Learn Prompting to get best out of the LLMs

I Thought ChatGPT Was the Product. Then I Realised Prompting Was the Real Skill.

When I first started using ChatGPT, I treated it like a search engine.

I’d ask a question.

It would answer.

I’d move on.

Sometimes the answer was amazing. Sometimes it felt generic. Other times, it completely missed what I wanted.

Like many people, I blamed the AI.

“It’s not smart enough.”

“It doesn’t understand.”

“It hallucinated.”

But after spending months building AI projects, working with LLMs, experimenting with RAG systems, and writing hundreds of prompts, I realised something that completely changed how I use AI.

The quality of the response isn’t determined only by the model. It’s largely determined by the instructions you give it.

The model is powerful.

Your prompt is the steering wheel.

Imagine Hiring the World’s Smartest Expert

Imagine you hire the smartest engineer in the world.

On their first day, you simply tell them the following:

“Build me an application.”

What happens?

They’ll probably ask dozens of questions.

• What kind of application?
• Who are the users?
• What technologies?
• What’s the deadline?
• What’s the budget?

Now imagine giving them this instead:

“Build a scalable AI chatbot using FastAPI, LangChain, FAISS, and OpenAI. It should support document uploads, RAG, conversation memory, and Docker deployment. Explain every step because I’m learning.”

Suddenly the results become dramatically better.

Large Language Models work exactly the same way.

They don’t magically know what version of the answer exists inside your head.

The more context you provide, the closer the output gets to what you actually want.

Prompting Isn’t a Cheat Code

A lot of people ask me,

“What’s the secret ChatGPT prompt?”

The truth is…

There isn’t one.

There are no hidden commands.

No developer backdoor.

No magic keyword that unlocks “GPT Ultra Mode.”

Instead, prompting is simply the skill of communicating clearly.

Just like giving instructions to another person.

The Same AI Can Become Many Different Experts

One thing most beginners don’t realise is that ChatGPT doesn’t have to answer everything in the same style.

You can completely change how it thinks and how it explains.

Instead of asking:

Explain neural networks.

You can ask:

Explain neural networks like a university professor.

Or…

Explain neural networks as if I’m five years old.

Or…

Explain neural networks from an interviewer’s perspective.

Same AI.

Completely different experience.

That’s when I realised I wasn’t just asking questions anymore.

I was choosing who I wanted the AI to become.

Here Are Some Prompt Styles I Use Every Week

1. Teacher Mode

Perfect when I’m learning something new.

Explain this like a university professor.

Start with intuition.

Then theory.

Then mathematics.

Then a real-world example.

2. Explain Like I’m Five (ELI5)

Perfect when concepts feel overwhelming.

Explain this as if I'm five years old.

Use simple words.

Use everyday examples.

Avoid technical jargon.

3. Mentor Mode

This is one of my favourites.

Instead of giving the answer immediately…

Don't give me the answer.

Guide me with hints.

Let me solve it.

Correct me only when necessary.

It feels like having a personal tutor instead of Google.

4. Interview Mode

Great for job preparation.

Act as a technical interviewer.

Ask one question at a time.

Evaluate every answer.

Increase the difficulty gradually.

It turns ChatGPT into a mock interviewer available 24/7.

5. Socratic Mode

Instead of answering…

It asks questions.

Only ask questions that help me discover the answer.

Don't reveal the solution unless I ask.

This surprisingly improves long-term understanding.

6. Devil’s Advocate

Whenever I think an idea is brilliant…

I use this.

Challenge every assumption I make.

Find weaknesses.

Present opposing viewpoints.

Suggest alternatives.

It’s like having someone review your thinking before the internet does.

7. Research Assistant

Perfect when reading papers.

Act like a research scientist.

Compare approaches.

Discuss limitations.

Provide evidence.

Mention future work.

Much more useful than a simple summary.

8. Senior Software Engineer

Whenever I finish coding…

Review my code like a senior engineer.

Find bugs.

Find security issues.

Find performance problems.

Suggest improvements.

This catches things I would’ve missed.

9. FAANG Interview Mode

Ideal for coding practice.

Act like a Google interviewer.

Focus on:

• Correctness

• Time complexity

• Space optimisation

• Edge cases

Don't reveal the solution immediately.

10. Project Builder

Instead of dumping thousands of lines of code…

Teach me by building the project one step at a time.

Explain every file.

Wait until I finish each milestone.

This completely changed how I learn new frameworks.

11. Debate Mode

Give arguments both for and against.

Present evidence.

Then conclude based on facts rather than opinions.

Great for understanding controversial topics objectively.

12. Strict Reviewer

Sometimes I don’t want encouragement.

I want improvement.

Be brutally honest.

Don't praise unless it's deserved.

Point out every mistake.

Explain how to improve.

My Favourite “Universal Prompt”

Whenever I don’t know exactly how to ask something, I start with this:

Act as an expert in this subject.

When answering:

• Start with intuition.

• Explain technically.

• Give a practical example.

• Mention common mistakes.

• Share best practices.

• End with a short summary.

• Suggest three follow-up topics I should learn next.

It’s simple.

But it consistently produces structured, high-quality responses.

What Makes a Powerful Prompt?

Over time, I realised the best prompts usually include five things:

1. Role

Who should the AI become?

Teacher?

Engineer?

Researcher?

Designer?

Recruiter?

2. Goal

What exactly do you want?

Learn?

Build?

Review?

Debug?

Summarise?

Compare?

3. Context

Why are you asking?

What’s your experience level?

What constraints do you have?

4. Format

How should the answer look?

• Bullet points
• Table
• Step-by-step guide
• Tutorial
• Flowchart
• Code
• Examples

5. Constraints

Should it be concise?

Beginner-friendly?

Production-ready?

Explain mathematics?

Avoid jargon?

The more specific you are, the more useful the response becomes.

Prompting Is Becoming a Core Skill

A few years ago, knowing how to search the internet efficiently gave people an advantage.

Today, knowing how to communicate effectively with AI is becoming just as valuable.

Whether you’re a student, developer, researcher, designer, marketer, or entrepreneur, your ability to ask better questions directly affects the quality of the answers you receive.

Prompting isn’t about tricking the AI.

It’s about thinking clearly enough to express what you actually need.

And that, I’ve realised, is a skill worth developing even outside the world of AI.

Final Thought

The biggest lesson ChatGPT has taught me isn’t about artificial intelligence.

It’s about communication.

The better I communicate my intent, the better the AI can help me.

So the next time ChatGPT gives you an answer that feels average, don’t immediately assume the model isn’t capable.

Pause for a moment and ask yourself:

Did I ask an average question?

Because sometimes, changing just a few words in your prompt doesn’t just improve the answer.

It changes the entire conversation.