Borrow the thinking, not the personality.
Borrowed Brain Pro is a decision intelligence system that applies evidence-backed thinking lenses, historical failure audits, and multi-lens panels to real-world decisions. Not celebrity roleplay — structured decision support.
# Clone into your AI agent skill folder
git clone https://github.com/DOTfei/borrowed-brain-pro.git ~/.claude/skills/borrowed-brain-pro
# Or paste borrowed-brain-bundle.md as a system prompt in ChatGPT, Claude, or any LLM| Platform | Method |
|---|---|
| ChatGPT / Claude.ai | Paste borrowed-brain-bundle.md as system prompt |
| Cursor / Windsurf | Copy .cursorrules into project root |
| Codex / Claude CLI | git clone into ~/.claude/skills/ |
| Ollama / Open WebUI | Paste bundle into Modelfile system field |
Four modes, one skill:
"Build a thinking profile for Jensen Huang" → Distill Mode
"Using Jobs\'s profile, what am I missing?" → Apply Mode
"Compare Jobs and Altman on my launch timing" → Compare Mode
"Convene the Boardroom on my pricing dilemma" → Boardroom Mode
Generic AI vs Borrowed Brain Pro:
| Generic prompt | Borrowed Brain Pro | |
|---|---|---|
| Input | "Think like Buffett" | "Apply Buffett's profile to my SaaS deal" |
| Output | Generic motivational quotes | Sourced principles + explicit failure boundaries |
| Value | Near zero | Actionable decision lens with concrete limits |
Say "Convene the Boardroom on [your dilemma]" — the system auto-selects 3–4 complementary lenses and runs a structured 3-round debate:
- Round 1 — Each lens states its opening position on your dilemma
- Round 2 — Lenses cross-examine each other directly by name
- Round 3 — Consensus, irreconcilable conflicts, and board blind spots
- Output — Action table + Smallest Useful Test
Example board for a product/startup dilemma: Jobs (simplification) × Graham (user validation) × Munger (inversion) × Hastings (fast feedback).
Every lens includes documented failure boundaries. Four audited case files ship with the repo:
| Case | Figure | What Failed | Rule Extracted |
|---|---|---|---|
| Alibaba Investment | Munger | Overestimated platform moat | Low valuation can't save an eroding moat |
| Qwikster Split | Hastings | Conviction suppressed internal dissent | Farm for dissent before major changes |
| NeXT Workstation | Jobs | Perfectionism ignored willingness-to-pay | Elegance can't overcome price ceiling blindness |
| Model 3 Hell | Musk | Automated before simplifying | Never automate a step that should be deleted |
Each audit includes Self-Check Questions so you can test whether you're repeating the same mistake.
| Person | Best for |
|---|---|
| Warren Buffett | Valuation discipline, saying no to a deal |
| Charlie Munger | Inverting problems, avoiding bias |
| Steve Jobs | Feature cuts, simplification, launch timing |
| Chris Voss | Negotiations, breaking deadlocks |
| Richard Feynman | Auditing logic, hunting self-deception |
| Cal Newport | Focus, deep work, time allocation |
| Reed Hastings | Building candor, strategic pivots |
| Sam Altman | Iterative shipping, speed vs. safety |
| Paul Graham | Early-stage validation, doing things that don't scale |
| Elon Musk | First principles, deleting requirements |
| Jensen Huang | Platform bets, conviction under pressure |
| Linus Torvalds | Systems design, pragmatic architecture |
| Travis Kalanick | Aggressive market entry, speed vs. compliance |
| Julia Evans | Technical writing, explaining hard things simply |
| Marcus Aurelius | Crisis leadership, stoic decision-making |
| Nietzsche | Adversity, re-evaluating inherited assumptions |
| Socrates | Epistemic humility, auditing false certainty |
| Laozi | Wu Wei, organic emergence, non-forcing |
Pre-configured advisory boards for common decision types:
| Pack | Members |
|---|---|
| Startup Founders | Jobs, Altman, Musk, Huang, Graham, Hastings |
| Investors & Capital | Buffett, Munger, Voss |
| Engineering | Torvalds, Musk, Feynman, Evans, Newport |
| Crisis Leadership | Aurelius, Munger, Feynman, Newport |
| Builder Decisions | Jobs, Graham, Munger, Hastings |
- No fabricated quotes — direct quotes capped at 15 words per source
- Every claim traceable to a verified public source
- Private individuals refused — public figures only
- Every response closes with what the lens misses in your specific situation
Contributions and new distilled profiles are welcome. See CONTRIBUTING.md.
# Rebuild the bundle after adding profiles or audits
python scripts/build_bundle.pyMIT License · DOTfei