Afrexai Strategic Thinking

Strategic Thinking & Mental Models Engine

The comprehensive decision-making methodology for founders, operators, investors, and leaders. 50+ mental models organized by when to use them, with templates and scoring systems.

Quick Start — /decide

When the user says "help me decide" or "analyze this decision":

1. Ask: What's the decision? (one sentence)
2. Ask: What type? (business / investment / hiring / product / personal / technical)
3. Ask: Reversibility? (easy to undo / hard to undo / permanent)
4. Ask: Time pressure? (minutes / days / weeks / no deadline)
5. Select the right framework(s) from the catalog below
6. Walk through step-by-step
7. Score using the Decision Quality Rubric (Phase 10)
8. Output a Decision Record (Phase 11)

/8 — Quick Decision Health Check

Score the current decision process (1-5 each):

≥32: Strong process — proceed with confidence 24-31: Decent — address weak dimensions before committing 16-23: Gaps — slow down and fill them ≤15: Stop — you're about to wing a consequential decision

Phase 1: Decision Classification

Not all decisions deserve the same process. Classify first.

Type 1 vs Type 2 (Bezos Framework)

The #1 mistake: Treating Type 2 decisions like Type 1. This creates organizational paralysis. Speed on Type 2 decisions is a competitive advantage.

Consequence Mapping

Before choosing a framework, map consequences:

decision: "[What you're deciding]"
type: 1 | 2
reversibility_cost: "$X / Y hours / Z reputation damage"
upside_if_right: "[Best realistic outcome]"
downside_if_wrong: "[Worst realistic outcome]"
time_to_know: "[When will you know if this was right?]"
asymmetry: "positive | negative | symmetric"
# positive = upside >> downside (bet freely)
# negative = downside >> upside (be cautious)
# symmetric = roughly equal (use expected value)

Phase 2: First Principles Thinking

Before reaching for frameworks, strip the problem to fundamentals.

The 5 Whys (Root Cause)

Don't solve symptoms. Ask "Why?" five times:

1. Why are we losing customers? → They churn after month 3.
2. Why month 3? → That's when the free premium features expire.
3. Why do they leave when features expire? → They haven't built habits around core features.
4. Why haven't they built habits? → Onboarding doesn't guide them to sticky features.
5. Why doesn't onboarding cover this? → It focuses on setup, not value realization.

Root cause: Onboarding design, not pricing or product gaps.

Inversion (Jacobi Method)

Instead of "How do I succeed?", ask "How would I guarantee failure?"

Template:

Goal: [What you want to achieve]

How to guarantee failure:
1. [Anti-pattern 1]
2. [Anti-pattern 2]
3. [Anti-pattern 3]
4. [Anti-pattern 4]
5. [Anti-pattern 5]

Therefore, avoid:
1. [Inverted actionable rule]
2. [Inverted actionable rule]
3. [Inverted actionable rule]

Regret Minimization (Bezos)

For life-altering Type 1 decisions:

"Project yourself to age 80. Which choice minimizes regret?"

Use when:

• Career changes (leave job to start company?)
• Major financial commitments
• Relationship decisions
• The analytical frameworks feel inadequate because values are at stake

Phase 3: The Core Mental Models Catalog

3.1 — Strategy & Business

Porter's Five Forces (Industry Attractiveness)

Score each 1-5 (1 = favorable, 5 = threatening):

Moat Assessment (Competitive Advantage)

Score each dimension 0-10:

OODA Loop (Speed Advantage)

For competitive situations where speed matters:

1. Observe: What's happening? Raw data, signals, changes.
2. Orient: What does it mean? Context, mental models, cultural factors.
3. Decide: What will we do? Select action from options.
4. Act: Execute. Then observe again.

Key insight: The winner isn't who has the best strategy — it's who cycles through OODA faster. If you can observe and orient faster than competitors, you'll always be inside their decision loop.

Wardley Mapping (Strategic Positioning)

Map components by:

• Y-axis: Visibility to user (top = visible, bottom = invisible)
• X-axis: Evolution stage: Genesis → Custom → Product → Commodity

Rules:

• Build what's in Genesis/Custom (your differentiation)
• Buy what's in Product/Commodity (don't reinvent wheels)
• Watch for components about to shift stages (opportunity/threat)

3.2 — Investment & Financial

Expected Value Calculation

For any bet or investment:

EV = (Probability of Win × Win Amount) - (Probability of Loss × Loss Amount)

Example:
- 30% chance of winning $100,000
- 70% chance of losing $20,000
- EV = (0.30 × $100,000) - (0.70 × $20,000) = $30,000 - $14,000 = +$16,000

Decision: Positive EV → take the bet (if you can afford the loss)

Kelly Criterion (optimal bet sizing):

Kelly % = (bp - q) / b
Where:
  b = odds received (win/loss ratio)
  p = probability of winning
  q = probability of losing (1 - p)

Example: 60% win rate, 2:1 payout
Kelly = (2 × 0.6 - 0.4) / 2 = 0.4 = 40%
Half-Kelly (safer): 20% of bankroll

Rule: Never bet full Kelly. Half-Kelly or quarter-Kelly in practice.

Margin of Safety (Graham/Buffett)

intrinsic_value: "$X (your best estimate)"
current_price: "$Y"
margin_of_safety: "(X - Y) / X × 100%"
# ≥30% for stable businesses
# ≥50% for uncertain/cyclical
# ≥70% for speculative/turnarounds

Application beyond investing:

• Hiring: Can this person do 30% more than the role requires?
• Timelines: Add 50% buffer to estimates
• Revenue projections: Plan for 70% of optimistic scenario
• Server capacity: Provision 2x expected peak

Asymmetric Risk/Reward

The best decisions have capped downside and uncapped upside:

Examples of asymmetric positive bets:

• Angel investing ($5K loss max, 100x upside possible)
• Content creation (time investment, infinite distribution upside)
• Learning a skill (months invested, decades of returns)
• Cold outreach (rejection cost = 0, deal value = $$$)

3.3 — Product & Prioritization

ICE Scoring (Quick Prioritization)

Score = Impact × Confidence × Ease

Calibration:

• Impact: Revenue, retention, or growth effect
• Confidence: How sure are you about Impact? (data-backed = 8+, gut = 3-5)
• Ease: 10 = hours, 7 = days, 4 = weeks, 1 = months

Jobs To Be Done (JTBD)

Template:

When [situation/trigger],
I want to [motivation/job],
So I can [expected outcome].

Functional job: [What they're literally trying to do]
Emotional job: [How they want to feel]
Social job: [How they want to be perceived]

Insight: People don't buy products. They hire them to make progress. Understand the job, and the product/feature decisions become obvious.

Eisenhower Matrix (Time/Priority)

Key insight: Most people spend 80% of time in Urgent (both quadrants). Winners spend 80% in Important/Not Urgent (Q2) — that's where compounding happens.

3.4 — Risk & Uncertainty

Pre-Mortem (Klein)

Before committing to a plan:

"Imagine it's 6 months from now. This decision was a disaster. What went wrong?"

Template:

decision: "[What we're about to do]"
pre_mortem_failures:
  - failure: "[What went wrong]"
    probability: "high | medium | low"
    severity: "catastrophic | major | minor"
    prevention: "[What we'll do to prevent this]"
    detection: "[How we'll know early if this is happening]"

Run with 3+ people independently, then combine. The exercise works because it gives permission to voice concerns that "positive thinking" culture suppresses.

Scenario Planning (Shell Method)

Don't predict the future. Prepare for multiple futures.

scenarios:
  optimistic:
    name: "[Descriptive name]"
    assumptions: ["[Key assumption 1]", "[Key assumption 2]"]
    probability: "X%"
    our_response: "[Strategy if this happens]"
    leading_indicators: ["[Signal 1]", "[Signal 2]"]

  base_case:
    name: "[Descriptive name]"
    assumptions: ["[Key assumption 1]", "[Key assumption 2]"]
    probability: "X%"
    our_response: "[Strategy if this happens]"
    leading_indicators: ["[Signal 1]", "[Signal 2]"]

  pessimistic:
    name: "[Descriptive name]"
    assumptions: ["[Key assumption 1]", "[Key assumption 2]"]
    probability: "X%"
    our_response: "[Strategy if this happens]"
    leading_indicators: ["[Signal 1]", "[Signal 2]"]

  black_swan:
    name: "[Descriptive name]"
    assumptions: ["[Unlikely but catastrophic event]"]
    probability: "<5%"
    our_response: "[Survival plan]"
    hedges: ["[Protection 1]", "[Protection 2]"]

Rule: If your plan only works in one scenario, it's not a plan — it's a prayer.

Antifragility Assessment (Taleb)

Score your system/business/portfolio:

Total: ≥4 = antifragile, 0 = robust, ≤-4 = fragile (fix immediately)

3.5 — People & Organizational

Circle of Competence (Munger)

Before any decision in a domain:

Domain: [Area of decision]

Inside my circle:
- [What I genuinely understand from experience]
- [Where I have real data and pattern recognition]
- [Decisions I've made successfully before in this space]

Edge of my circle:
- [What I know I don't know]
- [Where I'd need expert input]

Outside my circle:
- [What I'm completely unfamiliar with]
- [Where I'd be guessing]

Decision: Am I inside my circle for THIS specific decision?
If no → find someone who is, or do the homework first.

Hanlon's Razor + Steel Man

Before reacting to someone's behavior or proposal:

1. Hanlon's Razor: "Never attribute to malice what is adequately explained by incompetence" (or ignorance, busy-ness, different priorities)
2. Steel Man: Before arguing against a position, articulate the STRONGEST version of it. If you can't steel-man it, you don't understand it enough to disagree.

Second-Order Thinking

Every decision has consequences (1st order). Those consequences have consequences (2nd order).

Template:

Decision: [What we're doing]

1st order effects (immediate):
- [Direct result 1]
- [Direct result 2]

2nd order effects (weeks/months later):
- [Consequence of result 1] → [Further consequence]
- [Consequence of result 2] → [Further consequence]

3rd order effects (months/years later):
- [Systemic change 1]
- [Systemic change 2]

Counter-intuitive insight: [What becomes clear only at 2nd/3rd order]

Classic examples:

• Lowering prices (1st: more customers → 2nd: competitors match → 3rd: margin compression industry-wide)
• Remote work (1st: flexibility → 2nd: global talent pool → 3rd: global competition for your job)
• Firing quickly (1st: team relief → 2nd: hiring bar rises → 3rd: culture of accountability)

3.6 — Negotiation & Persuasion

BATNA Analysis (Fisher/Ury)

Before any negotiation:

my_batna: "[Best Alternative To Negotiated Agreement — what I do if we don't agree]"
my_batna_value: "$X or equivalent"
their_batna: "[Their best alternative]"
their_batna_value: "$Y or equivalent"
zopa: "[Zone Of Possible Agreement: range between our walk-away points]"
my_reservation_price: "[Minimum I'd accept]"
my_aspiration: "[What I actually want]"
their_likely_reservation: "[Best guess at their minimum]"

power_assessment: "I have more power | balanced | they have more power"
# Whoever has the better BATNA has the power

Cialdini's 6 Principles (Influence Audit)

For any persuasion situation, check which levers apply:

3.7 — Technical & Engineering

Build vs Buy Decision Matrix

Shortcut: If it's your core differentiator → build. If it's commodity → buy. Everything else → this matrix.

Reversibility-First Architecture

Design decisions by reversibility:

Phase 4: Cognitive Bias Defense System

Biases are the #1 threat to decision quality. Active defense required.

Daily Bias Checklist (Before Major Decisions)

• Have I actively sought disconfirming evidence?
• Am I anchored to someone else's number/frame?
• Am I continuing because of sunk costs?
• Would I make this same choice starting from zero?
• Have I considered the base rate, not just my situation?
• Has someone challenged this decision?

Phase 5: Decision-Making Under Uncertainty

Confidence Calibration

Before acting on any estimate:

Rule: Most people are overconfident. If you think you're 90% sure, you're probably 70% sure. Adjust down.

Information Value Assessment

Before spending time/money gathering more data:

Decision to make: [X]
Current best guess: [Y]
Current confidence: [Z%]

If I gather [this information]:
- Cost: [$X / Y hours]
- Would it change my decision? [yes / maybe / probably not]
- By how much would confidence increase? [+5% / +15% / +30%]

Value of information = (confidence gain × decision stakes) - gathering cost

Rule: Don't research a $1,000 decision for 40 hours. Match effort to stakes.

When to Decide (Timing Framework)

Phase 6: Group Decision-Making

Structured Disagreement Protocol

For team/partner decisions where people disagree:

1. Independent write-up: Each person writes their recommendation and reasoning (5 min, no discussion)
2. Share simultaneously: Everyone reveals at once (prevents anchoring)
3. Steel man opposition: Each person must articulate the best version of the opposing view
4. Identify cruxes: What's the ONE factual question where if resolved, you'd change your mind?
5. Resolve or decide: If crux is resolvable → get the data. If not → whoever has the best BATNA decides, or the person closest to the information decides.

RACI for Decisions

Common failure: No clear A. If two people think they're the decider, no decision gets made.

Phase 7: Compounding & Systems Thinking

Compounding Mental Model

Most people think linearly. Compounding is the most powerful force:

Linear: 1 + 1 + 1 + 1 = 4 (after 4 periods)
Compounding: 1 × 1.1 × 1.1 × 1.1 × 1.1 = 1.46 (after 4 periods)

But after 50 periods:
Linear: 50
Compounding: 117.39

Rule of 72: Years to double = 72 / growth rate%
- 10% growth → doubles in 7.2 years
- 20% growth → doubles in 3.6 years
- 1% daily improvement → 37x in one year

Application: Every decision should be evaluated for its compounding potential. A decision that creates a 1% improvement to a daily process is worth more than a one-time 50% improvement to an annual process.

Leverage Points (Meadows)

Where to intervene in a system, ranked by effectiveness:

1. Paradigms (most powerful) — Change the mindset/goals of the system
2. Goals — What the system is optimizing for
3. Rules — Incentives, constraints, punishments
4. Information flows — Who knows what, when
5. Feedback loops — Speed and accuracy of response
6. Structure — How components connect
7. Parameters (least powerful) — Numbers, budgets, quotas

Insight: Most people intervene at #7 (adjust the budget). The highest-leverage interventions are at #1-3 (change what we're optimizing for).

Phase 8: Personal Decision-Making

The Energy Audit

Not all decisions need the same energy:

high_energy_decisions: # Use frameworks, sleep on it
  - Career changes
  - Major financial commitments (>10% of net worth)
  - Hiring/firing
  - Market entry/exit
  - Relationship commitments

medium_energy_decisions: # 30-min analysis, then decide
  - Quarterly priorities
  - Tool/vendor selection
  - Pricing adjustments
  - Content strategy

low_energy_decisions: # Decide in <5 min or automate
  - What to eat, wear, read
  - Meeting attendance
  - Social media responses
  - Routine purchases

rule: "Match decision energy to decision stakes. Most people overthink low-energy decisions and underthink high-energy ones."

Default Rules (Eliminate Decision Fatigue)

Create personal defaults so you don't waste energy:

defaults:
  new_meeting_request: "Default NO unless clearly advances top 3 priorities"
  price_negotiation: "Never discount more than 15% — offer value instead"
  new_project: "Default NO unless it replaces something on current list"
  email_response: "Batch 2x/day. Respond in ≤3 sentences or schedule a call"
  investment: "Default index fund. Active only with genuine edge + margin of safety"
  delegation: "If someone can do it 80% as well, delegate"
  saying_yes: "If it's not a HELL YES, it's a no"

Phase 9: Decision Frameworks by Situation

Quick reference — which framework for which situation:

Phase 10: Decision Quality Rubric

Score any decision AFTER making it (or retrospectively):

≥80: Excellent process — outcome is in fortune's hands, not yours 60-79: Good — minor gaps but fundamentally sound 40-59: Mediocre — important dimensions skipped ≤39: Poor — outcome is a coin flip regardless of luck

Critical insight: Judge decisions by PROCESS quality, not outcomes. A good process can produce bad outcomes (variance). A bad process that produces good outcomes is dangerous — it teaches bad habits.

Phase 11: Decision Record Template

Document every significant decision:

decision_record:
  id: "DR-[YYYY-MM-DD]-[number]"
  date: "YYYY-MM-DD"
  decision: "[One sentence — what we decided]"
  type: "1 | 2"
  context: "[Why this decision was needed now]"
  options_considered:
    - option: "[Option A]"
      pros: ["...", "..."]
      cons: ["...", "..."]
    - option: "[Option B]"
      pros: ["...", "..."]
      cons: ["...", "..."]
    - option: "Do nothing"
      pros: ["...", "..."]
      cons: ["...", "..."]
  decision_rationale: "[Why we chose this option]"
  frameworks_used: ["[Framework 1]", "[Framework 2]"]
  key_assumptions: ["[Assumption 1]", "[Assumption 2]"]
  risks_accepted: ["[Risk 1]", "[Risk 2]"]
  success_criteria: "[How we'll know this was right]"
  review_date: "YYYY-MM-DD (when to evaluate)"
  quality_score: "X/100 (Phase 10 rubric)"
  decided_by: "[Name]"
  consulted: ["[Name 1]", "[Name 2]"]
  
  # Fill in at review_date:
  outcome: "[What actually happened]"
  lessons: "[What we learned]"
  would_decide_differently: "yes | no"
  why: "[If yes, what would we change about the PROCESS?]"

Phase 12: Advanced Patterns

Barbell Strategy (Taleb)

Combine extreme safety with extreme risk. Avoid the middle.

Portfolio: 85-90% ultra-safe (treasuries, cash, index) + 10-15% high-risk/high-reward (startups, crypto, moonshots)
Time: 80% predictable deep work + 20% wild exploration/experimentation
Products: Cash cow product (boring, reliable) + speculative bets (innovative, might fail)
Career: Stable income source + asymmetric side projects

Why no middle: The "medium risk" zone gives you medium returns with hidden tail risk. Better to KNOW you're safe on one side and gambling on the other.

Lindy Effect

The longer something has survived, the longer it will likely survive.

Applications:

• Books: A 100-year-old book is more likely relevant in 10 years than a 1-year-old book
• Technologies: SQL (50 years) will outlast this year's hot framework
• Business models: Subscription model (centuries old as concept) > novel monetization
• Advice: Wisdom from 2,000 years ago (Stoics, Sun Tzu) > last week's Twitter thread

Via Negativa (Subtract, Don't Add)

Often the best decision is what to REMOVE, not what to add:

• Remove a feature (focus)
• Remove a meeting (time)
• Remove a client (sanity, team morale)
• Remove a goal (clarity)
• Remove a bad habit (energy)
• Remove complexity (reliability)

Template: "What's the ONE thing I could eliminate that would improve everything else?"

Opportunity Cost Consciousness

Every yes is a no to something else:

If I do X:
- Direct benefit: [value gained]
- Time cost: [hours/days/weeks]
- Best alternative use of that time: [what I'm saying no to]
- Opportunity cost: [value of best alternative forgone]

Net value = Direct benefit - Opportunity cost

If net value is negative, you're destroying value by saying yes — even though it "feels productive."

10 Decision-Making Commandments

1. Classify before analyzing. Type 1 or Type 2? Match process to stakes.
2. "Do nothing" is always an option. Evaluate it explicitly.
3. Seek disconfirming evidence. The moment you like an idea, hunt for why it's wrong.
4. Separate process from outcome. Good process, bad outcome = fine. Bad process, good outcome = lucky.
5. Time-box decisions. Set a deadline. Perfectionism is a form of procrastination.
6. Write it down. Unwritten decisions can't be reviewed, learned from, or challenged.
7. One decider. Every decision needs exactly one person who makes the final call.
8. Sleep on Type 1 decisions. Your brain processes during sleep. Use it.
9. Review decisions. Quarterly, look at your decision records. What patterns emerge?
10. Compound decision quality. Each good decision process makes the next one better. This is the real edge.

10 Common Decision Mistakes

Natural Language Commands

• "Help me decide [X]" → Full decision walkthrough (Quick Start)
• "Score this decision" → Decision Quality Rubric (Phase 10)
• "Pre-mortem [plan]" → Pre-mortem exercise (Phase 4)
• "Is this inside my circle?" → Circle of Competence check
• "Bias check" → Daily Bias Checklist
• "Expected value of [bet]" → EV calculation
• "Map the second-order effects" → Second-order thinking template
• "BATNA analysis for [negotiation]" → Full BATNA template
• "Rate this market" → Porter's Five Forces scoring
• "How strong is the moat?" → Moat Assessment
• "Which framework should I use?" → Phase 9 situation lookup
• "Write a decision record" → DR template (Phase 11)