Prompts for everyday life and IT: Templates that yield better answers

In short: Good prompts state role, task, context, format and examples; with such templates you get clearer, more useful answers from AI assistants.

In more detail: A language model guesses what you want. The more precisely you describe task, background and desired form, the better the result. The templates below are starting points; adapt them and always check the answers.

In Depth

The basic frame

Role – task – context – format – constraints – example:

You are a patient IT trainer.
Task: Explain subnetting to me.
Context: I am in the first year of IT specialist training and understand binary numbers.
Format: 5 steps, then 2 practice tasks with solution.
Constraint: No technical terms without explanation, at most 250 words.

Stating the target audience, length and format (list, table, code) often brings the most.

Templates for learning

• Explain: “Explain [topic] once simply and once with technical terms. Give an everyday example.”
• Quiz: “Ask me one question after another on [topic], wait for my answer, rate it and explain mistakes.”
• Compare: “Compare X and Y in a table (advantages, disadvantages, use).”
• Summarise: “Summarise this text in 5 bullet points; mark uncertainties.”
• Feedback: “Check my text for clarity and argumentation, without rewriting it; name 3 concrete improvements.”

Templates for IT and work

• Find errors: “Here is my code and the error message (complete). Explain the cause and suggest a fix. Version: Python 3.12.”
• Explain code: “Explain this function line by line for a beginner.”
• Tests: “Write pytest tests for this function including edge cases.”
• Regex/SQL: “Write a regular expression for …, with 3 examples that match and 3 that do not.”
• Email: “Draft a polite email to [role] in which I describe [concern] with a deadline of [date]; max. 120 words.”
• Documentation: “Create from these bullet points a README with sections installation, usage, licence.”

Tips for better results

1. Iterate: criticise the result (“shorter”, “with more examples”, “different approach”).
2. Supply context: version, environment, goal; no confidential data.
3. Ask for step-by-step thinking: “Proceed step by step and state assumptions.”
4. Allow uncertainty: “If you are not sure, say so.”
5. Ask for sources and check them yourself.
6. Keep your own prompt collection (notes). Prompts are craft: practice and critical checking count more than “magic sentences”.

See also: Prompt engineering, Using AI assistants sensibly and safely, Learning with AI, AI and privacy in everyday life