Prompt Engineering Basics for Beginners

Emerrank Consultancy Team Apr 13, 2025 10 min read Lesson 5 — Beginner
AIBeginnerTutorial

Two students ask the same AI to help with an assignment. One types "write about AI." The other writes "Explain three real uses of AI in Indian agriculture in 200 words, simple language, bullet points." Guess who gets the useful answer?

Prompt engineering is the skill of talking to AI clearly so you get what you actually need. It is not magic words or secret hacks — it is good communication, like briefing a smart teammate who takes instructions literally.

What Is Prompt Engineering?

A prompt is everything you send to an LLM: your question, background context, examples, and formatting instructions. Prompt engineering means designing those inputs deliberately to improve accuracy, tone, and structure of the response.

Think of ordering food at a busy restaurant. "Something good" gets you a random dish. "Vegetarian thali, medium spice, no onion, extra roti" gets you exactly what you wanted. Same kitchen, different order clarity.

Anatomy of a Good Prompt

Strong prompts often include four parts:

  • Role — "You are a friendly engineering tutor."
  • Task — "Explain binary search to a first-year student."
  • Context — "They know arrays but not recursion yet."
  • Format — "Use one analogy, three bullet points, under 150 words."
System message  →  role + global rules
User message    →  task + context + format
Assistant reply →  the model's output

In apps, developers store the role in a system prompt and pass user questions separately — like giving a new employee a job description before each customer call.

Simple Techniques That Work

Be specific. Vague prompts get vague answers. Replace "summarise this" with "summarise in five bullet points for a manager who has two minutes."

Chain of thought. Add "think step by step" for math or logic problems. The model breaks reasoning into visible steps, which often improves accuracy.

Few-shot examples. Show one or two examples of input → desired output before your real question. The model copies the pattern.

Iterate. First reply not perfect? Refine: "Shorter." "Add an example." "Use Hindi terms with English explanations." Conversation is a loop, not one shot.

Delimiters. Wrap long context in markers like triple quotes or XML tags so the model knows what is reference material versus the actual task.

Before and After Example

Weak prompt: "Tell me about APIs."

Strong prompt: "Explain what a REST API is to a first-year CS student. Use a restaurant waiter analogy. Cover: what API stands for, one real example (food delivery app), and why developers use APIs. Max 120 words."

The second prompt controls audience, structure, length, and example type. You might be wondering if that is cheating — it is not. It is professional communication.

Real-World Example: Support Email Drafting

A SaaS company templates prompts for support agents:

System: You are a polite support agent for Acme Cloud.
Use only the FAQ below. If unsure, say you will escalate.
Never invent refund policies.

FAQ: [paste policy text]

User: Draft a reply to: "I was charged twice this month."

Agents paste customer emails as the user message. The LLM drafts consistent, policy-aligned replies. Prompt engineering here means guardrails — not creative writing tricks.

Common Misconceptions

"Longer prompts are always better." Fluff confuses the model. Be concise and structured.

"There is one perfect prompt." Prompts depend on model, task, and data. Test and iterate.

"Prompt engineering replaces coding." Apps still need APIs, validation, and security. Prompts are one layer.

"You must use English only." Many models handle Hindi and mixed language well — specify your language preference explicitly.

Quick Recap

  • Clear prompts = role + task + context + format.
  • Specificity beats cleverness.
  • Use examples, step-by-step instructions, and iteration.
  • System prompts set behaviour; user messages carry each request.

Summary

Prompt engineering is how you steer LLMs toward useful outputs. You already do a lighter version when you rephrase a Google search. With AI, the skill matters more because the model tries hard to please whatever you wrote — even if your request was ambiguous.

Think of yourself as a director on a film set. The actor (LLM) is talented, but needs a clear script. In Lesson 6, we meet the language most AI developers use daily: Python.

Frequently Asked Questions

A prompt is the text you send to an LLM — your question, instructions, and any context. Better prompts lead to better outputs.

A system prompt sets the AI's role and rules, like "You are a patient tutor. Use simple language." It guides behaviour across the conversation.

No. Start with ChatGPT, Copilot, or any LLM chat interface. Practice writing clear instructions and compare results.

You include a few input-output examples in the prompt so the model learns the pattern you want before answering the real question.

Prompts may be too long, vague, or contradictory. Put critical instructions first, be specific, and break complex tasks into steps.

Yes. Developers, product teams, and support engineers all write prompts for apps. Clear communication with LLMs is increasingly valuable.

Key Takeaways

  • Prompt engineering is clear instruction-writing for LLMs — role, task, context, format.
  • Specific prompts outperform vague ones; iteration is normal and expected.
  • System prompts define behaviour; user messages carry each request.
  • Techniques like chain-of-thought and few-shot examples improve complex tasks.
  • In production, prompts include guardrails — not just creative instructions.

Suggested Next Reads

Share: LinkedIn Facebook X

Need help implementing this in your organization?

Contact Emerrank Consultancy