AI
Prompt Engineering: How to Get Useful Answers from AI at Work
In short: prompt engineering is the skill of writing clear instructions for an AI tool such as ChatGPT, Claude or Gemini, so it gives you a useful answer the first time. Say who it should act as, what the task is, what it needs to know, what the answer should look like, and give an example. Then check the result like you would a junior colleague's draft.
Why prompts matter
An AI model only knows what you tell it in that conversation, plus what it learned in training. A vague request gets a vague, generic answer. A specific request, with the facts and the format you need, gets something you can actually use.
The five parts of a good prompt
- Role: "You are an experienced accountant writing to a small-business client."
- Task: "Explain why their GST input credit was reduced this quarter."
- Context: the facts, numbers and documents it needs (with private details removed).
- Format: "Five short bullet points, plain English, then one line on what they should do next."
- Example: a sample of the tone or structure you like.
Before and after
| Weak prompt | Better prompt |
|---|---|
| Write a reply to this customer. | You handle customer service for a Delhi furniture store. Write a polite reply in simple English to the message below. Apologise once, give the new delivery date (12 November), and offer a 5% discount code. Under 80 words. |
| Summarise this report. | Summarise this 20-page field report for our donor in 6 bullet points: what we did, numbers reached, two problems, and what we need next quarter. Quote figures exactly as they appear. |
| Make a lesson plan. | Make a 40-minute class 7 science lesson plan on reflection of light: a 5-minute hook, one hands-on activity with classroom items, three check questions, and homework. Use Indian examples. |
Habits that improve every answer
- Break big jobs into steps. First an outline, then each section.
- Ask it to say when it is unsure, and to list the facts it relied on.
- Ask for sources when facts matter, and open them yourself.
- Keep a library of prompts that worked for repeated tasks, and share it with your team.
- Edit the prompt, not just the answer. If the result is off, say what was wrong and try again.
Mistakes to avoid
- Pasting client names, Aadhaar or PAN numbers, medical records or passwords into public AI tools. Remove or replace them first.
- Trusting numbers, laws or quotes without checking them.
- Asking for "the best" answer without saying what best means to you.
- Using AI output under your name without reading it fully.
Try it on your own work
My free guide shows ten everyday tasks for nine professions, each with a ready-made prompt you can run in one click: The suit has changed, not the job. When a prompt becomes a task your office repeats every day, that is when an AI agent or a small pilot makes sense. Here is how I help with that.
Questions people ask
What is prompt engineering in simple words?
It is writing clear instructions for an AI tool so it understands the task, the context and the format you want, and gives a useful answer the first time.
Is prompt engineering a real job?
It is mostly a skill rather than a separate job. People in every profession who learn to write good prompts get much more out of AI tools, and some technical roles focus on designing prompts for products.
Do I need to know coding for prompt engineering?
No. Clear writing and knowing your own work well matter more than coding for everyday use.
Is it safe to put client data into ChatGPT or other AI tools?
Avoid putting identifiable client, patient or financial details into public AI tools. Remove or replace them, or use a business version whose data terms you have checked.