Introduction to Prompt Engineering — Prompt Engineering Course

What you'll learn

The same AI model can feel like a genius or a disappointment — and most of the difference is not the model, it's how you ask. Prompt engineering is the skill of asking well: designing the text you give a model so it reliably does what you actually want. This course teaches that skill from the ground up, with real prompts and the responses they produce on every page.

By the end of this first module you'll be able to:

Say what a prompt and prompt engineering really are
Explain why the wording of a request changes the answer so much
See — in one side-by-side example — the gap between a vague prompt and an engineered one
Adopt the iterative, experiment-like mindset the rest of the course builds on

What is a prompt?

A prompt is simply the text you give a language model to get a response. When you type a question into a chatbot, that message is a prompt. When an app asks an AI to summarize an email behind the scenes, the instructions and the email it sends across together form the prompt. Crucially, the model sees nothing but the prompt — no tone of voice, no context in your head, no memory of what you meant. Whatever you want it to know, you have to put into words.

Prompt

Suggest a name for a coffee shop.

AI response

Here are a few ideas: The Daily Grind, Bean & Gone, Steam & Cream, Morning Ritual, and Brew Haven. Want names with a particular vibe — cozy, modern, or playful?

A prompt in, a completion out. Everything the model knows about your goal came from those six words.

Note

Because the model only ever sees the prompt, prompt engineering is really just communication — the discipline of turning what you want into words a model can act on. No machine-learning background required.

What prompt engineering is (and isn't)

Prompt engineering is the practice of deliberately designing, testing, and refining prompts to get reliable, high-quality results from a model. It sits on top of a model you didn't train and can't change — so the prompt is your steering wheel. It matters just as much for a one-off chat as for a production feature that runs the same prompt a million times a day.

A few myths worth clearing up right away:

It isn't "magic words." There's no secret incantation. It's clear thinking, made explicit.
It isn't only for programmers. Most of this course is about language, not code — though we'll show how to prompt through an API when it helps.
It isn't a fixed trick. It's an iterative craft: you rarely nail the perfect prompt on the first try, and that's normal.

Why the way you ask matters so much

Here is the whole reason this skill exists, in one comparison. Both prompts below go to the same model. The only thing that changed is how the request was written — and the results are worlds apart.

✗Weak prompt

Prompt

write about dogs

Response

Dogs are domesticated mammals and one of the most popular pets in the world. There are hundreds of breeds, varying in size, shape, and temperament. Dogs are often called man's best friend because they are loyal and affectionate. They need food, exercise, and regular veterinary care…

✓Strong prompt

Prompt

Write a 60-word paragraph for a 3rd-grade science worksheet explaining why dogs have such a good sense of smell. Use one everyday example, and end with a question for the student.

Response

Dogs have an amazing sense of smell — much stronger than ours! Their noses have millions more smell detectors than a human's. That's why a dog can find a hidden treat in another room, or sniff out a friend before seeing them. Some dogs even help people by smelling for danger. Can you think of a job where a dog's powerful nose would be helpful?

Same model, same topic. The engineered prompt fixed the audience, length, format, and goal — so the answer is usable instead of generic.

Notice what the strong prompt actually did: it named the audience (3rd grade), the length (60 words), the format (a single paragraph), the content (one everyday example), and the ending (a question). It removed the guesswork. The vague prompt forced the model to guess all of that — and an average guess is exactly what it returned.

Key idea

A model can't read your mind — it can only read your prompt. Almost every improvement in this course comes from replacing something the model had to guess with something you stated on purpose.

The prompt engineer's mindset

The single most useful habit is to treat prompting like a small experiment rather than a wish. You draft a prompt, run it, look honestly at where the answer missed, change one thing, and run it again. Great prompts are almost always revised prompts.

The prompt iteration loop

Step 1

Draft

Write your first attempt at a prompt.

Step 2

Run

Send it to the model and read the answer.

Step 3

Inspect

Where did it fall short of what you wanted?

Step 4

Refine

Change one thing, then run it again.

You'll run this loop constantly — for a quick chat it takes seconds; for a production prompt it becomes a real testing process (Phase 7).

Tip

When an answer disappoints, resist the urge to rewrite everything. Ask "what did the model have to guess here?" and make just that one thing explicit. Changing one variable at a time is how you learn what actually works.

The map of this course

We build up in eight phases, each standing on the last — from "what is a prompt?" all the way to evaluating prompts and building AI agents. Early phases work entirely in the chat box; later ones move to the API where prompting powers real software.

Eight phases, beginner to advanced

Foundations

What prompting is, how models read your words, and the parts of a prompt.

The prompting toolkit

Clear instructions, roles, examples, structure & shaping output.

Making models reason

Chain-of-thought, decomposition & advanced reasoning.

Structured & reliable output

JSON & schemas, constraints & the sampling dials.

Grounding & knowledge

Beating hallucination with RAG, long context & context engineering.

Tools, agents & the API

Call the API, give the model tools, and build agents.

Production prompting

Evaluate, debug, secure & use prompts responsibly.

Applied prompt engineering

Reusable patterns, domain playbooks & what's next.

Every phase is taught with real prompts and the responses they produce — you learn by seeing what works.

You only need two things to follow along: the ability to type a request in plain language, and the curiosity to ask "why did it answer that?" A little Python helps for the API modules in Phase 6, but it's never required to understand the ideas.

Recap & quick check

Key takeaways

A prompt is the text you give a model — and it's all the model sees about your goal.
Prompt engineering is deliberately designing and refining that text to get reliable, high-quality results.
It's communication, not magic words: it makes what you want explicit instead of leaving the model to guess.
The same model gives wildly different answers depending on how you ask — the prompt is your steering wheel.
Treat prompting as an experiment: draft, run, inspect, refine — changing one thing at a time.

Quick check

1. What is a 'prompt'?

2. Why does the wording of a request change the answer so much?

3. Which statement about prompt engineering is TRUE?

4. An answer disappoints you. What's the recommended next move?

So the whole game is turning what you want into words a model can act on. To do that well, it helps to know a little about what's happening under the hood. Next up: Module 2 — How LLMs Actually Work.