Prompting basics

A good prompt is not clever -- it is clear. The difference between a useless AI response and a genuinely helpful one almost always comes down to how you asked. In this lesson, you will learn a simple four-part recipe that makes your prompts better immediately.

Why prompts matter

Remember our pattern machine from Lessons 1 and 2? The AI model has absorbed patterns from billions of examples. When you send it a prompt, it uses those patterns to generate a response that "fits" your input. The problem is: a vague input matches many possible patterns, so the model picks one at random. A specific input narrows the field, so the model produces something much closer to what you actually want.

Think of it like asking a friend for a restaurant recommendation. "Know any good places to eat?" could get you anything from a hot dog stand to a Michelin-starred restaurant. But "Know a quiet Italian place near downtown, good for a date, under 50 dollars per person?" -- that gets you a useful answer.

Prompting works exactly the same way.

The four-part recipe

Every good prompt has up to four parts. You do not always need all four, but knowing them gives you a toolkit for any situation.

1. Task -- What do you want the AI to do? Start with a clear verb: write, summarize, list, compare, translate, explain, brainstorm. The more specific the verb, the better the result.

2. Context -- What does the AI need to know about your situation? Who is the audience? What is the purpose? What has already been tried? Context is the information you would give a smart colleague who is helping you for the first time.

3. Format -- How should the output look? Bullet points? A single paragraph? A table? A numbered list? A specific word count? Telling the AI the shape of the answer saves you from getting a rambling essay when you wanted three bullet points.

4. Example -- What does a good answer look like? If you can show the AI one example of the tone, style, or structure you want, the quality jumps dramatically. You do not always need this, but when the AI keeps missing the mark, an example is usually the fix.

Bad prompt vs. good prompt

Let's see the recipe in action.

Bad prompt:

"Write me something about marketing."

What will you get? Probably a generic 500-word essay about marketing principles. It is technically correct but useless for any real purpose.

Good prompt:

"Write 3 Instagram caption options for a coffee shop launching a new cold brew. Tone: casual and fun. Each caption should be under 150 characters. Here is an example of the tone I like: 'Your iced latte just got a cooler cousin.'"

Let's break down what changed:

• Task: Write 3 Instagram caption options (specific verb, specific quantity)
• Context: for a coffee shop launching a new cold brew (the situation)
• Format: under 150 characters each (a clear constraint)
• Example: "Your iced latte just got a cooler cousin" (the tone we want)

The difference in output quality between these two prompts is enormous -- and all you did was be specific about what you wanted.

When the answer is not right: the debugging checklist

Sometimes even a decent prompt gives a mediocre answer. Before you throw your hands up, run through this checklist:

Is the task clear? Did you use a specific verb? Did you say exactly what you want? "Help me with this email" is vague. "Rewrite this email to be shorter and more direct while keeping a warm tone" is clear.

Is the context sufficient? The AI does not know what you know. If the answer feels off-base, you probably left out information the AI needed. Add it in your next message.

Did you specify the format? If you got a wall of text when you wanted bullet points, that is a format problem. Just say so: "Give me that as 5 bullet points, each one sentence."

Would an example help? If the tone or style is wrong, paste in one example of what you are looking for. "Here is the style I want:" followed by a short sample works wonders.

Most bad answers can be fixed in one or two follow-up messages by adding the missing piece. This back-and-forth is called iteration, and it is completely normal. Even experienced AI users rarely get the perfect answer on the first try. The skill is knowing which piece to add next.

A real iteration example

Say you asked: "Help me write a thank-you note to my manager."

The AI produces something formal and stiff. Not terrible, but not you.

Your follow-up: "That is too formal. Make it warm and genuine, like I am texting a mentor I respect. Keep it to 3-4 sentences."

Now the AI produces something much closer to your voice. Maybe one phrase still feels off.

Your second follow-up: "Great. Replace 'I am deeply grateful' with something less dramatic."

Done. Three messages. A thank-you note that sounds like you. That is what good prompting looks like in practice -- not a single magical sentence, but a quick, focused conversation.

Save your best prompts

Once you find a prompt that works well for a task you repeat, save it. Keep a note on your phone, a document, or a bookmarks folder. You are building a personal toolkit.

For example, if you regularly need to summarize meeting notes, your saved prompt might be: "Summarize the following meeting notes into 3 sections: decisions made, action items with owners, and open questions. Use bullet points. Keep it under 200 words."

You will use this idea heavily in Lesson 6 when we build your daily AI workflow with reusable templates.

What is next

You now know how to write prompts that get useful answers. But how do you know when to trust those answers? In the next lesson, we explore where AI shines and where it fails -- so you know when to lean in and when to double-check. Head to Strengths and limits when you are ready.

---