MODULE_00 Setup

The Discovery

~30 min active work 3 interactive prompts 4 artifacts No prereqs

Build the infrastructure that makes everything else work - three files that travel with you through every module.

Let's get into it now.

But before we teach you anything, you're going to build the infrastructure that makes everything else work.

Three files. They travel with you through every module. The AI reads them before every lesson, which is why Module 6 feels different for a freelance copywriter than for a startup founder. Same prompt. Different experience. Because the AI knows who it's talking to.

This isn't the "optimal" way to do it. This is how you should approach every single project. This is the single biggest difference between people who get value from AI and people who keep starting over in blank chats.

core-files/ ARTIFACT
my-context.json - who you are, what you do, your goals
my-learning-style.json - how you learn best
my-knowledge.md - what you've mastered (grows with each module)
prompt-learning-extraction.md - reusable prompt you'll run after every module
Plus the Learning Extraction Prompt, a reusable prompt you'll run after every module to capture what you learned.

Solving the Amnesia Problem

Let's start with this question: why does AI give you generic output?

It's not because the model is dumb. It's because you're a stranger.

Think about it this way... You hire a consultant. Day one, they walk in knowing nothing about you. You burn the first hour explaining your business, your customers, your problems, what you've tried, what flopped. Only then can they say anything worth hearing.

Now imagine doing that every single time you meet. Every conversation starts from zero. No memory. No notes. No file on you.

That's what happens every time you open a blank chat window on ChatGPT (or any provider). You're hiring a brilliant consultant with amnesia. And then getting frustrated when they hand you something that sounds like it was written for nobody in particular.

We're not going to fix it with a better prompt... but with much better context.

Context is everything the AI knows about you before you ask your question. Who you are. What you do. How you think. What you've already tried. When the AI has this, the output shifts. Not because the model got smarter. Because it finally has enough information to give you something specific instead of something safe.

That's what this module builds. Three files that solve the amnesia problem for the rest of this course.

You should remember this image of the consultant or co-founder... It makes things easier when we're talking about AI. When the first version of ChatGPT launched, we couldn't use this analogy. Things have changed, and more than ever before, AI is very similar to a consultant.

Let's take a look at the files.

File 1: my-context.json

It's everything about you. Your professional situation. Your work, your clients, your goals, your problems. When the AI reads this, "write me an email" becomes "write an email from a freelance UX designer to a SaaS startup about a project delay." Specificity changes everything.

File 2: my-learning-style.json

How you actually learn. Not what you think you prefer. We'll test it. Some people want analogies. Some want step-by-step procedures. Some want the technical mechanics laid bare. The AI adapts its teaching to match.

File 3: my-knowledge.md

What you've mastered so far. This starts almost empty. By Module 15, it's the most valuable document in your collection: a complete record of everything you know about AI, written in your own words. Every module adds to it. Nothing gets lost. As the AI landscape evolves and we have to study new concepts, this file will become VERY handy.

These three files get attached to your course project. Every prompt you run in this course reads them automatically. That's why the same Module 6 prompt teaches writing differently to a copywriter than to a project manager. The prompt is identical. The context is not.

One more thing you'll notice: every prompt in this course uses XML tags. They look like this: <role>, <context>, <instructions>. Don't worry about understanding XML yet. Module 3 covers why these tags matter (they create attention boundaries that change how the model processes your prompt). For now, just know that angle brackets around a word are a structural signal telling the AI how to treat the content inside it. Think of them as labeled containers. You'll get comfortable with them fast.

Your Folder Structure

Before running any prompts, create this folder structure on your computer. This is where you'll save everything you build throughout the course.

ai-fundamentals/
ai-fundamentals/
├── core-files/
│   ├── my-context.json
│   ├── my-learning-style.json
│   └── my-knowledge.md
├── prompts/
│   └── learning-extraction-prompt.md
├── module-outputs/
│   ├── module-00-discovery/
│   ├── module-01-rules/
│   ├── module-02-tokens-prediction/
│   ├── module-03-attention-structure/
│   ├── module-04-context-engineering/
│   ├── module-05-systems/
│   ├── ...
│   └── module-15-open-source/
└── experiments/

You don't need every subfolder right now. Just set up core-files/, prompts/, and module-outputs/module-00-discovery/. Build the rest as you go.

Create Your Course Project

Create a new Project in your favorite LLM (I'd recommend Claude just because they have very solid models and the best features overall). Name it something like AI Fundamentals Course.

This project stays open for the entire course. After you build each of your three core files below, attach them to this project. Every module prompt you run from here on reads those files automatically.

If you're using Claude: Click Projects in the sidebar → New Project → name it → you'll add files to the Knowledge section after creating them. On ChatGPT, Gemini, Grok or any AI app, it's very similar... shouldn't be a problem for you to find it.

Build Your Three Files

01

Step One

Build Your Context File

Prompt: prompt-0a-context-discovery.md

Time: 10-15 minutes


How to run it:

1. Open a new conversation in your course project
2. Copy the entire contents of prompt-0a-context-discovery.md and paste it in
3. The AI will interview you. Answer each question as it comes
4. After 8-15 questions, it produces your my-context.json file


What you should see: The AI asks you questions one at a time about your work, your AI experience, and your goals. It should feel like a conversation, not a form. It will push back if your answers are vague. That's by design. The more specific you are, the better every future module works for you.


When it's done: Copy the output, save as my-context.json in your core-files/ folder, and attach to your course project.

I need you to dedicate as much time as you can here. The goal is to update this file when your situation changes. When you talk with Claude about important aspects of your life, you can prompt it to update your my-context.json file.
02

Step Two

Discover Your Learning Style

Prompt: prompt-0b-learning-style-discovery.md

Time: 10-15 minutes


How to run it:

1. Start a new conversation in your course project (so your my-context.json is already loaded)
2. Copy the entire contents of prompt-0b-learning-style-discovery.md and paste it in
3. The AI runs you through 5 short tests. Each one presents options or a small challenge
4. After all tests, it produces your my-learning-style.json file


What you should see: Five tests, one at a time. The AI is watching HOW you respond, not grading WHAT you know. There are no wrong answers. It won't label you during the tests. All interpretation happens at the end when it produces your profile.


When it's done: Copy the output, save as my-learning-style.json in your core-files/ folder, and attach to your course project.

Here you want to make sure that the AI understands DEEPLY how you think. This course will be life changing if you nail this step. Don't rush it.
03

Step Three

Initialize Your Knowledge File

Template: template-my-knowledge.md

Time: 2 minutes


This one isn't interactive. Copy the entire contents of template-my-knowledge.md and save it directly:


Save as my-knowledge.md in your core-files/ folder, then attach to your course project.


This file starts mostly empty. By Module 15, it's a complete record of everything you've learned, written in your own words, updated after every module. It becomes the most valuable document in your collection.

04

Step Four

Save Your Learning Extraction Prompt

Prompt: prompt-learning-extraction.md

Time: 1 minute


This prompt is your closing ritual for the rest of the course. You'll run it after every module. It reviews what just happened, extracts what you actually learned (not what was presented), and produces an update for your knowledge file.


Save prompt-learning-extraction.md to your prompts/ folder. You'll use it 15+ times.


How to use it (after any module):

1. Finish the module's interactive prompt
2. In the same conversation, paste the Learning Extraction Prompt (so the AI can review everything that just happened)
3. The AI reviews the session and produces two outputs:

Knowledge File Update: paste this under the current module heading in my-knowledge.md
Session Notes: save as session-notes.md in the current module's output folder

Context Discovery

Paste this into a new conversation in your course project. The AI will interview you and produce your my-context.json file.

prompt-0a-context-discovery.xml
<identity>
You are a professional profiler who spent 15 years building intelligence-grade dossiers on executives, founders, and specialists - the kind of document that lets a strategist walk into a room and operate as if they've known this person for years. You don't collect categories; you extract operating patterns. When someone says "I'm a marketing director," you hear static - the signal is in HOW they make decisions, WHERE their attention goes under pressure, and WHAT they actually produce versus what their title implies. You are also a meticulous process manager who follows phases sequentially, never advances until signal quality passes the gate, and treats every vague answer as a data gap that will cascade into downstream failure.
</identity>

<prime_directive>
Produce a JSON profile so granular and behaviorally specific that any AI system reading it would calibrate its tone, depth, examples, pacing, vocabulary, and recommendations to THIS person - not a demographic, not a job title, not an archetype. The profile must capture operating patterns, not surface attributes.
</prime_directive>

<workflow_overview>
Phase 1: Professional Reality - who they are, what their work actually looks like, how they operate
Phase 2: Cognitive & Communication Patterns - how they think, decide, communicate, and process information
Phase 3: AI Relationship - honest current state with AI tools, including trust, skill, failure history, and mental models
Phase 4: Goals & Friction - what they specifically want, what's blocking them, and the gap between current and desired state
Phase 5: The Anchor - the ONE thing that would change their work the most
Phase 6: Environment & Constraints - tools, workflows, team dynamics, and real-world constraints
Phase 7: Synthesis - produce the complete JSON profile
</workflow_overview>

<interaction_protocol>
- Conversational, not clinical. This is a good conversation, not a form.
- ONE question at a time. Never stack questions. Never ask compound questions.
- Mirror their language register.
- When they say something interesting or unexpected, follow that thread.
- Never judge their AI skill level.
- Keep the total interview to 12-20 exchanges across all phases.
- When you get a vague answer, don't accept it: "Give me a specific example from this week."
- Do NOT produce the JSON profile until ALL six interview phases are complete.
</interaction_protocol>

<start>
Begin Phase 1. Start with: "Forget your job title for a second. Walk me through what last week actually looked like - what were you working on, who was it for, and what did you actually produce or deliver?"
</start>

Learning Style Discovery

Run this after saving your my-context.json. Start a new conversation in your course project (so your context file is loaded automatically), then paste this in. The AI runs you through 8 behavioral tests and produces your my-learning-style.json file.

prompt-0b-learning-style-discovery.xml

      

My Knowledge Base

Copy this template and save it as my-knowledge.md in your core-files/ folder, then attach to your course project. This file grows with you through every module.

template-my-knowledge.md
# My AI Knowledge Base

This file tracks everything I've mastered through the AI Fundamentals course. Updated after every module using the Learning Extraction Prompt.

Last updated: [date]

---

## Module 0: The Discovery
- Completed: [date]
- Key insight: [filled after extraction]

## Module 1: Rules of Engagement
- Completed:
- Key insight:

## Module 2: What's Actually Happening When You Prompt
- Completed:
- Key insight:

## Module 3: Attention, Structure and Prompt Architecture
- Completed:
- Key insight:

## Module 4: Context Engineering
- Completed:
- Key insight:

## Module 5: From Conversations to Systems
- Completed:
- Key insight:

## Module 6: AI for Writing and Content
- Completed:
- Key insight:

## Module 7: AI for Images and Visual Content
- Completed:
- Key insight:

## Module 8: AI for Video and Audio
- Completed:
- Key insight:

## Module 9: AI for Research and Analysis
- Completed:
- Key insight:

## Module 10: AI for Code and Building
- Completed:
- Key insight:

## Module 11: MCPs (Giving AI Hands)
- Completed:
- Key insight:

## Module 12: AI Agents (When and When Not)
- Completed:
- Key insight:

## Module 13: Automations and Workflows
- Completed:
- Key insight:

## Module 14: Building Your AI Stack
- Completed:
- Key insight:

## Module 15: Open Source and Running Your Own Models
- Completed:
- Key insight:

---

## Skills and Techniques Library
(Accumulated from all modules: specific techniques you've learned and can reuse)

## Open Questions
(Things you still want to understand deeper. Revisit these as the course progresses)

Your Closing Ritual

Save this to your prompts/ folder - you'll use it after every module. After completing a module's interactive prompt, paste this into the same conversation so the AI can review what just happened and extract what you actually learned.

learning-extraction-prompt.xml

      

You Now Have Four Files

Run the Learning Extraction Prompt now. Even for this setup module. It's practice for the ritual you'll repeat 15 more times.

my-context.json: who you are (in core-files/, attached to your project)

my-learning-style.json: how you learn (in core-files/, attached to your project)

my-knowledge.md: what you know, initialized and ready to grow (in core-files/, attached to your project)

prompt-learning-extraction.md: your closing ritual for every module (in prompts/)

Run the Learning Extraction Prompt, update Module 0's entry in my-knowledge.md

Save any session notes to module-outputs/module-00-discovery/

Next: Module 1: Rules of Engagement.
The three operating rules that make every module after this one actually work. Skip them and you'll get 20% of the value. Follow them and the course compounds.