Guide
What Is Prompt Engineering?
Prompt engineering is the skill of writing instructions that get reliable, useful output from an AI model. This guide explains what it is, the techniques that actually work, and how prompting fits into building apps and AI agents — with copy-ready examples.
By Abdul Rahman Azam, founder of AI Season · Updated
Key facts
| What it is | Designing the instructions, context and examples you give an AI model |
|---|---|
| Core techniques | Clear role and task, context, examples, output format, step-by-step reasoning |
| Used for | Chatbots, content, code, data extraction, RAG and AI agents |
| You need | Nothing to start; basic Python to use prompts inside apps |
What is prompt engineering?
A prompt is the input you give an AI model. Prompt engineering is the practice of designing that input — the instructions, context, examples and output format — so the model produces what you need, consistently. In apps and agents, most prompts live in code as system prompts that every user message is combined with.
It matters because the same model can give a vague answer or an excellent one depending on how you ask. Good prompting is cheaper and faster than switching to a bigger model.
The anatomy of a good prompt
| Part | What to write | Example |
|---|---|---|
| Role | Who the model should act as | You are a patient tutor for first-year CS students. |
| Task | Exactly what to do | Explain recursion using one everyday example. |
| Context | Facts the model needs | The student knows loops but not functions calling themselves. |
| Constraints | Limits and rules | Under 150 words. No code longer than 6 lines. |
| Format | The shape of the answer | A short explanation, then a Python example, then one practice question. |
Techniques that actually work
- Be specific — say who the audience is, how long the answer should be and what to leave out.
- Give examples (few-shot) — two or three input-and-output examples teach a format better than a paragraph of rules.
- Ask for structure — request JSON with named fields, then validate it in code.
- Separate instructions from data — wrap pasted text in clear markers so the model does not mistake content for instructions.
- Ask it to reason first — for multi-step problems, ask the model to work through the steps before giving the final answer.
- Let it say "I don't know" — explicitly allow uncertainty, and ask it to cite the context it used.
- Iterate with test cases — keep a small set of inputs and re-run them after every prompt change.
Example: a system prompt for an app
You are a study assistant for university students in Pakistan and India.
Rules:
- Answer only from the course notes between <notes> tags.
- If the notes do not contain the answer, say "This isn't in your notes."
- Reply in simple English; add a short Urdu/Hindi gloss for key terms.
Output JSON: {"answer": string, "sources": [string], "confidence": "high" | "low"}
<notes>
{retrieved_notes}
</notes>Prompt engineering for AI agents
In an agent, prompting goes beyond one message. You write the system prompt that sets the agent's goal and rules, the tool descriptions the model reads to decide which tool to call, and the format for tool results. Clear tool descriptions often fix more agent bugs than any change to the main prompt.
Prompts are also an attack surface: text inside a web page or document can try to override your instructions (prompt injection). Treat retrieved content as data, limit what tools can do, and require approval for risky actions. See how to build an AI agent for these guardrails in code.
Is prompt engineering a career?
On its own, rarely. The demand is for engineers who can prompt and build: connect models to data with RAG, give them tools, evaluate them and deploy them. Prompting is module 3 of the free generative AI course for exactly that reason.
The AI Season bootcamp teaches prompting as part of the full agent stack, live in Urdu and English over 6 weeks. Cohort 02 starts 1st January 2027.
Frequently asked questions
What is prompt engineering in simple words?
It is writing clear instructions for an AI model — saying who it should act as, what to do, what context to use and what the answer should look like — so the output is useful and consistent.
Can I learn prompt engineering for free?
Yes. The techniques in this guide, plus practice with a free model tier, are enough to get good. Test each prompt on several inputs and improve it step by step.
Do I need coding for prompt engineering?
Not to write good prompts in a chatbot. To use prompts inside apps, automations or agents, basic Python helps a lot.
What is few-shot prompting?
Giving the model a few examples of the input and the output you want before the real input. It is one of the most reliable ways to control format and style.
What is prompt injection?
An attack where text the model reads — in a web page, email or document — tries to override your instructions. Defend against it by separating instructions from data and limiting what the model's tools can do.