Prompt Engineering Examples: 8 Prompts That Work

The difference between a useless AI answer and a genuinely helpful one is usually the prompt. These prompt engineering examples show the pattern behind good prompts — a clear task, useful context, and the format you want — with before-and-after pairs you can copy and adapt.

The pattern behind every good prompt

Almost every weak prompt is missing one of three things: the task (what to do), the context (what the AI can't know unless you say), and the format (what the answer should look like). Add those and results jump.

1. Writing an email

Weak:  write an email about the meeting

Strong: Write a short, friendly email to my team of 8 moving Thursday's
meeting from 3pm to 4:30pm because a client call overran. Ask anyone who
can't make it to reply by Wednesday noon. Under 80 words, no jargon.

2. Explaining a concept

Weak:  explain compound interest

Strong: Explain compound interest to a 15-year-old with one worked example
using 1,000 rupees over 3 years. Then give me two questions to test myself.

3. Summarising a document

Summarise the report below for my manager, who has two minutes. Give the
main finding in one sentence, then 3 bullets on what it means for our team,
then any number I should double-check.

[paste the report]

4. Getting a table you can use

Compare these three laptops for a student. Table columns: Model | Price |
Weight | Battery hours | Best for | Main drawback. Use only the specs I
paste; write "not given" if missing.

[paste specs]

5. Getting structured data (for spreadsheets or code)

Extract the event details from this email as JSON only:
{"title": string, "date": "YYYY-MM-DD" or null, "location": string or null}
Use null if a detail isn't in the email. Don't invent anything.

[paste email]

6. Using examples to fix the format

Turn my notes into flashcards in this format:
Q: What is osmosis?
A: Water moving across a membrane from low to high solute concentration.

Now do the same for these notes: [paste]

7. Asking for the reasoning (so you can check it)

We have 3 vans, each doing 12 drops a day. Tomorrow: 41 orders, 6 must
arrive before 11am. Work through it step by step, show the arithmetic,
then give a plan per van.

8. Getting honest pushback

Here's my plan to switch our team to a 4-day week: [plan]. Argue against
it as strongly as a sceptical manager would, then list what evidence
would change your mind either way.

Two rules that make every example better

• Add "If information is missing, ask me instead of guessing." It turns a confident wrong answer into a useful question.
• Remember prompting gets better answers, not necessarily true ones — always check any fact you'll rely on.

Go from examples to a real skill

These examples show the pattern; our Prompt Engineering Course teaches it properly — a reusable framework, examples, structured output, working with your own documents, and staying safe — with a free first lesson that's a plain ChatGPT tutorial. Developers who want to call AI from their own code should see AI for Developers.