a visual course

How to Talk
to an AI

(Prompt Engineering)

Why the same AI gives one person gold and another mush — the craft of writing effective prompts, in about half an hour.

A free, visual course on prompt engineering — how to get dramatically better answers from ChatGPT, Claude and any LLM. Why prompts steer a next-word predictor, being specific, showing examples, letting the model think, structuring prompts, and iterating.

in the real world

Better prompts mean better emails, code, analysis and ideas from the AI you already use — and the same craft builds the system prompts inside every serious AI product.

Start with a bad prompt →

Part 1 — What a prompt really is

1

You're steering an autocomplete

The model doesn't read your mind — it continues your text. Your prompt is the steering wheel.

2

Say what you actually want

Vague in, vague out. Add audience, constraints and format — and watch the same model sharpen up.

Part 2 — The playbook

3

Show, don't tell

Two or three examples teach the model your pattern better than a paragraph of description ever will.

4

Give it room to think

Forced to answer instantly, the model guesses. Asked to reason step by step, it gets it right.

5

The anatomy of a great prompt

Role, context, task, examples, format, guardrails — the six blocks serious prompts are built from.

Part 3 — Prompting in practice

6

Iterate, then put it to work

Prompts are drafts. Test, tweak one thing at a time — and see where this craft pays off in the real world.