A Small Prompt Library for Everyday Tasks
Reusable instruction templates for summarising, explaining, drafting, checking and learning. Copy, fill in the brackets and adapt.
A good prompt states the task, gives the necessary context and describes the result you want. The templates below follow that pattern. Replace the parts in square brackets, and remove anything that does not apply.
These templates work with most AI assistants. Results will still vary, so read every answer critically and check facts against original sources, as described in A Practical Guide to Working with AI Assistants.
Summarise
Summarise the text below for [audience, e.g. a busy manager / a 12-year-old].
Use at most [number] bullet points.
Keep names, numbers and dates exactly as written.
If something in the text is unclear or contradictory, say so
instead of guessing.
Text:
[paste text]
Explain a concept
Explain [concept] in three parts:
1. One sentence a beginner would understand.
2. A short paragraph with one everyday analogy.
3. The most common misconception about it, and why it is wrong.
Say where experts disagree, if they do.
Draft
Draft a [email / note / announcement] to [recipient].
Purpose: [what the reader should know or do].
Tone: [friendly / formal / neutral]. Length: about [number] words.
Must mention: [points].
Avoid: [anything to leave out].
Learn with examples
Rewrite each sentence in plain English, following the examples.
Example: "Utilise the aforementioned procedure." -> "Use this method."
Example: "Prior to commencement, ensure compliance." -> "Check the rules before you start."
Now rewrite:
[sentences]
Giving a few worked examples in the prompt is one of the most reliable ways to show an assistant what you want.
Check your own work
Review the text below as a careful editor.
List: (a) statements that need a source, (b) unclear sentences,
(c) arguments that do not follow. Do not rewrite the text;
only list the problems with a short reason for each.
Text:
[paste text]
Plan
Help me plan [goal] within [time / budget / constraints].
First list the assumptions you are making.
Then give a numbered plan with no more than [number] steps.
Mark any step where I should get expert advice.
Sources
- Brown, T. B. et al. Language Models are Few-Shot Learners. NeurIPS 2020.
- UK Government. Artificial Intelligence Playbook for the UK Government, 2025.