Giving the AI Good Context

Lesson 7 · taught by Pip

Hello again, friend. Pip here. 🐙

We've now touched every big lever once — the goal, the context, the shape, the example, the role. This lesson is the one I'd hang above the door if I could only keep one, because it ties two of them together and makes them stick: an AI is only ever as good as what you tell it.

The AI doesn't know your world

You met this idea back when we asked a stranger for directions. Here's the same truth from a different angle, because it's worth nailing down. When you ask an AI a question, it knows a great deal about the world in general — but it knows nothing about your world. It can't see your desk, your customers, your last email, or what you're really trying to do.

So when you ask a bare question, the AI has to guess at all the missing pieces. And a guess is just that: a guess. Sometimes it's lucky. Often it isn't.

Context simply means the background details you give the AI so it doesn't have to guess — who you are, what you're working on, who the answer is for, and what "good" looks like to you.

The new-employee picture

Imagine a brand-new employee on their very first morning. They're sharp and willing — but they've never met your customers and don't know how you do things.

If you say "write a thank-you note," they'll write something. But it'll be generic, because they don't know your business yet.

If you say "write a thank-you note to a wedding client named Maria who booked our garden venue, keep it warm but not gushy, and mention we'll send photos next week," now they can write something that sounds like you.

The AI is exactly that new employee. The more you tell it, the less it has to invent.

The strongest context is an example

You'll remember the tailor from a few lessons back: bring in the jacket you love and the fit comes out right. An example is context too — the richest kind. If you have an email you loved sending last month, paste it in and say "write the new one in this style." That one sample carries the tone, the length, and the rhythm all at once, far better than any description of them.

So when you gather your background details, count a good example as the heaviest one in the bag.

A simple recipe for good context

You don't need anything fancy. Before you hit send, try adding:

• Who it's for — "this is for a busy restaurant owner who hates jargon."
• What you're trying to do — "I want them to book a call."
• Any facts the AI couldn't know — your prices, your dates, your name.
• An example — "make it sound like this: [paste]."

Even two of these will lift your answer noticeably. You don't have to do all four every time.

Your turn

Take a request you'd normally type in one line — say, "write a follow-up message to a customer."

Now rewrite it with three pieces of context: who the message is for, what you want them to do, and one real detail only you would know (a name, a date, a product). Notice how much more yours the next answer feels.

That difference is the whole lesson in your hands.

Onward

You're learning to hand the AI the right ingredients — and that's most of the battle. Next we'll look at what to do when the first answer isn't quite right: how to nudge, correct, and steer it toward exactly what you meant. See you there. 🔦