Have you ever wondered why your AI assistant sometimes hits the mark — and other times misses it completely? The secret lies in how you ask. Welcome to prompt engineering: the art of formulating input so AI delivers the best possible results.
Think of prompting as instructing a skilled actor — not programming a computer.
If you ask:
“Tell me about cybersecurity”, you’ll get a generic answer.
But if you ask:
“Name three common cybersecurity threats and explain how Microsoft Defender addresses them”
…you’ll receive a far more concrete and useful response.
The difference? Specificity and structure.
Break complex prompts into clear components:
Define the AI’s role → “You are a marketing copywriter specialising in B2B”
Give clear instructions → “Write in an enthusiastic tone and max. 200 words”
Add relevant context → “The target audience is energy companies…”
Specify the output format → “Plain text, no markdown”
For more advanced prompts, you can use structured formats such as XML for even greater precision. See good examples here:
https://github.com/disler/marimo-promptlibrary/blob/main/prompt_library/chapter-gen.xml
https://cookbook.openai.com/examples/gpt4-1_prompting_guide#5-general-advice
“Few‑shot prompting” can significantly improve results.
Example:
“Can you help me?” → “Could you please assist?”
You can then ask the AI to rewrite content in the same style. IBM offers an excellent guide to this technique:
https://www.ibm.com/think/topics/prompt-engineering-guide
For complex tasks, explicitly write:
“Think step by step”
This helps the AI break the problem down and improves accuracy — especially for analytical tasks.
Avoid pleaser AI
AI systems tend to be overly accommodating.
If you want honest feedback, try writing for example:
“Be direct and brutally honest. Identify every flaw in this code.”
This leads to more critical and valuable responses.
Experiment with short, creative prompts that pack a lot of meaning into very few words.
Examples:
“Respond as a panel of experts”
“Respond as if you are 25 years in the future”
These can unlock new perspectives and more creative outputs.
Find more micro‑prompts here:
https://www.reddit.com/r/PromptEngineering/comments/1k6fmje/a_collection_of_absurdly_useful_microprompts/
Ready to improve your AI interactions? Start exploring Fabric
https://github.com/danielmiessler/fabric/tree/main/patterns/
an extensive prompt library with proven patterns for extracting insights, analysing claims, and much more.
If you want to go even deeper, Anthropic’s official documentation on prompting offers thorough guidance:
https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview
Remember: better prompts lead to better results — and ultimately to better business outcomes. You’ve read this now, so there’s no excuse not to start optimising your prompts and experience how your AI productivity improves.
