PUBLIC_KEY_PEM before distributing.
Context is all of the information you feed the machine at once. Not just what you ask — everything surrounding the ask. Bigger, better context gets better results. This module teaches you how to build it deliberately.
By the end of this module, you will have:
In Module 2 you saw how word choice changes output. Context is that idea scaled up. Every time you send a message, everything in the conversation goes in together — the whole thing, from the beginning. The machine is not remembering. It's processing everything at once, every single time you hit send.
When you open a new chat and hit send, the machine is not remembering your prior messages. It is resending everything in the chat window — all of it, all at once — every single time you press send. The conversation is not a back-and-forth in the way you have conversations with people. It's a growing block of text that gets reprocessed from scratch with every message.
That means you can front-load the conversation. Instead of iterating through twenty exchanges to get to a good result, you build the full picture once and send it all at once. One message. One result. That's the goal.
When you front-load a conversation with everything the tool needs, you get more accurate results because the context doesn't degrade over multiple exchanges. Each back-and-forth message dilutes your original intent with the tool's interpretations. One well-constructed message gives you the clearest, most controlled output possible.
Every context you build has three parts. These are the three types of information to include when you want the best possible result:
Sets the persona, expertise level, and communication style. The tool takes on this role for everything that follows.
Your raw material — documents, data, background, situation. What the tool is actually operating on.
Specific, concrete task instructions. What to produce, how to structure it, what to include or exclude.
"You want to aim to one-shot it. Go in, paste in one message and get the perfect result every time." Context-building is how you get there. Not because you get lucky — because you gave the tool everything it needed before it started.
The Context Builder is a simple standalone tool that helps you assemble the three parts of a context into a single block, ready to paste into any AI tool. It runs in your browser, stores nothing, and sends nothing anywhere.
Download it now. Keep it — you'll use it in every module from here on. It does not need the internet to work.
Standalone HTML tool — opens in any browser, no internet required.
The Context Builder is bundled inside this module file. It is not on the website and not available anywhere else. If you lose it, come back to this page and download it again.
Each exercise uses the same three-part structure. What changes is how much is provided versus how much you write. In Exercise 1, you fill in only the content — the role and instructions are given to you. In Exercise 2, the content is provided and you write the role and instructions. By Exercise 3, all three parts are entirely yours.
Use the Context Builder to assemble each one. Paste the assembled context into your AI tool. Copy the full response. You'll paste both into the submission fields below each exercise.
Copy the role and instructions below into the Context Builder. Then fill in the content section with something you actually want to work on — a situation, document, problem, or question from your own work or life. Assemble and send.
This is the middle section — your actual situation, document, or problem. Write a paragraph or two. It can be anything: a decision you're weighing, background on a project, a scenario you want advice on. Keep it real and specific.
This time the middle content is provided. You write the role and the instructions. Think carefully about what kind of advisor or expert would produce the best result for this content — and what exactly you want them to do with it.
Open the Context Builder. Write a role that sets up the best possible advisor for this situation. Write instructions that direct the tool to produce something genuinely useful for this meeting. Then assemble and send.
Choose something you actually care about — a real challenge, project, or decision in your work or life. Write all three parts of the context yourself: the role, the content, and the instructions. Assemble in the Context Builder. Send it. See what you get.
This is the exercise that stays with you. The tool you built in the first two exercises, you now use on your own terms.
Choose your topic — something with real stakes for you, not a throwaway test.
Write the role — who is the ideal expert for this? What do they know and how do they communicate?
The role sets the lens through which the tool processes everything else. Think about what kind of expert would give you the best answer for your specific situation. Some examples:
Write the content — the full picture. Background, constraints, what's at stake. The more real information, the better the result.
Write the instructions — exactly what you want back. Specific format, specific focus, specific output.
Assemble in the Context Builder. Paste into your AI tool. Copy the full response.
Your work is saved locally. Your work has been saved. Keep this file for offline reference.