🔥 Why This Matters (Read This First)

Most of us work like this:

• We read articles

• We skim documentation

• We take scattered notes

• We try to piece things together

• Then we start building training

The problem:

👉 We lose most of what we learn

👉 We repeat research every time

👉 We manually connect everything

💡 What This System Does Instead

This system:

• Turns your research into structured knowledge

• Connects ideas automatically

• Lets you ask questions against your own content

• Saves every answer so it builds over time

• 
  

🧠 The Big Idea

You are not just taking notes.

You are building a system that gets smarter every time you use it.

🔁 How It Works

• You add articles, docs, notes

• AI turns them into structured knowledge

• You ask questions

• AI generates answers based on YOUR sources

• Those answers get saved and improve future answers

👉 It gets smarter every time you use it

🆚 How This Is Different from NotebookLM

NotebookLM:

• You upload documents

• You ask questions

• You get answers

This system:

• Builds a structured knowledge base

• Connects ideas across everything

• Saves outputs permanently

• Improves over time

👉 It’s not just answering questions

👉 It’s building a brain

🧱 Phase 1: Setup Your Second Brain

🎯 Goal

By the end of this section, you will:

• Have Obsidian installed

• Have Codex Desktop installed

• Have a vault (folder) created

• Have Codex connected to that vault

• Be ready to start building your second brain

🧠 What You’re Doing (Simple)

You are creating:

👉 A folder (your brain)

👉 Obsidian (to view it)

👉 Codex (to build and think with it)

Step 1: Install Obsidian

• Go to: https://obsidian.md

• Click Download

• Install it like any normal app

• Open Obsidian

Step 2: Create Your Vault (Your “Brain”)

This is the most important step.

Option A (Recommended – simplest)

• On your computer, create a new folder on your Desktop:

Obsidian Vault

• Open Obsidian

• Click “Open folder as vault”

• Select your Obsidian Vault folder

  
  

✅ You now have your vault

🧠 What just happened

That folder is now:

👉 your entire knowledge system

Everything lives there:

• notes

• sources

• AI outputs

Step 3: Install Codex Desktop

• Download Codex Desktop

• Install it

• Open Codex

Step 4: Connect Codex to Your Vault

When Codex opens:

• Click "Add new project"

• Select your:

Obsidian Vault

✅ Now Codex can:

• read your files

• create new files

• update your knowledge base

  
  

🧠 What just happened

You connected:

• Codex → your vault (brain)

• Obsidian → your vault (viewer)

So now:

👉 Codex writes

👉 Obsidian displays

⚙️ Before You Start Using Codex

Before running your first prompt, set Codex to:

• Model: Best available 

• Effort: Medium 

• Mode: Planning (Shift+Tab) 

🧠 Why This Matters

Planning Mode helps Codex follow multi-step instructions more reliably.

Step 5: Initialize Your Knowledge Base

Now we let Codex build your structure for you.

👉 Paste this into Codex:

You are operating inside an Obsidian vault.

Create the following folder structure

00_inbox/

01_raw/
01_raw/articles/
01_raw/notes/
01_raw/images/

02_sources/

03_concepts/

04_entities/

05_maps/

06_outputs/
06_outputs/answers/
06_outputs/slides/
06_outputs/visuals/

07_system/

Create the folders only. Do not overwrite existing content.

When complete, confirm:
"Structure created successfully."

Step 6: Add System Rules

👉 Paste this into Codex:

Create the following files inside 07_system/.

---

File: style_guide.md

Content:

# Style Guide

- Use clear, simple language
- Use headings
- Use bullet points
- Link notes using [[ ]]

---

File: ontology.md

Content:

# Note Types

- Source
- Concept
- Entity
- Map
- Output

---

File: naming_rules.md

Content:

# Naming Rules

- source_<title>
- map_<project>
- answer_<topic>

---

File: maintenance_checklist.md

Content:

# Maintenance

- Find gaps
- Identify duplicates
- Improve connections

Do not overwrite existing files.

Confirm when complete

✅ You’re Done with Setup

At this point, you have:

• A vault (your brain)

• A structure for organizing knowledge

• Codex connected and ready

• Obsidian showing everything

🧠 What You Just Built

A system where AI can read, write, and improve your knowledge over time.

👉 Next Step

Now you’re ready to:

👉 Add sources

👉 Process them

👉 Start asking questions

🧠 Phase 2: Add and Process Research

Step 7: Add Your First Sources

You can add content in two ways to 01_Raw folder.

• Use the Obsidian Web Clipper (recommended)

Save all content into:

01_raw/articles

🔧 (Recommended) Set Up Web Clipper Location

To make this automatic:

• Click the Obsidian Web Clipper icon in your browser

• Click the ⚙️ Settings icon

• Go to Templates

• Click on Default

• Find “Note location”

• Change it to:

01_raw/articles

✅ Now every time you clip an article, it will go directly into the correct folder.

🧠 Why This Matters

This ensures:

• all your sources go to the right place

• Codex can process them easily

• your system stays clean and consistent

🧠 When to Use Each Raw Folder

Most of the time, you will use articles.

Use the other folders only in specific situations:

📄 articles (default)

Use this for:

• web pages

• KB articles

• SOPs

• internal documentation

👉 If you’re unsure, put it here

📝 notes

Use this for:

• your own notes

• meeting notes

• SME conversations

• copied Slack messages

👉 Anything created by you or your team

🖼️ images

Use this for:

• screenshots

• diagrams

• visual references

📚 papers (rare)

Use this for:

• research papers

• long-form reports

  
  

💻 repos (rare)

Use this for:

• code repositories

• technical documentation tied to code

💡 Simple Rule

• Web content → articles

• Your thinking → notes

• Visuals → images

Everything else is optional.

💡 Tip

Don’t overthink what you add.

Start with:

• 3 to 10 articles

• anything relevant to your course or topic

Step 8: Process Sources

👉 Paste into Codex:

Process all new files in 01_raw/articles/.

For each file:
- Create a source note in 02_sources/
- Summarize the content
- Extract key ideas
- Identify entities
- Identify related concepts

Update source_index.md with all sources.

Do not reprocess existing files.

Confirm when complete:
"Sources processed successfully."

🧠 Phase 3: Build Understanding

Step 9: Generate Concepts

👉 Paste into Codex:

Analyze all source notes in 02_sources/.

Identify recurring ideas that appear across multiple sources.

Create concept notes in 03_concepts/ with:
- summary
- key ideas
- supporting sources

Only create concepts that appear in at least 2 sources.

Update concept_index.md.

Confirm:
"Concepts generated successfully."

Step 10: Generate Entities

👉 Paste into Codex:

Analyze all source notes in 02_sources/ and concept notes in 03_concepts/.

Identify important named entities such as:
- tools
- systems
- teams
- policies
- products

Create entity notes in 04_entities/ only when they are clearly important or appear more than once.

Each entity note should include:
- what it is
- why it matters
- related concepts
- related sources

Update entity_index.md.

Confirm:
"Entities generated successfully."

🧠 What Are Entities?

Entities are the real-world things in your knowledge base.

Examples:

• tools

• systems

• teams

• policies

• products

💡 Why They Matter

Entities help Codex connect ideas to real-world context, which makes answers more grounded and useful.

⚠️ Keep It Simple

You do not need to think too hard about this step.

Codex will handle most of it automatically.

🧠 Phase 4: Start a Project (CRITICAL STEP)

⚠️ This example is written for instructional design, but this applies to any role or project.

🧠 What You’re Doing

You are creating a Map

A map is:

• your project workspace

• the boundary for AI thinking

• how you keep everything organized

💡 Examples by Role

Instructional Design

map_bird_course
map_returns_training

Customer Service / Ops

map_refund_process
map_escalation_handling

Leadership / Strategy

map_customer_experience_strategy

🧠 The Rule

Every project gets its own map.

Step 11: Create Your Map

👉 Paste into Codex:

Create a new map file in 05_maps/.

Name it:
map_<your_project_name>.md

Include sections:
- Overview
- Goals
- Key Concepts
- Source Material
- Open Questions
- Outputs

Confirm:
"Map created successfully."

Step 12: Create Output Folder for This Project

👉 Paste into Codex:

Create the following folders:

06_outputs/<project_name>/
06_outputs/<project_name>

🧠 Why this step matters

This keeps all your answers organized from the start, instead of cleaning things up later.

Step 13: Strengthen Your Map (Recommended)

👉 Paste into Codex:

Review all source notes in 02_sources/.

Identify which ones are relevant to [[map_<your_project_name>]].

Update the map to include:
- relevant sources
- related concepts

Do not remove existing content.

Confirm:
"Map updated with relevant knowledge."

This step strengthens your map by explicitly connecting it to relevant sources and concepts. While the system can work without this, doing this improves the quality and focus of your answers.

🧠 Why Maps Are So Important

Without a map

The AI will:

• search across your entire vault

• mix unrelated topics

• pull in irrelevant information

👉 Your answers may feel:

• too broad

• unfocused

• less useful

With a map

The AI will:

• focus only on knowledge connected to that map

• ignore unrelated content

• stay aligned to your project

👉 Your answers become:

• focused

• relevant

• immediately usable

💡 Simple way to think about it

• Without a map = searching your entire brain

• With a map = working inside a specific workspace

🧠 One-line takeaway

A map controls what the AI thinks with, so your answers stay focused on your project.

🧠 Phase 5: Ask Questions (Where the Magic Happens)

Step 14: Generate Your First Answer

👉 Paste into Codex:

You are answering a question using the knowledge base.

Using [[map_<course_name>]], answer:

What are the most important things agents need to know and do?

Create a file in:
06_outputs/<course_name>

🧠 What’s Happening Behind the Scenes

When you ask a question, Codex:

• Reads sources (facts)

• Uses concepts (patterns)

• References entities (context)

• Follows the map (scope)

Then:

👉 writes a structured answer back into your vault

🔥 High-Value Questions You Could Use

Training Design

Using [[map_<course_name>]], what should agents learn?

Mistakes

Using [[map_<course_name>]], what mistakes will agents make?

Scenarios

Using [[map_<course_name>]], what scenarios should agents practice?

Learning Objectives

Using [[map_<course_name>]], what should the learning objectives be?

Gaps

Using [[map_<course_name>]], what is missing or unclear?

🚀 Streamline the Process Once You Understand It

Now that you understand how the system works, you do not need to repeat every step manually for every new project.

Once your vault is set up, your workflow can be much simpler:

1. Add source material to 01_raw/articles/

2. Choose a project name

3. Paste one prompt into Codex

4. Let Codex prepare everything for you

5. Start asking questions

🧠 What This Streamlined Workflow Does

Instead of manually:

• processing sources

• generating concepts

• creating a map

• creating output folders

• connecting knowledge to the map

You can have Codex do all of it in one pass.

This is the fastest way to go from:

👉 raw articles

to

👉 a project that is ready for focused question-answering

✅ Best Time to Use This

Use this workflow when:

• your vault is already set up

• you understand the basic structure

• you are starting a new project

• the raw content you added is mostly related to one topic

⚠️ Important Note

This works best when the content you add to 01_raw/articles/ is mostly for the same project.

If you add a mix of unrelated content, Codex can still process it, but the project map may be less focused.

Best practice:

Add a batch of related source material for one project, then run the prompt.

🔧 Codex Mega Prompt for Future Projects

Paste this into Codex and replace the placeholder before running it.

🔧 How to Use This Prompt

1. Replace this line at the top:

PROJECT NAME: <project_name>

with your project name, for example:

PROJECT NAME: bird_course

1. Do NOT manually update the rest of the prompt.

Codex will use this project name throughout the process.

⚠️ Quick Check

After the prompt runs, quickly confirm:

• your map is named correctly (e.g., map_bird_course)

• your output folder is correct (e.g., 06_outputs/bird_course/)

If anything looks off, you can fix it manually or rerun the prompt.

🧠 What This Prompt Does

This prompt will:

• create your project map

• create your output folders

• process new sources

• generate concepts

• connect everything together

👉 So your project is ready for question-answering immediately

🚀 Paste This Into Codex

You are operating inside an Obsidian knowledge base.

Your job is to take newly added raw source material and prepare this project so it is ready for focused question-answering.

PROJECT NAME: <project_name>
PROJECT MAP NAME: map_<project_name>

Always reuse the PROJECT NAME exactly as defined above. Do not invent variations.

Use this exact project name consistently in filenames, folders, and links.

Your goal is to complete all of the following steps without overwriting existing user content.

---

STEP 1: CREATE OR VERIFY PROJECT MAP

In 05_maps/, create this file if it does not already exist:

map_<project_name>.md

If it already exists, update it without removing existing useful content.

The map should include these sections:

- Overview
- Goals
- Key Concepts
- Source Material
- Open Questions
- Outputs

---

STEP 2: CREATE OR VERIFY PROJECT OUTPUT FOLDERS

Create these folders if they do not already exist:

06_outputs/<project_name>/
06_outputs/<project_name>/answers/

Do not overwrite or delete existing files.

---

STEP 3: PROCESS NEW RAW SOURCES

Review all files in:

01_raw/articles/

Identify any raw files that do not yet have corresponding source notes in 02_sources/.

For each unprocessed raw file:

- Create a source note in 02_sources/
- Summarize the content
- Extract key ideas
- Identify entities
- Identify related concepts

Use clear source note formatting and do not duplicate already processed work.

Update source_index.md with any newly created source notes.

---

STEP 4: GENERATE OR UPDATE CONCEPTS

Review all source notes in 02_sources/.

Identify recurring ideas that appear across multiple relevant sources.

Create or update concept notes in 03_concepts/ with:

- summary
- key ideas
- supporting sources

Only create concepts that are meaningful and supported by at least 2 relevant sources.

Update concept_index.md.

---

STEP 5: CONNECT PROJECT MAP TO RELEVANT KNOWLEDGE

Review all source notes and concept notes.

Identify which sources and concepts are relevant to [[map_<project_name>]].

Update the map to include:

- relevant sources in the Source Material section
- relevant concepts in the Key Concepts section
- useful open questions based on gaps or ambiguity
- outputs section placeholders if needed

Do not remove existing useful links.

---

STEP 6: KEEP PROJECT SCOPE CLEAN

When deciding what belongs in [[map_<project_name>

💡Example

If your project is called:

bird_course

Then replace:

<project_name>

with:

bird_course

This will allow Codex to prepare:

• [[map_bird_course]]

• 06_outputs/bird_course/answers/

• relevant sources

• relevant concepts

• a project structure ready for questions

✅ What You Do Next

Once Codex finishes, you can immediately ask questions like:

Using [[map_bird_course]], what are the most important things agents need to know and do

Using [[map_bird_course]], what mistakes are agents most likely to make

Using [[map_bird_course]], what scenarios should agents practice

🧠 Final Takeaway

Once your system is set up, most projects follow the same flow:

👉 add sources

👉 run one prompt

👉 ask better questions

🧠 What This Actually Is

You are not:

❌ asking AI random questions

You are:

✅ building a system that stores knowledge, connects ideas, and answers better over time

🧠 One-Line Value

This turns your research into a system that helps you design training faster and smarter.

If you have any questions or ideas on how to improve this, feel free to reach out. This is an evolving system and your feedback helps make it better.