing its functionality”. It will break up long functions and simplify the logic.
- To debug errors: Be clear about the problem. “This Python code doesn’t produce the expected result. What logical changes should I make?”. If it’s a syntax issue, ask it directly: “Find and fix the syntax errors in this snippet”.
? How to think when writing a good prompt (without overcomplicating it)
Forget rigid formulas. Think about these four things every time you’re going to ask for something, and the results will improve a lot:
- What is it for? (The end goal: “A proposal for a client”, “a script for a short video”).
- Who is going to see it? (The context: “For a technical blog”, “for my boss”, “for high school students”).
- What do you want it to give you back? (The format: “A table”, “a list of 5 points”, “a 100-word email”).
- How should it sound? (The tone and style notes: “Professional but approachable tone”, “avoid technical jargon”, “use a practical example”).
? A few more examples so you can see the pattern
- For a local business: “Act as a communications advisor. Write a 200-word proposal for a client who wants to improve her visibility on Instagram. Her business is a yoga studio in Madrid for women over 40. Use a warm and professional tone.”
- To create training content: “Design a 1-hour training session on [topic] for adults with no prior knowledge. Include a block-by-block structure, objectives, and a final activity.”
- For analysis: “Acting as a data analyst, examine these performance metrics. Identify patterns and give 3 recommendations to improve ROI.”
? A couple of final tips
- Iterate and correct: Your first prompt is almost never the final one. Read the response, see what’s missing or what’s too much, and ask for adjustments: “Cut this in half”, “make it more formal”, “expand more on the second point”.
- For complex tasks, ask it to think step by step: This is called chain-of-thought. For a difficult problem, tell it: “Solve this by explaining each step of your reasoning”. You’ll see how it reaches the solution and it’s easier to verify if it makes a mistake.
- Watch out for hallucinations: AI makes things up with conviction. Don’t take specific data (dates, figures, quotes) at face value without verifying them. Use it to generate ideas, structures, or first drafts, not as an absolute source of truth.
In the end, it’s like working with a super-fast but literally-minded intern: you have to give them crystal-clear instructions and review their work. The better you explain what you need, the less you’ll have to redo yourself.