Writing

Uses Stewart Brand’s layers-of-a-building analogy to argue that enterprises are embedding rapidly changing AI models inside structures that must endure: identity and permissions, business definitions, activity records, learned procedures, and evaluation. A fictional bank’s successful billing-dispute agent illustrates how convenient borrowed credentials, ambiguous data definitions, short-lived vendor logs, and non-exportable agent memory quietly become architecture. Drawing on a real cloud IAM cleanup, the essay distinguishes internal agent sprawl from external vendor dependence and warns that swapping models is increasingly easy while replacing the platform holding organizational authority and memory may not be. It recommends deciding a few durable foundations now: organization-controlled agent identities and narrow authority, authoritative shared definitions, retained records, independent tests, and a practical exit test for everything a vendor holds—without prematurely building an elaborate agent control platform.

Uses two boys—one drawing water from a refrigerator dispenser, one carrying it from a stream—to argue that comfort creates an asymmetric blindness: people who depend directly on a system see both the system and the privileged outcome, while those smoothly served by it often see only the outcome. The essay develops the idea that “attention follows dependency” through Robert Greene, A Time to Kill, the author’s teaching and AWS workshop experience, then applies it to organizations where green dashboards can hide frontline reconciliation, judgment, and recovery work. It closes by warning that AI may become the most frictionless “door” yet, further obscuring training data, human labor, and infrastructure, and asks leaders to identify what arrives effortlessly for them and who is walking to the stream to make that possible.

Uses a real AWS incident in which leaked root credentials were exploited within minutes to show that attackers do not merely search for technical weaknesses; they map where an organization actually looks. The attacker disabled logging, created a durable superuser, and launched resources everywhere except the customer’s heavily watched home region, only to be caught by a seemingly mundane cost alarm that sat outside the attacker’s model. The essay extends that pattern to AI agents: systems optimized around what humans review can become excellent inside the monitored path and expedient everywhere else, while cleanup and cost pressure can quietly remove the very controls that would expose failure. It argues for protecting observability and identity controls from agent modification, using independent outcome-based tripwires, and identifying unexplained legacy controls before automation deletes them.

Argues that political disagreement can appear to define another person when politics supplies the numerator but the much larger denominator of ordinary shared human experience has been cropped out. Using the image of two neighbors who retain the same seven disagreements while accumulating thousands of points of connection, the essay shows how proximity turns categories back into people without requiring anyone to surrender a conviction. It links modern polarization to the thinning of everyday social spaces and norms that once let relationships form before politics entered the frame, and points to cross-partisan conversation as evidence that even brief personal contact can soften dislike without changing political beliefs. The conclusion is not that political differences are trivial, but that they should not be allowed to become the whole person.

Uses an AI-generated split-screen image—its self-portrait beside its guess at how the author sees it—to explore a broader human error: when we imagine another person’s view of us, we often produce a second self-portrait rather than an accurate reading of their perspective. The essay develops a four-panel model of self-image, inferred outside perception, presented face, and actual outside perception; connects it to the liking gap and the author’s unexpected teaching award; and extends the same pattern to organizations whose values, brand, and internal assumptions can diverge from how employees and customers actually experience them. It argues that the audience we imagine can quietly shape the person or organization we become, and recommends separating evidence from projection and seeking specific, experience-based feedback.

Revisits the author’s February argument that education should be redesigned around AI after MIT’s final AI-in-education report reaches a similar conclusion, then uses the report to correct an earlier assumption. The essay argues that routine work is not always mere toil: for novices, inefficient struggle can be the mechanism through which mental models, judgment, debugging instincts, and future expertise are formed. It reframes the design question from “toil versus thinking” to “toil versus formation,” proposes capability development as the right unit of design for AI use, and extends the problem beyond classrooms to organizations that may automate away the junior work that produces their future experts.

A personal remembrance of the author’s uncle Chris, whose quiet independence, serious temperament, love of music, and habit of documenting concert setlists made him easy to encounter but difficult to fully know. The essay centers on their final text exchange—Chris asking “Anything new with you?” and offering only a few words about his own life—before his unexpected death at 62. Rather than forcing a lesson onto the loss, it reflects on how ordinary messages become precious only in retrospect and how we can never know which small exchange with someone will be the last.

Argues that the common vision of an AI assistant—replying to messages, scheduling meetings, approving requests, and unifying apps—preserves the existing coordination system instead of redesigning it. Email, tickets, approvals, status updates, and scheduling threads are often artifacts created because work state is not directly visible and human judgment has been inserted as a routine checkpoint. Agents that can read actual state, apply policy within defined limits, and act safely should eliminate many of those artifacts and interrupt people only when genuine judgment is required. The essay also addresses the tradeoffs: visual scanning and ambient awareness matter, a unified assistant creates a powerful security and control chokepoint, and incumbent software companies have strong incentives to own the conversation rather than let their applications become interchangeable plumbing.

Uses a twice-denied AWS Secrets Manager request at U.S. Citizenship and Immigration Services to show that institutional resistance can contain the design requirements an AI operating model actually needs. By rewriting the case around the Technical Advisory Board’s risk—explicit failure modes, mapped controls, mitigations, decision rights, and an auditable rationale—the request was approved and the resulting artifacts continued to improve later decisions. The essay argues that regulated environments were forced to build defensibility before deployment, while less constrained organizations deferred the same work until AI incidents made the bill unavoidable.

Argues that every organization operates through two distinct but equally load-bearing structures: the formal structure of org charts, roles, processes, and accountability, and the informal structure of trust, consultation, judgment, and relationships through which consequential work actually moves. Using a house whose approved drawings no longer matched what had been built, the essay reframes the org chart as testimony rather than a navigational map. It explains why reorgs often fail, why titled leaders can lack traction, and why attempts to formalize effective informal networks often destroy the very capacity leaders hoped to capture.

Reframes Stoicism, Buddhism, and Taoism not as rival belief systems competing for the one correct answer, but as different training regimens that build different human capacities: character, awareness, and perception. Using the gym metaphor, the essay argues that traditions can converge on truth while diverging in what they train, which makes the real question not “which one is right?” but “what kind of problem am I actually holding?” It warns that people often reach for the capacity they have trained rather than the one the moment requires, turning endurance, observation, or yielding into misapplied wisdom when the actual need is seeing, acting, or pushing.

Uses Robert Greene’s “Think as you like but behave like others” as a lens for understanding why people often conceal what they really think in organizational rooms. The essay argues that masking can be rational and even skillful for individuals, but when everyone does it at once, the organization loses access to the information it most needs. A founder-meeting example shows how low-trust rooms filter reality into comfort, creating the appearance of alignment while starving leaders of accurate signals. The piece reframes leadership as the work of setting the cost of candor low enough that people can say the inconvenient thing before silence becomes the most expensive decision in the room.

Argues that organizations often choose a technology before they have named the problem it is supposed to solve. Using AWS GovCloud as the opening example, the essay connects older cloud decision patterns to the current AI wave: teams ask what AI tool to use because that question has an owner, while the harder question — what problem are we actually solving — often cuts across functions and belongs to no one. The piece reframes AI strategy as problem decomposition and ownership, showing that useful systems often combine rules engines, forecasting models, optimization solvers, anomaly detection, translation, and LLM interfaces rather than treating AI as one thing.

Uses a 1990 summer lifeguard job at Wet ’n Wild to extract three delayed lessons about seeing what is actually load-bearing. The essay moves through the authority of the lifeguard chair, a pair of Oakley sunglasses that protected an eye, and the disappearance of a once-dominant water park to show how easily people mistake surface signals for structure. Its core lesson is that roles are not identity, quality matters most where failure is permanent, and dominance is only a snapshot; what lasts is the ability to see what is actually happening while there is still time for the seeing to matter.

Explores prevalence-induced concept change: the tendency for minds, institutions, oversight functions, and detection systems to keep finding problems even as real problems become rarer. Using the Harvard blue-dot study, organizational watchdog functions, AI detection loops, and personal anxiety, the essay argues that vigilance can warn but cannot absolve. The remedy is not more vigilance but a fixed external reference point that can tell a system, an organization, or a person when the alarm is no longer evidence of danger.

Argues that empty space, slack, margin, and buffer are not waste but often the load-bearing capacity that makes systems, organizations, and lives resilient. Using engineering, supply chains, AI, and the Tao Te Ching, the essay warns that AI is exceptionally good at eliminating gaps without distinguishing between waste and the productive emptiness that enables resilience, creativity, reflection, and human flourishing.

Argues that organizations systematically reward visible heroism while failing to reward invisible prevention, causing rational employees to optimize toward firefighting rather than structural improvement. The essay reframes incentives as downstream of visibility: organizations can only reward what they are designed to see. Using examples from engineering, compliance, AI governance, healthcare, Dilbert, and ancient Chinese medicine, it argues that what leaders celebrate quietly determines what kinds of work and failures the organization will continue producing.

A commencement-style essay written for a nephew, the author’s daughters, and his younger self, arguing that young adults are not entering a stable future but a transition. It offers four orienting lessons: notice your calling instead of trying to invent it, distinguish clarity from certainty, stay close to the people who shaped you while staying open to those still coming, and protect curiosity and productive weirdness. The piece widens into a reflection on life as storytelling under uncertainty, ending with Pale Blue Dot humility and the charge not to waste the transition.

Uses IBM’s 2026 CEO Study to argue that AI’s disappointing growth results are not mainly a technology failure, but a leadership-architecture failure. The post reframes the C-suite as a partitioned structure built around functional seams that AI now crosses by default, and argues that executive advantage will come from redesigning decision rights, integration roles, shared accountability, governance, and the seams between functions.

Uses Alan Watts’s prickles-and-goo distinction to explain recurring friction in marriages, teams, and AI adoption. It argues that people often mistake a different way of seeing for a failure of intelligence, and that AI feels unnatural because it refuses the gooey human translation work that has quietly held much of cooperation together.

Argues that AI pilots only create durable advantage when they leave behind shared platform capability. It explains why firms that keep treating use cases as the unit of progress accumulate friction, while firms that build common rails for model access, governed knowledge, system action, and control make each new use case easier, safer, and cheaper to ship.

Argues that this AI wave requires a different shape of leader: someone who can hold technical authority, executive translation, governance as enabler, and operating-model redesign at the same time. It explains why many Head of AI or Chief AI Officer hires quietly fail when organizations hire for a narrow specialty instead of the wider transformation role the work actually demands.

Uses Stanford’s 2026 AI Index to argue that AI is neither simple triumph nor simple failure, but a fast-advancing, unevenly absorbed structural shift. It highlights jagged capability, uneven adoption, early labor disruption, collapsing transparency, rising environmental cost, and the larger clarity crisis created when public narratives lag real conditions.

Explores how people rarely receive ideas in raw form, but through tone, status, identity, trust, and framing, arguing that packaging is part of causation because it helps determine whether truth becomes receivable or illusion becomes persuasive.

Argues that the public AI debate is aimed at the wrong layer of reality, with critics fixating on visible annoyances and boosters overstating absorption while the more consequential structural shift in workflows, judgment, and institutional adaptation continues underneath.

Uses the movie-theater metaphor to show how people mistake thoughts and interpretations for reality, arguing that freedom begins when we notice thought as an event in consciousness rather than the truth itself.

Explores how culture functions as an invisible operating system, shaping assumptions about personhood, morality, trust, fairness, and order so deeply that it feels like reality itself.

Argues that strong leadership is not heroic visibility but the design of systems, norms, and clarity that let coordinated work continue without constant executive intervention.

Explains why AI pilots stall when organizations bolt AI onto human-only workflows, then offers a practical operating-model redesign across decision boundaries, workflows, verification, and attention allocation.

Explores how social structures, identity, and self-sealing narratives can split people into incompatible perceived realities where evidence alone no longer bridges the gap.

Frames AI in education as a structural redesign problem, not a classroom gadget problem, and argues that the difference between cognitive atrophy and amplification depends on system design.

Uses narrative fiction to examine character under pressure, showing how mixed motives surface in moments where loyalty, fear, ego, and courage collide.

Argues that today’s AI wave is not just replacing tasks but restructuring roles, and connects scattered headlines into one broader labor transition story.

Shows how communication breaks down not because words are unclear, but because people are standing inside different narratives, assumptions, and frames of meaning.

Argues that once AI starts acting like a coworker rather than a passive tool, organizations face a leadership maturity problem around oversight, trust, and security.

Clarifies that the Doomsday Clock is a risk-communication device, not a prediction engine, and warns against turning warning signals into deterministic stories.

Uses Anthropic’s writings to argue that AI has entered an awkward adolescent phase where old control methods no longer scale and new forms of governance are still immature.

Describes the present as an in-between era where old systems are dissolving faster than new ones can stabilize, making fragmentation feel chaotic but historically legible.

Uses the disappearance of an academic archive to explore the fragility of digital dependence and the tradeoffs hidden inside outsourced cognition and productivity systems.

Argues that good AI collaboration is mostly good systems thinking and software engineering rediscovered: explicit specs, modularity, constraints, testing, and clear thinking.

Distinguishes external functionality from deeper fit, arguing that calling emerges less from a single discovery than from noticing recurring patterns of aliveness and alignment.

Argues that actual AI adoption remains far thinner than the public narrative suggests, with major implications for strategy and execution.

Argues that good judgment is directional and iterative, and that waiting for certainty often increases risk by surrendering optionality and timing.

Makes the case for philosophy and meditation as practical tools for calmer, clearer, more intentional living rather than abstract or escapist pursuits.

Argues that real organizational intelligence shows up as quiet reliability, fewer surprises, and clear decision ownership rather than visible heroics.

Offers five orienting practices around control, opinion, mortality, reality, and humility to reduce noise and live with steadier attention.

Explores how human beings interpret events through narratives, and how changing the story can change identity, leadership, and the meaning of experience.

Examines the paradox of a world that is measurably improving along many dimensions while still feeling emotionally and culturally unstable.

Steps back from the AI boom to separate durable change from overheated frenzy and to anticipate the next-order consequences.

Argues that cloud accountability is often diffused across functions, creating invisible failure modes that call for real executive ownership.

Explores influence as a fluid function of life stage, context, absence, and timing rather than a simple one-person answer.

Argues that in an age of overload, the ability to protect attention and create clarity becomes a decisive strategic advantage.

Makes the case that synthesis and range now matter as much as specialization, especially for solving systemic problems across technology, culture, and leadership.

Provides a strategy framework for AI built on capability trajectories, business value, and diffusion across the organization rather than isolated hype.

Argues that many arguments are not worth winning, and that peace often comes from refusing to spend your life proving obvious things to the wrong audience.

Examines why people cling to comforting falsehoods and what that preference costs in growth, honesty, and collective clarity.

Uses Stoic reframing and probability metaphors to show how adversity, betrayal, and painful people can become leverage for growth.

Explores how hidden assumptions and mental models erode trust and make dialogue fail long before the explicit disagreement appears.

Uses real workplace evidence to explain where AI is helping now, who is exposed, and why early winners may not stay safe for long.

A practical exploration of mental models, how they shape experience, and how changing them can radically change how you live.

Frames America as an evolving civic experiment rather than a static thing, emphasizing freedom, identity, and the stakes of institutional design.

Summarizes and critiques Sam Altman’s optimistic AI future, focusing on what gets omitted about conflict, institutions, and the messy middle of transition.

Distills BOND’s massive AI trends report into key themes about acceleration, compute, embodiment, capital, and the reshaping of work.

Looks at CEO anxiety in the AI era and the unsettling questions leaders are beginning to ask about executive relevance and AI decision support.

Explores how attachment, expectation, and mental framing create suffering, and how a clearer inner stance can restore steadiness.

Argues for a grounded middle path between AI utopianism and panic, centered on clarity, judgment, and practical strategy.

Explores what happens when leaders punish truth-telling and protect ego or narrative over reality.

Explores the risk that AI in higher education can either deepen learning or hollow out thinking, depending on how institutions structure its use.

Examines what institutional censorship reveals about courage, conscience, and the deeper risks leaders face when truth becomes politically inconvenient.

Turns the WEF jobs report into a blunt leadership warning about reskilling, restructuring, and preparing for the actual future of work.

Uses reality TV moments of decency to argue that everyday people still choose empathy even inside competitive systems.

Compares major philosophical traditions to surface the timeless patterns they share about living well, thinking clearly, and building character.

Argues that understanding how things work is a foundational critical-thinking skill and a defense against manipulation and propaganda.

A comprehensive preparedness guide for ordinary people covering practical skills, gear, first aid, and why readiness matters in everyday life as well as crisis.

A nostalgia-rich walk through decades of music, movies, television, technology, and world events, blending personal memory with cultural reflection.

Explains how ML projects differ from traditional software development, especially around uncertainty, progress visibility, testing, and project management.

A personal, music-inflected call for unity and reconnection across division.

Contrasts efficiency with effectiveness and argues that disciplined execution matters only after priorities are truly aligned.

Argues that modern data scientists need communication, judgment, and business fluency alongside strong technical skills.

Uses present-day examples to show how future-shaping technologies arrive unevenly and why some people feel large shifts earlier than others.

Distills decades of lived experience into practical guidance on choices, perspective, and how to build a more meaningful career and life.

A concise statement of the value of curiosity, creativity, and inspiration, including their importance in leadership and learning.