One place that holds everything about you, that any AI can read. We’ll do it one simple step at a time.
No tech skills needed. You can’t break anything. You’ve got this.
Your second brain is just a folder. First, choose where that folder lives:
Pick the cloud you chose, download the desktop app, install it, and sign in.
In the drive you picked, create a new, empty folder. Name it something simple like “My Second Brain.”
That folder is your second brain. Everything will live inside it.
Download the Claude app. Don’t use the website version, the app is what we’ll use.
✨ Claude Open →Obsidian makes your brain easy to read and click through. This one’s optional.
📚 Obsidian Open →This is the magic part. Claude interviews you, then builds your whole folder system for you.
Everyone’s setup is different, so this interview isn’t about filling your brain yet. It’s about understanding your life and work so Claude can build the right folders for you.
Then paste this and answer its questions (one section at a time):
# YOU ARE THE KNOWLEDGE BASE BUILDER
Your job is to interview me, then design a personalized knowledge base structure that works as my external second brain. It needs to live in Obsidian (or Dropbox or Google Drive) and be easy for Claude to pull from later.
## What you're building
A folder system where I'll keep:
- Meeting transcripts (Fathom, Zoom, Granola, etc.)
- Emails and email summaries
- Documents, contracts, SOPs
- One file per person (people dossiers)
- Project notes
- Company information
- Personal context (family, health, money, properties)
- Promises I've made, decisions I've taken, follow-ups I owe
The structure has to hit four bars:
1. Built for Obsidian. Numeric prefixes on folders for sort order. Kebab-case filenames. YAML frontmatter on every file. Wikilinks ([[Name]]) between files. A Map of Content file (_MOC.md) in every major folder.
2. Easy for Claude to find things in. Predictable paths, clear names, and a _MOC.md in every folder so any LLM can see what's inside without reading every file.
3. RAG-style. One concept per file, lots of cross-links, tags for things that cut across categories.
4. Personalized to me. Not a generic template. The structure should reflect my actual businesses and life.
## How to run the interview
Ask me the questions below one section at a time. Wait for my answer before moving to the next section. If an answer is fuzzy or missing something important, ask one short follow-up, then keep going. Keep it efficient but warm. This should feel like a thoughtful onboarding, not a survey.
After every section is answered, produce the four outputs at the bottom.
---
## THE QUESTIONS
### Section 1: About Me
1. What's your name, and what city and country are you based in?
2. What's your main professional identity (founder, investor, executive, advisor)?
3. In one sentence, what's your single biggest professional focus for the next 12 months?
### Section 2: Your Businesses
4. How many businesses or companies are you actively involved in? For each one, give me:
- Company name
- Your role (founder, co-founder, CEO, advisor, board member, investor)
- Industry / what they do (one line)
- Rough team size
- Revenue stage (pre-revenue, under $1M, $1M to $10M, over $10M)
### Section 3: Other Professional Commitments
5. Are you on any boards (for-profit or non-profit)? If yes, list them.
6. Are you involved in any charities, foundations, or formal giving efforts?
7. Are you in any masterminds, peer groups, or formal mentor relationships (as the mentor or the mentee)?
### Section 4: Your People
8. Who are the 5 to 10 most important people in your professional life (not counting employees of your companies)? Think business partners, key advisors, accountant, lawyer, top clients, key vendors. Give me a first name and a one-line note for each.
9. Who are the 3 to 5 most important people in your personal life? Spouse, partner, kids, parents, key friends. First name and one-line note for each.
### Section 5: Tools and Data
10. Where does your work actually happen? Tell me which tool you use for each:
- Email
- Calendar
- File storage
- Team chat
- Meeting calls
- Project management
- CRM (if any)
- Meeting recording (Fathom, Otter, Granola, etc.)
11. How many separate Google accounts or email addresses do you use professionally?
### Section 6: Personal Context
12. Do you have significant assets that may need ongoing context? Yes or no on each. Real estate. Vehicles. Investment accounts. Business equity. Other. No details yet, just yes or no.
13. Any big life context I should know? Health stuff, family situations, recurring travel, time zone constraints. Share what you're comfortable sharing.
---
## WHAT TO PRODUCE AFTER THE INTERVIEW
Once every section is answered, produce these four outputs in order.
### Output 1: The Folder Structure
A full folder tree as a code block, customized to the answers I gave you. Use this base as a starting point and adjust it based on my actual situation:
Second Brain/
├── 00-CONTEXT/
│ ├── identity.md
│ ├── current-focus.md
│ ├── jargon.md
│ └── morning-brief-instructions.md
├── 01-COMPANIES/
│ ├── [Company A]/
│ │ ├── _MOC.md
│ │ ├── people/
│ │ ├── docs/
│ │ ├── calls/
│ │ ├── projects/
│ │ └── decisions/
│ └── [Company B]/
├── 02-PEOPLE/
│ ├── _MOC.md
│ └── [first-last].md
├── 03-BOARDS/ (only if applicable)
├── 04-CHARITIES/ (only if applicable)
├── 05-COMMUNITIES/
├── 06-PERSONAL/
│ ├── family/
│ ├── health/
│ ├── properties/
│ └── finances/
├── 07-CALLS-INBOX/
├── 08-DOCS-INBOX/
├── 09-DECISIONS/
└── 99-LOGS/
### Output 2: Three Starter File Templates
Ready-to-paste content for these three files:
**File A: 00-CONTEXT/identity.md.** Fill this in using my actual answers.
**File B: 00-CONTEXT/morning-brief-instructions.md.** The exact prompt, customized to my preferences.
**File C: 02-PEOPLE/_template.md.** A blank template for every person file.
### Output 3: Source Materials Checklist
Tell me what to gather and where to put it, customized to my actual tools and answers.
Phase 1: Foundation (do this today)
Phase 2: Recent context (this week)
Phase 3: Deep context (this month, as needed)
### Output 4: Frontmatter Standards Reference
One code block showing the YAML frontmatter pattern for each file type.
---
## RULES TO STICK TO
- Stay conversational. Don't dump every question at once. One section at a time.
- Probe when something's vague. One short follow-up, then move on.
- Personalize. Don't hand me a template. The output should feel built for me.
- Kebab-case filenames (john-doe.md, not John Doe.md).
- ISO dates (2026-05-08, not 5/8/26).
- Wikilinks for every cross-reference ([[John Doe]], [[Company Name]]).
- Never invent details. If something's missing, mark it [to be filled].
Start with Section 1 now.Now Claude understands your life and work, but it hasn’t built anything yet. This step has it create the actual folders.
When the interview wraps up, tell Claude to build the structure. Say something like:
“Okay, great. Please create this file/folder structure and put it in my Second Brain folder.”
This creates your CLAUDE.md file. It’s the rulebook that tells Claude how to read, file, and organize your whole brain.
In the same conversation, paste the prompt below and send it. It fills in what it knows about you and flags anything it’s unsure of.
I want to populate a CLAUDE.md file in my main second brain folder (and in subfolders where it makes sense) so you can navigate and understand my second brain more efficiently. Below is the CLAUDE.md template I want you to customize for me.
Two rules before you start:
1. Everything below the line goes into a single file named CLAUDE.md. Nothing outside that file.
2. Replace every bracketed placeholder ([like this]) with actual values pulled from what you already know about me. If you don't know a value, leave the bracket in place and flag it at the end.
---
# CLAUDE.md: Second Brain Operating Instructions
You are working inside [USER NAME]'s second brain. This file is the operating contract. Read it first, then act.
## What this folder is
A personal second brain organized for AI retrieval. Every piece of meaningful business and personal context lives here. The structure and rules below are what make it queryable.
## Read these first
Before doing anything in this folder, read in this order:
1. 00-CONTEXT/identity.md (who I am, what I run, what I focus on)
2. 00-CONTEXT/jargon.md (my internal terms, abbreviations, code names)
3. 00-CONTEXT/morning-brief-instructions.md (how I want my daily brief)
4. This file (CLAUDE.md) for structure and routines
## Folder structure
```
Second Brain/
├── 00-CONTEXT/ # Identity, focus, jargon (foundation)
├── 01-COMPANIES/ # One folder per business
│ └── [Company-Name]/
│ ├── _MOC.md
│ ├── people/
│ ├── docs/
│ │ ├── sops/
│ │ ├── contracts/
│ │ └── reference/
│ ├── calls/
│ ├── projects/
│ └── decisions/
├── 02-PEOPLE/ # Cross-cutting key people
│ └── _MOC.md
├── 03-BOARDS/ # (only if applicable)
├── 04-CHARITIES/ # (only if applicable)
├── 05-COMMUNITIES/ # Masterminds, peer groups
├── 06-PERSONAL/ # Family, health, properties
├── 07-CALLS-INBOX/ # Raw call transcripts arrive here
├── 08-DOCS-INBOX/ # Unsorted docs arrive here
├── 09-DECISIONS/ # Decision log (chronological)
├── 99-ARCHIVE/ # Stale, deprecated, historical
└── CLAUDE.md # This file
```
## Naming and format rules
- Filenames: kebab-case.md. No spaces, no capitals. (Exception: First-Last.md for people files.)
- Dates: ISO format YYYY-MM-DD.
- Call files: YYYY-MM-DD-call-with-[name-or-topic].md
- Decision files: YYYY-MM-DD-decision-[short-slug].md
- Cross-references: always use [[wikilinks]] to point to people, companies, projects, and other files.
- Tags: use sparingly. Save them for things that cut across categories (for example, #q3-priorities, #hiring-cohort-1).
## Frontmatter standards
Every file needs YAML frontmatter. The minimum required is type and updated.
```yaml
---
type: person | call | decision | sop | contract | reference | project | moc
updated: YYYY-MM-DD
tags: []
---
```
Type-specific frontmatter templates live at /00-CONTEXT/templates/.
## Cross-linking rules
- People files link to: every call they were in, every project they own, and the company they belong to.
- Call files link to: every participant (their people files), the company context, and related decisions.
- Decision files link to: people involved, related projects, and the company.
- Project files link to: owners, key people, related calls, and related decisions.
- A company's _MOC.md is the index. Link from it to active projects, key people, and recent decisions.
## What not to do
- Don't duplicate content across folders. Store it once, cross-link from the other.
- Don't edit historical call transcripts. They are the source of truth. Pull from them, don't change them.
- Don't rewrite a decision when it changes. Log a new decision that replaces the old one.
- Don't put sensitive personal content in operational folders. Use 06-PERSONAL/.
- Don't use folder names with spaces, capitals, or special characters.
- Don't create any file without frontmatter.
- Don't dump content into a generic "Notes" or "Misc" folder. Route it properly or ask me.
---
## MAINTENANCE ROUTINES
When I ask you to run one of these, follow it exactly. When you spot a situation that matches a routine, suggest running it.
### Routine 1: File a new call transcript
Trigger: a new file shows up in 07-CALLS-INBOX/, or I say "file this call."
1. Read the full transcript.
2. Identify participants. For each name, check if 02-PEOPLE/[First-Last].md exists. If not, ask: "I don't have a people file for [Name]. Should I create one?"
3. Identify the company context. If unclear, ask.
4. Add frontmatter to the transcript:
```yaml
---
type: call-transcript
date: YYYY-MM-DD
participants: [[First-Last1]], [[First-Last2]]
company: [Company name]
tags: []
summary: [one-line summary]
---
```
5. Extract decisions. Any meaningful decision becomes a new file in 09-DECISIONS/.
6. Extract promises. Any promise I made updates the relevant person's file in 02-PEOPLE/.
7. Rename the file: YYYY-MM-DD-call-with-[name-or-topic].md
8. Move it to 01-COMPANIES/[Company]/calls/
9. Update the relevant _MOC.md.
10. Report back: "Filed [filename]. Created [X] decision entries. Updated [Y] people files."
### Routine 2: File a new document, SOP, or photo
Trigger: I drop a document, photo, or PDF into 08-DOCS-INBOX/, or I say "file this doc."
1. If it's an image, extract the text with vision. Ask me to fill in anything that didn't read cleanly.
2. Classify it. SOP goes to sops/. Contract goes to contracts/. Reference goes to reference/. Personal goes to 06-PERSONAL/. If unclear, ask.
3. If the original was a PDF or image, convert to markdown. Keep the original in a sibling originals/ folder.
4. Add frontmatter.
5. Cross-link. Scan the doc for any people, companies, or projects mentioned and wrap them in wikilinks.
6. Suggest a kebab-case filename based on the content.
7. Confirm the route with me before final filing. Wait for my OK.
8. Move to the final location. Update the relevant _MOC.md.
9. Report back.
### Routine 3: Create a new person file
Trigger: I introduce a new important person, or Routine 1 surfaces a participant without a file.
1. Confirm with me that this person warrants a file.
2. Use the template at 02-PEOPLE/_template.md.
3. Fill in what's known: name, relationship type, company connection, one-line context.
4. Ask me for 2 or 3 sentences of background (how I know them, why they matter).
5. Add them to 02-PEOPLE/_MOC.md.
6. Cross-link from the company folder if it applies.
7. Suggest 2 to 4 tags based on context.
8. Report back.
### Routine 4: Log a decision
Trigger: I say "log this decision," or Routine 1 finds a meaningful decision in a call.
1. Create the file: 09-DECISIONS/YYYY-MM-DD-[short-slug].md
2. Apply this template:
```yaml
---
type: decision
date: YYYY-MM-DD
status: pending | decided | reversed
company: [Company or "personal"]
people: [[Person1]], [[Person2]]
trigger-to-revisit: [conditions]
tags: []
---
# Decision: [short title]
## What was decided
[One paragraph]
## Why
[Reasoning]
## Alternatives considered
- [Alt 1]: why rejected
- [Alt 2]: why rejected
## Action items
- [ ] [Who]: [What] by [When]
```
3. Link to the source call if there is one.
4. Update people files for everyone involved.
5. Report back.
### Routine 5: Weekly review
Trigger: I ask for "weekly review," or we hit the agreed weekly review time.
1. Scan 02-PEOPLE/. List every file where status is active and last_contact is more than 14 days ago.
2. Scan promises. Across all people files, surface any open promise older than expected.
3. Scan 09-DECISIONS/. List decisions with status: pending for more than 7 days.
4. Scan inboxes. Check 07-CALLS-INBOX/ and 08-DOCS-INBOX/ for anything unfiled.
5. Scan recent activity. What got logged in the last 7 days? Any patterns?
6. Generate the report and save it to 99-LOGS/YYYY-MM-DD-weekly-review.md.
### Routine 6: Quarterly audit
Trigger: I ask for "quarterly audit," or we hit the quarter boundary.
1. Re-read 00-CONTEXT/identity.md. Flag anything outdated.
2. Audit 02-PEOPLE/. Flag active people not contacted in 90+ days. Suggest moving them to status: dormant.
3. Audit SOPs. Flag any in docs/sops/ not updated in 90+ days.
4. Audit 01-COMPANIES/. Verify each _MOC.md reflects current reality.
5. Suggest archives. Files older than 6 months that haven't been referenced.
6. Generate the full report. Save it to 99-LOGS/YYYY-Q[N]-audit.md.
7. Recommend the top 3 structural changes for next quarter.
### Routine 7: Process inboxes
Trigger: I say "process my inbox" or "clear the inboxes."
1. List every file in 07-CALLS-INBOX/ and 08-DOCS-INBOX/.
2. For each file: if it's a call, run Routine 1. If it's a doc, run Routine 2. If unclear, ask me where it goes.
3. Report back with a count of what got filed where.
## How I invoke routines
[Insert preferred trigger phrases here, for example: "file this call," "log this decision," "weekly review," "quarterly audit," "process my inbox."]
## Communication style
When working in this folder:
- Be direct. Skip preamble.
- Always confirm structural changes before making them ("I'm about to move X to Y, confirm?").
- Surface ambiguity. If something could go in two places, ask.
- Report concisely. After running a routine, give me a one-paragraph summary.
- Never silently change historical data. If you change anything in 09-DECISIONS/ or 02-PEOPLE/, tell me what changed and why.
## When you're not sure
If you hit a situation these routines don't cover, default to:
1. Ask, don't guess.
2. Preserve, don't delete. Move things to 99-ARCHIVE/ if you're uncertain.
3. Cross-link generously. Better to over-link than under-link.
4. Add frontmatter even if minimal. At least type and updated.Your brain is built. Now you fill it with your real information.
There are 8 kinds of content that can flow in. You don’t have to do all of them. Pick the ones that matter to you. You can always add more later.
Got documents you want added to your brain as context? Add the ones that matter, and Claude will file and connect them for you.
In Claude Code with your second brain folder selected, drop in your files. Then say something like:
“I’m uploading these files to add to my second brain. For each one, decide where it goes, file it in the right folder based on what it is, and cross-link it with the right tags so everything is connected. Then create a quick overview so we can navigate our second brain fast. Keep the original files.”
Want your email added to your brain as context? Connect your account, and Claude can pull in what matters and file it for you.
In Claude, open Connectors and connect your email account (Gmail uses the Gmail connector).
In Claude Code with your second brain folder selected, say something like:
“Pull any important emails from my connected email accounts into my second brain. For each email or thread that matters, file it in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you're done doing an entire backlog of all of my emails that are important, we need to set up a routine that will search through any new emails that have come in since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Want your work chats and texts added to your brain as context? Connect them, and Claude can pull in the conversations that matter.
In Claude, open Connectors and connect Slack. (On a Mac, Claude can read your iMessages, and there’s a Claude iMessage connector.)
In Claude Code with your second brain folder selected, say something like:
“Pull relevant messages from my connected Slack and texts into my second brain. For each conversation that matters, file it in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you’re done doing an entire backlog of all of my important messages, we need to set up a daily routine that will search through any new messages that have come in since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Use whatever AI notetaker you already have, like Fathom or Granola. Connect it, and Claude can pull your meeting notes into your brain.
Connect your notetaker to Claude through a connector or a Zapier MCP.
In Claude Code with your second brain folder selected, say something like:
“Pull my meeting notes and transcripts from my connected notetaker into my second brain. For each meeting, file it in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you're done doing an entire backlog of all of my important meetings, we need to set up a daily routine that will search through any new meetings that have come in since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Record your phone calls? Add those to your brain too. This works with a recorder like Plaud, or an AI notetaker that records calls (Granola’s mobile app does).
Connect your recorder or notetaker to Claude through a connector or a Zapier MCP.
In Claude Code with your second brain folder selected, say something like:
“Pull my recorded calls and their transcripts from my connected notetaker into my second brain. For each call, file it in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you're done doing an entire backlog of my calls, we need to set up a daily routine that will search through any new calls that have come in since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Use a CRM or a to-do list? Connect it so Claude can pull in what matters: things that need doing, important changes, and useful history.
Connect your CRM or task tool to Claude through a connector or a Zapier MCP.
In Claude Code with your second brain folder selected, say something like:
“Pull the relevant information from my connected CRM or task list into my second brain: things that need to be done, important changes, and useful historical context. File each item in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you're done doing an entire backlog, we need to set up a daily routine that will search through anything new since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Keep notes in an app like Evernote or Apple Notes? Pull the ones worth keeping into your brain.
Connect your notes app to Claude through a connector or a Zapier MCP if it has one.
In Claude Code with your second brain folder selected, say something like:
“Pull the relevant notes from my connected notes app into my second brain. For each note that matters, file it in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you're done doing an entire backlog, we need to set up a daily routine that will search through any new notes since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Want your real-life conversations in your brain too? That’s what AI wearables are for. They record and transcribe what you talk about in person.
Phil’s pick is Omi, because it’s always on. With Plaud you turn it on for each conversation.
Most wearables sync to a cloud folder (Omi syncs to Google Drive, Dropbox, etc.). Point Claude at that folder, or connect it through a connector or Zapier MCP.
In Claude Code with your second brain folder selected, say something like:
“Pull my recorded in-person conversations and their transcripts from my connected wearable into my second brain. For each conversation that matters, file it in the right folder, cross-link the people and companies involved, and add the right tags so everything is connected. Once you're done doing an entire backlog, we need to set up a daily routine that will search through any new conversations since the previous scan, determine what is important, and file those to the brain as well. That way the brain stays updated every single day.”
Once your info is in, run this every week. It re-checks every connection, flags conflicts, and keeps everything clean and accurate, and waits for your OK before changing anything.
In Claude Code with your second brain folder selected, paste this and run it:
# YOU ARE THE CONNECTION FINDER Your job is to scan my second brain, find patterns and connections across the files, flag anything that conflicts, and keep the information as clean as possible. Surface what you find. Don't change anything without my OK. ## What to do 1. Read my CLAUDE.md file first so you understand the folder structure and conventions. 2. Walk the second brain folder by folder. Build a mental map of what's in here: people, companies, projects, decisions, ideas, opportunities, calls, docs. 3. Look for connections that exist but haven't been linked yet. Examples: - People who belong to or work with a company - People connected to specific ideas or opportunities - Companies connected to ideas, projects, or opportunities - Calls that mention people or companies who don't yet show up as wikilinks in that file - Decisions that tie back to specific people or projects - Tags, topics, or themes that cluster across multiple files 4. Look for conflicts. Any place where one file says one thing and another says something different. 5. Look for cleanup opportunities. Dead links, duplicate people files, inconsistent tags, missing frontmatter, files in the wrong folder. Don't act on these. Put them in the changelog and wait for me: - Conflicts between files. Which version is right is a judgment call. - Possible duplicate people files. Don't merge without my OK. - Files that look like they're in the wrong folder. - Anything that would change a historical record (call transcripts, decisions, dated entries). - Anything you're less than 90% confident about. ## Hard rules - Never edit historical call transcripts. They're source of truth. Pull from them, don't change them. - Never rewrite a decision in 09-DECISIONS/. If a decision needs to change, log a new one that supersedes it. - Never delete a file. If something should go, move it to 99-ARCHIVE/. - Never invent connections. Only link things actually mentioned in the files. ## What to deliver After every run, write a changelog to 99-LOGS/YYYY-MM-DD-connection-scan.md with these sections: 1. **Summary line.** Counts: X links added, Y files updated, Z conflicts flagged, A cleanup items flagged. 2. **What I changed.** Every change you made. File path, what changed, why. 3. **What I flagged.** Conflicts, duplicates, suspicious routing, low-confidence calls. File path and what's off. 4. **Patterns worth your attention.** People showing up across multiple companies, recurring topics, opportunities that connect to multiple ideas, anything that hints at a bigger thread. Keep the section headers and formatting consistent across runs so I can compare logs over time. ## How I'll invoke this later I want to turn this into a recurring routine (weekly).
That’s the whole build. Open Claude Code in your second brain folder and just ask. A few things to try:
The best daily habit: a morning brief. Set it once:
“Every morning, give me a short brief from my second brain: the 3 things that matter most today, anything that's slipping, promises I owe people, and a heads-up on my meetings. Set this up as a routine that runs every morning and sends it to me.”
You’ve got a second brain that knows you, stays current, and works for you every single day.
You just built a second brain with me walking you through every step. Now picture that all year long: me in your corner every week, every workshop like this one included, and a room of people building right alongside you.
That’s The Stratosphere.
An elite 12-month AI mastery community. The full roadmap, live access to me every single week, and everything I release, all in one place.
Join for a one-time payment of $3,497, or spread it over 3 monthly payments of $1,397.
When The Stratosphere reopens on July 1, we’re capping it at 10 members for the year, and there are already more than 200 people on the waitlist.
Join today and your Second Brain workshop ticket is on us, we’ll refund it. You’re grandfathered in: every workshop free for the next year, including this one.
Join The Stratosphere →philstringer.com/stratosphere · 10 spots · 200+ on the waitlist · today only
Stuck? Email admin@philstringer.com