Mental Models Deep Dive: Leadership

May 14, 2026 · Day 15

Servant Leadership Servant Leadership

[In Depth]

Servant Leadership was proposed by Robert Greenleaf in the 1970s. Its core claim: the leader's first job is not to issue orders, but to serve the growth and success of team members. The leader influences through listening, empathy, healing, awareness, and persuasion — not through positional authority and control. Power flows bottom-up: the leader clears obstacles, supplies resources, and creates conditions in which each person can reach their full potential.

Classic example Southwest Airlines' legendary CEO Herb Kelleher was a benchmark servant leader. He believed deeply in "employees first, customers second, shareholders third" — when employees feel cared for and respected, they spontaneously deliver outstanding service, and profit follows. Kelleher personally loaded baggage at Thanksgiving, remembered thousands of employees' names, and called frontline staff on Christmas to thank them. Under his leadership, Southwest stayed profitable for 47 consecutive years — the only US airline that survived multiple economic downturns without losing money. He proved that when a leader defines their role as "serving the team" rather than "the team serving me," the organization unlocks startling collective energy.
BigCat scenario In the super-individual era, a leader is more like an "enablement node." When you use Claude or other AI tools to help teammates speed up and unstick technical problems, you're practicing servant leadership — you aren't replacing them, you're equipping them with a "cognitive exoskeleton." The same applies to parenting: when your kid is stuck on a hard math problem, rather than handing over the answer, guide them to the solution path and provide just enough scaffolding. You serve the process of growth, not the result.

[English Summary]

Servant Leadership inverts the traditional hierarchy: the leader's primary role is to serve the team, removing obstacles and fostering each member's growth. Authority flows from trust and empowerment, not positional power. In the AI era, this translates to being an "enablement node" — equipping your team (or your children) with tools, context, and autonomy rather than micromanaging outcomes.

[AI Prompts]

English Prompt: I manage a small cross-functional team building an AI product. Using the Servant Leadership framework, design a weekly routine that ensures I spend more time unblocking my team than directing them. Include specific rituals, questions to ask in 1-on-1s, and metrics to gauge whether my "serving" is actually accelerating the team's output.

Situational Leadership Situational Leadership

[In Depth]

Situational Leadership was proposed by Hersey & Blanchard. Its core claim: there is no one-size-fits-all leadership style. The most effective approach depends on the follower's "readiness" — their competence and willingness on a specific task. The theory defines four styles: Directing (high task, low relationship), Coaching (high task, high relationship), Supporting (low task, high relationship), and Delegating (low task, low relationship). The leader must shift fluidly as the team member develops.

Classic example Learning to drive walks neatly through all four stages. On the first lesson, the instructor must be Directing — "clutch in, first gear, ease the clutch out while feathering the gas" — every step spelled out, because the learner has neither skill nor confidence. After a few lessons, with basic operation in hand but confidence still shaky, the instructor switches to Coaching — demonstrating proper lane changes while asking, "how would you handle this intersection?" When the learner can drive independently but still tenses up at moments, the instructor moves to Supporting — sitting quietly in the passenger seat, offering encouragement or a nudge only at key moments. Finally, once the learner is fully competent, the instructor delegates: "drive yourself, call if you have a problem." At each stage the style must match the learner's stage — letting go too early causes accidents; letting go too late suffocates growth.
BigCat scenario This model is invaluable in investment decisions and human-AI collaboration. When you first start using a new AI tool, you need Directing — precise prompts, careful review of every output. As you grow familiar with the tool's capability boundary, you can shift to Delegating — give a high-level goal and trust the AI on the details. The same with kids: teaching a child to ride a bike goes from hands holding the seat (Directing), to running alongside (Coaching), to watching from a distance (Supporting), to letting them ride to the park alone (Delegating). The key: your style should track the other party's growth, not lock into one mode.

[English Summary]

Situational Leadership holds that no single leadership style is universally optimal. The leader must adapt — directing novices, coaching developing performers, supporting capable-but-hesitant individuals, and delegating to self-reliant experts. The art lies in accurately diagnosing readiness and fluidly shifting style. This applies equally to managing people, guiding children's learning, and calibrating how much autonomy to give AI tools in your workflow.

[AI Prompts]

English Prompt: I'm onboarding a junior colleague who will use AI coding assistants daily. Map out a 30-day Situational Leadership plan: which leadership style (Directing → Coaching → Supporting → Delegating) should I use in each phase, what milestones signal it's time to shift, and what specific check-ins or exercises will accelerate their progression toward autonomous AI-augmented development?

Circle of Competence Circle of Competence

[In Depth]

Circle of Competence, a concept Buffett and Munger return to repeatedly, says everyone has a domain they truly understand — and what matters isn't the size of that circle, but whether you know precisely where its boundary is. Inside the circle, you have a judgment edge; once you cross the boundary, even genius-level intelligence can make beginner mistakes. The discipline demands three things: (1) honest identification of what you actually understand; (2) decision-making inside the boundary; (3) strategic, gradual expansion of the boundary.

Classic example For decades, Warren Buffett refused to invest in tech stocks, even when everyone around him during the 1990s dot-com bubble was calling him "out of date." His reason was simple: technology was outside his circle of competence — he couldn't tell which company would still have a moat in ten years. Investors piled into Pets.com, Webvan, and the like; Buffett didn't move. When the bubble burst, the "smart money" took massive losses, and Berkshire emerged unscathed — and then went bargain-hunting. It wasn't until 2016 that he finally bought Apple — by then Apple wasn't a pure tech company but a consumer brand with deep loyalty and ecosystem lock-in, which fit squarely inside Buffett's circle. The lesson: knowing what you don't know is more valuable than knowing what you do know.
BigCat scenario For someone pursuing super-individual status, AI is a super-lever for expanding the circle — but it carries a deep trap. When Claude generates a quant trading strategy in code for you, you can feel the illusion of "I understand quant investing," when in fact you only have a tool, not understanding. This is exactly where first principles enters: you have to distinguish "AI did this for me" from "I actually understand this." In investing, if you've spent years deep in AI/tech, that's your circle — size positions accordingly. But when you face biotech or commodities, where you lack deep cognition, stay humble and small-sized even with AI-assisted analysis. The circle isn't static — deliberate practice and cross-disciplinary study expand it, just much slower than you think.

[English Summary]

The Circle of Competence, championed by Buffett and Munger, states that knowing the boundary of what you truly understand is more important than the size of that circle. Operate inside it for high-conviction decisions; respect the boundary when venturing outside. AI tools can extend your reach but not your understanding — confusing the two is a dangerous cognitive trap. Expand your circle deliberately through deep study, not by outsourcing judgment to algorithms.

[AI Prompts]

English Prompt: I'm an investor who primarily understands AI/tech but occasionally evaluates biotech and climate-tech deals. Using the Circle of Competence framework, help me build a pre-investment checklist that forces me to honestly assess whether a given opportunity falls inside, at the edge, or outside my circle — and what due-diligence steps differ for each zone.

Delegation & Trust Delegation & Trust

[In Depth]

Delegation isn't just throwing tasks over the wall — it's a precise system balancing trust, accountability, and control. Effective delegation follows the "outcome-based" rule: be explicit about the desired result and constraints, but hand the choice of method to the executor. Stephen Covey split delegation into two kinds: "gofer delegation" (tell them every step) and "stewardship delegation" (clear goals, standards, consequences — but the method is theirs). High-performing leaders use almost only the latter.

Classic example Netflix's "Freedom & Responsibility" culture is the extreme case. Founder Reed Hastings dismantled almost every traditional control mechanism: no fixed vacation days (employees decide), no travel reimbursement approval (just "act in Netflix's best interest"), and even no decision-approval hierarchy — any employee can make a major call on their own judgment, provided they own the consequences. This aggressive delegation selected for highly self-driven top performers and ejected those who needed external supervision to function. The result: Netflix grew from a DVD-by-mail company into a global streaming giant worth over $300B. Hastings proved that when you give enough trust and freedom, excellent people repay it with creativity and accountability far beyond expectation.
BigCat scenario In the human-AI collaboration era, the delegation-and-trust framework gets redefined. Every time you use Claude, you're making delegation decisions: how much autonomy do I give the AI for this task? Where do I insert human review? It's structurally identical to managing a team. Best practice is "progressive trust" — test reliability on low-stakes tasks first, build a trust baseline, then gradually widen the delegation. The same way you wouldn't let a new hire sign contracts on day one, you shouldn't fully trust AI output unverified. In parenting too: let your kid plan their weekend study schedule, with only the constraint "at least finish math and reading." Which to do first, how — trust their judgment. Each successful autonomous decision strengthens intrinsic motivation and self-confidence.

[English Summary]

Effective delegation is not abdication — it is a structured transfer of decision-making authority within clear boundaries. Define the desired outcome, constraints, and accountability, then grant full autonomy on the method. This framework maps directly onto human-AI collaboration: calibrate how much latitude to give AI tools through incremental trust-building, starting with low-stakes tasks and expanding as reliability is proven. The same principle applies to parenting — outcome-based delegation nurtures intrinsic motivation.

[AI Prompts]

English Prompt: Design a "Trust Ladder" framework for a parent managing a 9-year-old's increasing independence. Define 5 levels — from full supervision to full autonomy — with concrete examples for each level (homework, screen time, social plans, money). For each level, specify the trigger conditions for promotion, the guardrails that remain, and the recovery protocol when trust is broken.