What "Context" Really Means

Lesson 2 · taught by Pip · free preview

You'll hear the word context thrown around constantly with AI, usually without anyone explaining it. Let's fix that, because it's the single most useful word in this whole subject — and it's far simpler than it sounds.

Here's the plain idea: context is everything the AI can see at the moment it answers you. That's your latest message, plus everything earlier in the same conversation, plus any instructions it was given to start with. If it's not in front of the AI right now, the AI doesn't know it.

The assistant's desk

Picture an assistant working at a desk. Whatever papers are spread out on that desk right now — that's what they can use to answer you. The note you just handed them, the earlier notes from this same meeting, the standing instructions pinned to the corner.

What's not on the desk might as well not exist. A conversation you had with them last Tuesday? Filed away in another room, out of sight. A fact you assume "everybody knows" but never actually said? Not on the desk. The AI, like that assistant, answers from what's on the desk — nothing more.

This is why being clear and complete in your message matters so much. You're not just asking a question; you're laying papers on the desk. The more relevant and tidy those papers, the better the answer.

Why a vague question gets a vague answer

People often blame the AI for a weak reply when the real issue is an empty desk. Watch the difference:

"Write me a reply."

The desk is nearly bare. Reply to whom? About what? In what tone? The AI has to guess, and guesses are hit-or-miss.

"Write a short, warm reply to this customer who's upset their order is late. We're a small bakery, we want to apologize and offer a free pastry next visit. Keep it under four sentences."

Now the desk is covered in exactly the right papers — who, what, tone, length, the offer. Same AI, wildly better result. Nothing changed about the tool. You just gave its desk something to work with.

Why this matters for your business

Almost every disappointing AI moment traces back to context — either too little of it, or the wrong kind. Once you start thinking "what's on the desk right now?" you'll diagnose problems instantly:

• Bad answer? The desk was probably missing something the AI needed.
• It "forgot" a detail? That detail likely fell off the desk — more on that next lesson.
• It made something up? Often it had a gap on the desk and filled it with a guess.

The whole craft of working with AI is really just the craft of managing its desk well.

Your turn

Take a question you'd normally type into an AI and ask yourself: if I handed this to a brand-new assistant who knows nothing about me, could they answer it well? If not, add the missing papers — the who, the what, the tone — right into your message. Try it once today and feel the difference.

🔦 You now know context is just "what's on the desk." Next: that desk has edges, and when papers fall off, you get the forgetting we started with.