How to Set Up Your Human Stack (in 30 Minutes, from Anywhere)
What you'll need: A computer (Mac or Windows), about 30 minutes, and any text editor you already have. No coding. No new apps to install.
You've probably got a hundred important things in a hundred random places.
Notes in Apple Notes. Drafts in Google Docs. Half-finished AI conversations across three different tools. Outputs from those conversations sitting in your downloads folder. Frameworks you've built scribbled on the back of meeting notes. A decision you made six weeks ago that you can't quite remember why.
You know what you think. You just have nowhere to put it.
And now AI is making it worse, not better. Every session produces more outputs. More drafts. More ideas. More things to file and forget. The tool was meant to help you think, and it's flooding the room with thinking you don't know how to organise.
The Human Stack is the structured home that fixes this. Five folders. Three starter documents. A place for the thinking you already do, the work you already produce, the decisions you already make. As a side effect (a big one), every AI you talk to starts to know who you are, how you work, and what you're focused on this week. They stop guessing. You stop re-explaining yourself.
This is the operating manual. By the end of it you'll have a real Human Stack on your real computer, ready to use with whatever AI you prefer.
What You're Building
A folder on your computer called [Your Name]'s Vault. Inside it, five subfolders:
Colin's Vault/
1 Identity/
2 Systems/
3 Projects/
4 Pulse/
5 Thinking/
Three of those folders get a starter document each. The other two stay empty until you have something to put in them. That is the entire setup.
Five folders is the architecture. Three starter documents is the work. The empty folders are not a bug. They are seats reserved for the work that grows over time.
The five layers are organised by how often they change. Identity (who you are) changes yearly. Systems (how you work) changes quarterly. Projects monthly. Pulse weekly. Thinking daily. Build from the bottom up. The base is the most permanent and the most important. The top is the most fluid and the most fresh.
This is the architecture I teach in full at colinscotland.com/human-stack. What you are about to set up is the practical version: the same five layers, on your real computer, ready to use today.
There is no one true folder structure. The five-folder version above is mine. You can use a simpler three-folder version, or PARA-style, or invent your own. The architecture (Identity → Systems → Projects → Pulse → Thinking) is what matters. Where you put it on your computer is up to you.
Step 1: Make Your Vault (2 minutes)
Open Finder (Mac) or File Explorer (Windows).
Pick where you want this to live. Documents is a good default. iCloud Drive or Google Drive work fine if you want it to sync across machines. Desktop is okay if you want to see it every day. Anywhere on your computer is fine.
Create a new folder. Call it [Your Name]'s Vault. Use your real first name. Mine is Colin's Vault.
Open that folder. Inside it, create five new folders. Name them exactly like this:
1 Identity
2 Systems
3 Projects
4 Pulse
5 Thinking
The number prefixes keep them in the right order on screen. Without them your computer sorts alphabetically and the layers end up in the wrong sequence.
You'll know it's working when you open the Vault folder and see five subfolders in this order: Identity, Systems, Projects, Pulse, Thinking.
A note for the file-allergic. If creating folders on your computer feels like too much, you can do this in Apple Notes, Google Docs, Notion, or any app you already use. Make a section or notebook called Vault, and create five sub-pages with the same names. The architecture is the content, not the format. We will cover how to give it to your AI either way in Step 6.
But folders on your computer is the simpler path for almost everyone. You already know how to make folders. You don't need to learn anything new.
Step 2: Write Your BRIEF (10 minutes)
Goes in: 1 Identity
Your BRIEF is the master document about you. Voice, values, expertise, the way you work, what you never want to see in your output. It is the single most important document in the whole system. Without it, every AI you talk to is guessing. With it, every AI knows who you are before you say a word.
You have two ways to write yours.
Easy mode. I built a free tool that walks you through the questions and produces your BRIEF for you. It runs on ChatGPT, but the output is a plain document you can save anywhere. Open the Amplify OS Builder, answer its questions (about 10 minutes), copy the result, and save it as a new file called BRIEF.md (or BRIEF.txt if .md feels strange) inside your 1 Identity folder.
Manual mode. Open a new document in any text editor. TextEdit on Mac, Notepad on Windows, or whatever you use. Save it as BRIEF.md (or .txt) in your 1 Identity folder. Then write answers to these questions:
- Who are you? Name, role, what you actually do.
- What is your core thesis? The one big idea your work is built on.
- What are your core values? Three to five.
- How would you describe your voice in five words?
- What words do you never use? (Mine: never "delve", never "explore", never "leverage".)
- What are the frameworks or methodologies that define how you work?
- What kind of output do you never want from an AI?
- What are two or three short examples of your actual writing voice?
Be specific. "I'm a coach" is useless. "I'm a leadership coach for founders of companies between 10 and 50 people, and I work primarily through six-month engagements that combine weekly 1:1s with quarterly intensives" is useful. The more specific, the better the AI gets.
You'll know it's working when you open the file and read it back, and it sounds like you wrote it about yourself. Not a marketing bio. The real version.
Step 3: Write Your INDEX (5 minutes)
Goes in: 2 Systems
Your INDEX is the routing table. It tells you (and any AI you give it to) what to load for what kind of work. Think of it as a map of your own working world: when I do this kind of thing, here is what I reach for.
INDEX scales with whatever you actually run. A solo creator routes by task type (writing, strategy, client work). A coach routes by client or by programme. A business owner routes by functional area (sales, marketing, operations, finance, hiring). A team routes by team member or by initiative. Same architecture, different categories. You decide what the categories are.
Open a new document. Save it as INDEX.md in your 2 Systems folder. Here are two starter templates. Pick whichever fits how you actually work, and edit it.
For an individual or solo creator:
# Systems Index
Routes task types to the files, frameworks, and tools needed.
## Writing
Load: BRIEF.md (voice section), examples of past writing
Apply: Direct, simple, real. No fluff. No buzzwords.
## Strategy and Planning
Load: BRIEF.md, current project goals
Apply: My usual decision-making approach (intuition first, logic second)
## Client Work
Load: BRIEF.md, the relevant project file in 3 Projects/
Apply: Match the client's voice, not mine
For a business owner or team lead:
# Systems Index
Routes functional areas to the files, processes, and people involved.
## Sales
Load: BRIEF.md, our sales playbook, current pipeline
People: who owns what
Apply: Our actual qualification process, not a generic one
## Marketing
Load: BRIEF.md, brand voice doc, content calendar
Apply: Our positioning, our audience, our messaging rules
## Operations
Load: BRIEF.md, current SOPs, ongoing process docs
Apply: How we actually run things, not how a textbook would
Three categories is plenty to start. The rule is: the kinds of work that take up most of your week. You can add more as you go. INDEX is meant to grow.
If you have your own frameworks (a coaching model, a sales process, a methodology you teach, a way you run hiring), name them here. Even a one-line reference is enough. Future-you will thank you.
You'll know it's working when you can look at your INDEX and answer: "if I'm writing today, what does my AI need to know first?"
Step 4: Write Your NOW (5 minutes)
Goes in: 4 Pulse
NOW is the simplest document of the three, and the one with the highest weekly value. It captures what is alive in you right now. Priorities. Decisions. Energy. What is blocked. Where your head is.
There is no right or wrong way to write this. It does not need to be comprehensive. It does not need to be polished. It is meant to evolve. Capture what is true for you this week, and let the document grow with you.
Open a new document. Save it as NOW.md in your 4 Pulse folder. Here is a starter template. Edit it freely. Throw away anything that doesn't fit. Add anything that does.
# Current State - Week of [today's date]
## Priority Order
1. [Top priority this week]
2. [Second]
3. [Third]
## Active This Week
- [What you are actually working on]
- [Deadlines that matter]
- [Anything else top of mind]
## Recent Decisions
- [Things you've decided in the last week or two that an AI helping you should know]
## Blocked
- [Anything stuck and why, or "Nothing currently blocked"]
## Energy
[One line on where your head is. Sprint mode? Recovery week? Deep focus? Distracted? Be honest.]
This is the document you update every week. Five minutes on a Sunday or a Monday morning. Nothing fancy. The act of writing it is half the value (you get clearer about your own week), and the AI benefit is the other half.
You'll know it's working when you read your NOW and feel a small click of "yes, that's where I am."
Step 5: Leave the Other Folders Empty (For Now)
You now have three filled documents. BRIEF in Identity. INDEX in Systems. NOW in Pulse. The other two folders (3 Projects and 5 Thinking) stay empty for now. This is intentional.
Projects (Layer 3) is where you create one document per active project, when a project gets serious enough to deserve its own context. Most people start with zero of these and add their first one when they realise they keep re-explaining the same project to the AI. That is the right pattern.
Thinking (Layer 5) is where daily journal entries, meeting notes, captured insights, and bits of raw material go. It is not a place to write up front. It is a place that fills up over time as you do the actual work. Empty today, full in three months.
Both folders are seats, not assignments. They mark the architecture so you know where things go when they show up. You don't have to invent things to put in them.
This is the part most "get organised" guides get wrong. They tell you to populate every folder before you start. That is how systems die in week two. Build the house. Move into three rooms. Furnish the others when you have furniture.
What this looks like once it has grown
Here is roughly what mine looks like now, after a year of using it. Yours will not look like this on day one. It does not need to. This is what it works towards as you actually do the work.
Colin's Vault/
1 Identity/
BRIEF.md
anti-patterns.md
2 Systems/
INDEX.md
Amplify OS.md
Four Capacities.md
Audience-First Marketing.md
People/
3 Projects/
humane-business/
CONTEXT.md
coaching-com/
CONTEXT.md
colinscotland-com/
CONTEXT.md
book/
CONTEXT.md
4 Pulse/
NOW.md
PARKED.md
5 Thinking/
journal/
insights/
meetings/
philosophy/
Two things to notice. First, every folder has grown from the same five-folder starting point. Nothing here was bolted on later. The architecture I started with is the architecture I still use. Second, this is just one shape. You can absolutely structure yours differently. Different folder names, different categories, different format. There is no right or wrong way to do it. The five layers are the spine; the rest is yours.
Step 6: Give It To Your AI
Now the AI part. You've built a structured home for your thinking. Time to make sure your AI of choice can actually read it.
Same Vault, same three documents, every platform. Only the loading mechanism changes. Here is the table for the platforms most people use. Find yours, follow the instructions for that row, and you are done.
| Platform | Where it goes | What to do |
|---|---|---|
| Claude (chat) | A Claude Project | Open claude.ai → Projects → Create project. Call it [Your Name] Working. Add files. Drag in BRIEF, INDEX, NOW from your Vault. Every conversation in that Project loads all three. |
| Claude Cowork | The whole Vault folder | In Claude Desktop, switch to Cowork mode. Click the folder icon and point it at your [Your Name]'s Vault folder. Cowork now reads everything inside, including the empty Projects and Thinking folders as they grow. |
| Claude Code | The whole Vault folder | In your terminal, run cd "[path to your Vault]" and then claude. Claude Code starts inside your Vault and can read every file. Best path for power users who want to script and automate against the stack. |
| ChatGPT | A ChatGPT Project | Open chatgpt.com → Projects → New project. Call it [Your Name] Working. Upload BRIEF, INDEX, NOW. Every chat in the Project sees them. |
| Gemini | A Gem | Open gemini.google.com → Gems → Create new Gem. Open BRIEF, INDEX, NOW. Copy each in turn and paste into the Gem's instructions field. Save. |
| Custom GPT | The instructions field | Same pattern as Gemini. Copy the contents of the three documents and paste into the GPT's instructions or system prompt field. Save. |
| Anything else (universal fallback) | Top of every chat | Open BRIEF, INDEX, NOW. Copy the contents. Paste them at the start of every new chat. More friction per session, works everywhere. |
A note on the Claude options. Claude (chat) is the standard web interface. Best for everyday conversations. Claude Cowork is for delegating actual work that runs on your machine in the background. Claude Code is the developer-facing terminal version, good if you want to start scripting against your Vault. All three read the same Vault. Pick what fits the kind of work you do most.
You'll know it's working when you start a new chat (or assign Cowork a task) and the response references your BRIEF, your priorities, and your voice without you having to paste anything.
Step 7: Prove It Works (1 minute)
Open a new chat in your AI of choice. Paste exactly this prompt:
Based on what you know about me, what should I focus on this week and why?
Read the response.
If it sounds like you, references your priorities from NOW, and talks like the BRIEF says you talk: it worked. Your stack is loaded.
If it sounds generic, like a stock motivational AI response, something didn't connect. Common fixes:
- The files didn't actually upload. Check your project's file list. Re-upload if needed.
- Your BRIEF is too vague. Open it and add specifics. "I'm a coach" needs to become "I'm a leadership coach for founders running 10 to 50 person teams, working through six-month engagements that combine weekly 1:1s with quarterly intensives." The same rule applies whatever you do: name the thing, name the audience, name the shape of the work.
- Your NOW is empty or stale. It should have actual priorities for this week, not placeholders.
Run the prompt again after the fix. Iterate until it sounds like you.
Maintaining It
The maintenance load is small. That is by design.
Weekly: NOW.md. Five minutes. Sunday night or Monday morning. Update priorities. Note decisions made in the last week. Refresh the energy line. This is the file that decays fastest, and the file that keeps the AI calibrated to where you actually are right now.
Monthly: Project files (when you have them). When you start filling the 3 Projects folder, each project gets reviewed monthly. Goals still right? Phase still accurate? Decisions captured?
Quarterly: INDEX.md. When the kinds of work you do change, the routing table needs to change too.
Yearly: BRIEF.md. Voice and values don't change often, but they do drift. A yearly read-and-update is enough.
Daily, when the urge takes you: Thinking. Journal entries. Meeting notes. Captured insights. There is no schedule here. There is no template. The Thinking folder is the place you go when you have something to put down. Some weeks you put nothing in it. Other weeks you put something in it every day. Both are fine.
Total maintenance load for a typical week: about 5 minutes.
Where It Gets Interesting
Here is where the Vault stops being a folder and starts becoming a system that runs itself.
Once your Vault is set up, you can point an agentic AI tool at it and have the AI maintain it for you. You finish a coaching call, and a meeting transcript automatically lands in 5 Thinking/meetings/. You make a decision in conversation, and an agent appends it to your Recent Decisions in NOW. You publish a blog post, and the agent updates the relevant project's CONTEXT.md to reflect the change. You wake up Monday morning and your NOW has been refreshed by an overnight scheduled task that read your week's calendar, your inbox, and the latest entries in your journal.
This is the loop. You think and create. The AI organises and maintains. The Vault gets sharper every week without you doing the maintenance work yourself.
I wrote a full walkthrough of how this works with Claude Cowork (the working partner mode I use most). It covers the scheduled tasks, the connectors, the agentic patterns that turn your Vault into a living system instead of a static archive: Claude Cowork: Your AI Working Partner.
The five-folder Vault is the foundation. Agentic AI is what turns it into a flywheel.
Why This Is Bigger Than AI
You haven't just set up an AI. You have built a structured home for your thinking.
The BRIEF you wrote is yours. It is the clearest single document about who you are and how you work that you have ever produced. That alone is worth the 30 minutes.
The INDEX is a map of your own working brain. Future-you will read it and remember things present-you was about to forget.
The NOW is a weekly mirror that catches drift before it becomes a quarter you can't account for.
The empty Projects and Thinking folders are seats waiting for the work that builds your authority over the next year.
All of this exists on your computer. In folders you own. Not locked inside any AI tool. If you switch from Claude to ChatGPT to whatever comes next in 18 months, your Vault moves with you. If the AI you use today disappears tomorrow, your Vault is unaffected. The folders outlast the tools.
This is your second brain. AI is the first thing that benefits from it. There will be a tenth thing.
Level Up When You're Ready
You have the foundation. Three filled documents, two seats reserved, one structured home for your thinking. Don't add anything else until this version has been working for two weeks.
When you're ready, here's what comes next:
Fill the Projects folder. When a project gets big enough that you find yourself re-explaining it to your AI every time, it deserves its own document. Create 3 Projects/[project-name]/CONTEXT.md. Goals, phase, decisions, stakeholders, what is blocked, what is next. One file per project. Add it to the AI's context when you work on that project.
Start filling Thinking. A daily journal note (5 Thinking/journal/YYYY-MM-DD.md). A folder for insights you want to keep. Meeting notes from calls that mattered. The Thinking folder is the long-term memory of the system. Let it grow naturally.
Add more skills to your INDEX. Every time you do a new type of work, write a routing entry for it. Future sessions get better automatically.
Read the deeper version. The full architecture is at colinscotland.com/human-stack. It explains why each layer exists, what breaks when one is missing, and how the system feeds itself over time.
Want help building yours? The Amplify OS Builder generates your BRIEF in about 10 minutes. If you want a deeper, guided build of the whole stack with live support, that is what the Amplify Intelligence Accelerator is for.
The Whole Thing in One Sentence
Open Finder, make a folder called [Your Name]'s Vault, create five subfolders for the five layers, write three starter documents (BRIEF, INDEX, NOW), give them to your AI as a Project, and watch what happens.
30 minutes. No code. Any computer. Any AI.
That's the whole setup.
Start with the Amplify OS Builder to write your BRIEF in 10 minutes.