Knowledge Map
The foundational mental model for organizing projects, context, and collaboration.
The 3-Object Model
Goals
What we want to achieve
"Why does this exist?"
Context
Knowledge required to achieve goals
"What do we need to know?"
Collaborators
Humans who contribute
"Who helps make this happen?"
The file system is derived from this model. Each folder exists to serve a goal, contains relevant context, and is accessible to relevant collaborators.
Core Principles
AI-First Collaboration
Minimize human collaborators to the essential few.
Claude Code is the primary collaborator. Human collaborators are the exception, not the default.
| State | Description |
|---|---|
| Ideal | Goal achieved with Claude Code alone |
| Acceptable | Human collaborator provides significant long-term value |
| Avoid | Human collaborator out of habit, convenience, or "that's how it's done" |
When Humans Are Essential
- Relationships requiring human presence (clients, events, local markets)
- Judgment requiring domain expertise Claude Code lacks
- Execution requiring physical presence or legal identity
- Long-term institutional knowledge that compounds over time
Knowledge-First Principle
Default to achieving goals with knowledge (md files) + AI agents. Only write code when that fails.
| Layer | What | When |
|---|---|---|
| Knowledge | Markdown files โ context, playbooks, notes | Always start here |
| AI Agents | Claude Code operating on knowledge | Default execution layer |
| Code | Repos, apps, custom software | Last resort โ when knowledge + AI hits a wall |
Why Knowledge-First
- Markdown is native interface for AI โ lowest friction
- Code adds maintenance burden (dependencies, hosting, updates)
- Knowledge compounds over time; code often needs rewrites
- Faster iteration (edit md vs. ship feature)
- Forces clarity โ if you can't write it down, you don't understand it
Inside/Outside Game Model
| Game | Who | What |
|---|---|---|
| Inside | You + Claude Code | Content creation, strategy, execution, systems |
| Outside | Human collaborators | Relationships, representation, local presence, judgment in rooms AI can't enter |
Directory Structure
Identity Groups (Top-Level)
| Group | Purpose | Sharing |
|---|---|---|
self/ | Personal growth, learning, legacy for children | Private |
family/ | Shared family resources, legal, financial | Family |
community/ | Public contribution, open source (reserved) | Public |
work/ | Professional projects, income generation | Project-based |
Work: Three Businesses
Worca
Talent Agency
Serve clients with quality talent. 13 team members + 2 coaches.
HIHQ
AI & Data Research Lab
Tech stack, research agenda, product vision. AI-first development.
Angel
Investment
Find generational founders, achieve 1000X returns. AI-first analysis.
Goals โ Projects
self/
| Project | Goal |
|---|---|
research/ | Build knowledge base, preserve insights for children |
learning/ | Skill development, experiments |
family/
| Project | Goal |
|---|---|
tax/ | Tax compliance, optimization |
children-529-plan-and-roth/ | Education funding |
travel/ | Family travel planning |
anna-playground/ | Teach Anna to code |
2788-extension/ | Home extension project |
work/
| Project | Goal | Status |
|---|---|---|
worca/ | Operate talent agency | Active |
hihq/ | AI & Data Research Lab | Active |
angel/ | Investment pipeline | Active |
writings/ | Build public presence | Pending |
File Tree
/Code
โโโ CLAUDE.md # Cross-cutting conventions
โโโ KNOWLEDGE_MAP.md # 3-object model, meta, collaborators
โโโ OPPORTUNITIES.md # Action items
โโโ AI_TASKS.md # AI routines
โ
โโโ self/ # Private
โ โโโ research/ # Knowledge base, legacy
โ โโโ learning/ # Skill development
โ
โโโ family/ # Family only
โ โโโ 2788-extension/
โ โโโ anna-playground/
โ โโโ children-529-plan-and-roth/
โ โโโ tax/
โ โโโ travel/
โ
โโโ work/ # 3 businesses
โโโ worca/ # Talent Agency
โ โโโ CLAUDE.md
โ โโโ community/
โ โโโ repos/
โ
โโโ hihq/ # AI & Data Lab
โ โโโ CLAUDE.md
โ โโโ repos/hihq-monorepo/
โ
โโโ angel/ # Investment
โ โโโ CLAUDE.md
โ โโโ active/
โ โโโ portfolio/
โ โโโ passed/
โ
โโโ writings/ # Public presence (AI-first)
โโโ CLAUDE.md Collaborators
Worca Team (13 + 2 coaches)
LA (2)
| Name | Role | Inside/Outside | Notes |
|---|---|---|---|
| Alec Fang | Operations | Outside | Client relationships, financial ops |
| Linda Chen | External Relations | Outside | Brand ambassador, PR, media |
Linda's Highest-Value Activities
- External representation (podcasts, events, panels, media)
- Relationship development (journalists, influencers, partners)
- Client-facing PR (crisis management, sensitive comms)
- Industry intelligence (signals from conversations AI can't access)
Linda is the "outside game" โ you + Claude Code handle the "inside game."
Taiwan (4)
| Name | Role | Inside/Outside | Notes |
|---|---|---|---|
| Lydia Huang | Head of Talent | Outside | Events, bench relationships. Discord should be AI-first. |
| Wilson Yeh | Head of Ops | Outside | Government, compliance, local ops |
| Jim Chang | Tech Lead (PT) | Evaluate | Could be AI-first? |
| Fenny Lai | PM (PT) | Evaluate | Could be AI-first? |
Philippines (5)
| Name | Role | Inside/Outside |
|---|---|---|
| Alvin Calma | Account Manager | Outside |
| Ivan Gonzales | Head of Talent | Outside |
| April Larkin | Training Manager | Outside |
| Alexjandra | Intern (PT) | Evaluate |
| Izak | Intern (PT) | Evaluate |
Coaches (2)
| Name | Location | Role | Inside/Outside |
|---|---|---|---|
| Luke Welch | SF | GTM strategy | Outside |
| Doyle Slayton | Dallas | Sales & closing | Outside |
Re-evaluation Queue
| Collaborator | Current Role | Question |
|---|---|---|
| Linda Chen | Writings, PR | Is she creating content or executing/scheduling? Could CC draft, Linda publish? |
| Lydia Huang | Discord community | Does she have community-building expertise? If not, why is she doing this? |
Case Studies
| Project | Current State | Assessment |
|---|---|---|
| HIHQ | Built with Claude Code, no human devs/PMs | Success Proof that AI-first works. |
| Writings | Relying on Linda | Re-evaluate Should be AI-first. |
| Worca Discord | Relying on Lydia | Failing Clear candidate for AI-first. |