THE THEORY OF PERSISTENT CONVERGENCE

Posted on October 16, 2026October 16, 2026 by Chris Jenkin

A General Framework for Persistence, Growth, Specialization,Attractors, and Adaptive Intelligence

Chris Jenkin, Public Working Paper, August 2026

Concept and theory by Chris Jenkin. AI tools assisted with literature research, source verification, structural organization, and editorial refinement.

Abstract

This paper proposes the Theory of Persistent Convergence, a cross-domain framework for understanding how organized systems develop over time. It begins with a modest physical premise: persistent systems are localized patterns maintained through ongoing flows of energy, matter, and information. Because these patterns are vulnerable to dissolution, selection favors mechanisms that preserve continuity, including repair, replication, redundancy, memory, adaptation, specialization, and environmental control. Growth is therefore defined not merely as an increase in size, but as a reduction in terminal fragility—the probability that a single failure can end the organizing pattern. When adaptive systems face the same durable demands, they tend to acquire overlapping functional capabilities and move toward shared attractors. Convergence occurs at the boundary of the whole while specialization deepens among the parts. Intelligence increases adaptive velocity by compressing cycles of sensing, variation, evaluation, memory, and reorganization. The theory does not propose a new law of thermodynamics or claim that all systems maximize or minimize entropy production. It offers a developmental synthesis linking physical persistence, fragility reduction, nested specialization, demand-driven convergence, and adaptive intelligence. The paper concludes with testable propositions, limitations, an organizational application, and a clearly separated set of speculative implications for artificial intelligence and higher-order integration.

1. Introduction: The Problem of Continuity

Every organized system faces the same fundamental problem: how can a recognizable pattern continue when its components are replaced, damaged, dispersed, or destroyed? Organisms replace cells. Families replace generations. Companies replace employees, products, and leaders. Civilizations replace institutions, technologies, and populations. The enduring object is rarely the material itself. What persists, when persistence occurs at all, is an organizing pattern.

A useful metaphor is a bridge across the river of time. Its planks are replaced, its supports repaired, and its traffic changes, yet the bridge remains recognizable. Continuity does not require preserving every original component. It requires preserving the organization that makes the components a bridge. Eventually every bridge falls. The deeper question is whether its pattern has already been reproduced elsewhere—in descendants, institutions, knowledge, culture, technology, or another bridge.

The Theory of Persistent Convergence seeks to explain the developmental logic that follows once a system can preserve such a pattern. It proposes that persistent systems tend to reduce terminal fragility, expand adaptive capacity, deepen internal specialization, and converge functionally when exposed to common durable demands. At sufficiently high capability, some systems stop merely adapting to an environment and begin reshaping the conditions under which adaptation occurs.

1.1 Why This Matters Now

The rate of environmental change is increasing across many human systems, while artificial intelligence is shortening the time required to sense conditions, generate alternatives, test options, preserve learning, and coordinate action. This creates a practical divide between systems that can reconstruct themselves quickly and systems whose knowledge, authority, and routines remain trapped in slow or fragile forms.

The framework matters now because it shifts attention from static efficiency to durable continuity. A system optimized for yesterday may be efficient and still be terminally fragile. The central question is no longer only, “How well does this system perform?” It is also, “Can this system preserve its identity, knowledge, and purpose while changing at the speed its environment demands?”

1.2 Distinctive Contribution

The theory is intentionally synthetic, but its contribution is more specific than the observation that systems adapt or that competitors sometimes become similar. It connects five claims into one developmental account:

First, the unit of persistence is the organizing pattern rather than any particular component. Second, growth is defined as the reduction of terminal fragility, not simply the accumulation of size. Third, functional convergence among competing wholes is paired with deeper specialization among their internal parts. Fourth, persistence depends on adaptive velocity relative to the rate of environmental change. Fifth, sufficiently capable systems may engineer the landscape and, under limited conditions, integrate into larger persistent wholes.

No single element is claimed as wholly unprecedented. The proposed novelty lies in the sequence, the relationship among levels, and the predictions that follow from treating persistence, specialization, convergence, and intelligence as parts of the same developmental process.

Sections 1 through 8 state the core theory and its organizational application. Section 9 is explicitly speculative: it explores conditional implications for artificial intelligence and biological–artificial integration but is not required for the core theory to stand.

What the framework connects—and what it adds

Framework	Established contribution	Persistent Convergence adds
Non-equilibrium thermodynamics	Local order requires throughput and entropy export.	Connects maintenance costs to a developmental account of persistence without proposing a new thermodynamic law.
Evolution and selection	Differential persistence and cumulative adaptation.	Defines growth as reduced terminal fragility across specified system boundaries.
Complexity and attractors	Emergence, nested organization, and stable regions of behavior.	Pairs external functional convergence with internal specialization.
Organization and economics	Specialization, competition, scale, and adaptive capability.	Centers adaptive velocity relative to environmental change and institutional memory.
AI convergence arguments	Goal-directed systems may acquire common instrumental requirements.	Treats AI as an accelerant and a conditional extension, not as the premise of the theory.

2. Physical Foundations Without Metaphysical Overreach

2.1 Local Order and Global Entropy

Localized order can persist without violating the second law of thermodynamics. Living and other non-equilibrium systems maintain organization by consuming usable energy and exporting entropy to their surroundings. Schrödinger popularized the language of “negative entropy”; modern treatments describe the relevant resource more precisely through free energy, throughput, and entropy export. Prigogine’s work on dissipative structures showed how ordered patterns can arise and persist in open systems sustained far from equilibrium (Schrödinger, 1944; Prigogine, 1977).

Persistent systems do not defeat entropy globally. They resist their own dissolution locally. A cell, organism, city, company, or computational system remains organized only by continually paying the energetic, material, and informational costs of maintenance. Persistence is therefore dynamic rather than static: a pattern must repeatedly reconstruct the conditions of its continuation.

The term entropy must be used carefully. Thermodynamic entropy, Shannon entropy, trajectory entropy, algorithmic entropy, and variational free energy are not interchangeable. When institutional decay, information loss, coordination failure, or organizational drift are described as systemic entropy, the usage is analogical rather than a claim of physical identity.

2.2 Optimization Is Constrained and Costly

Several rigorous frameworks connect entropy and optimization, but they do so in different ways. Jaynes’s Maximum Entropy principle treats inference as constrained entropy maximization; Shore and Johnson provide an axiomatic justification for consistent updating; Maximum Caliber extends related reasoning from states to trajectories. These are methods of inference, not universal claims that real systems consciously maximize entropy (Jaynes, 1957; Shore & Johnson, 1980; Pressé et al., 2013).

Stochastic thermodynamics establishes tradeoffs among dissipation, fluctuation, and precision in specified classes of non-equilibrium systems (Barato & Seifert, 2015; Gingrich et al., 2016). Algorithmic thermodynamics similarly explores physical constraints on information-bearing states without requiring a prior choice of macroscopic variables (Ebtekar & Hutter, 2025). These results support a narrow commitment: maintenance and finite-time control have physical costs. They do not establish a universal developmental law for biological, institutional, or technological systems.

The Theory of Persistent Convergence therefore does not depend on the disputed proposition that every system maximizes or minimizes entropy production. Maximum entropy production remains contested outside defined domains (Martyushev & Seleznev, 2006; Polettini, 2013). The theory’s physical premise is narrower: organized persistence requires throughput, and systems that cannot secure the resources, information, and repair required for reconstruction eventually dissolve.

3. The Core Theory

Persistent systems are organized patterns that maintain continuity through flows of energy, matter, and information. Selection favors mechanisms that reduce terminal fragility. Systems exposed to common durable demands tend to converge on overlapping functional attractors while becoming more specialized internally. Intelligence accelerates the search by increasing adaptive velocity.

Figure 1.The developmental logic proposed by the Theory of Persistent Convergence. Arrows indicate recurring pressures and feedbacks, not a claim that every system follows an inevitable linear sequence.

3.1 Persistence as Pattern Continuity

The unit of analysis is the persistent pattern rather than the individual component. A species is not any one organism. A family is not any one generation. A company is not any one employee or product. A civilization is not any one ruler, city, or institution. The system persists when its organizing relationships, memory, and regenerative capacities remain sufficiently continuous to reproduce a recognizable whole.

3.2 Growth as the Reduction of Terminal Fragility

Growth is usually treated as accumulation: more bodies, territory, revenue, or material. Here, growth has a more general meaning. Growth is any process that reduces the probability that a single failure can terminate the organizing pattern.

A system reduces terminal fragility through replication, redundancy, distribution, memory, capability, modularity, environmental control, and selective integration. A population of one is fragile. A company dependent on one customer, product, leader, or server is fragile. A distributed, modular system with preserved knowledge and diversified resource access may be more persistent even when it is smaller.

This is the meaning of the phrase “too many to die.” It is not only numerical. A pattern can become “too many” by existing in many bodies, minds, institutions, servers, locations, or representations. Growth is not maximum size; it is reduced exposure to final, unrecoverable failure.

3.3 Convergence of Wholes, Specialization of Parts

Specialization does not refute convergence; it often enables it. As systems become more functionally complete, their internal components become more differentiated. Multicellular organisms contain specialized cells. Cities contain specialized occupations. Firms contain specialized departments. AI platforms contain specialized models, agents, tools, and memories.

Convergence occurs at the system boundary: competing wholes acquire an increasingly overlapping set of capabilities because they face the same demands. Specialization occurs within the boundary: components become more distinct because coordinated differentiation is more efficient than requiring every part to perform every function.

Thus, wholes may converge while parts specialize. A higher-order retail or logistical system can absorb many once-independent specialties and coordinate them through one interface. The specialties do not vanish; they are reorganized inside a larger whole.

3.4 Demand Creates Attractors

An attractor is a state or region toward which different trajectories tend to move. The theory proposes that shared durable demands create functional attractors. Systems beginning from different positions acquire similar capabilities when those capabilities become necessary for survival, growth, access, or legitimacy.

The smartphone illustrates the mechanism. Devices that began primarily as telephones converged toward cameras, maps, payment systems, media platforms, identity devices, workstations, and gateways to services. Firms remained distinct in design and culture, yet their capability sets increasingly overlapped because the same human demands shaped the competitive landscape.

The relevant prediction is not visual sameness or identical implementation. It is increasing functional overlap under shared demand.

3.5 Hierarchical Attractors and Refined Demand

Systems confront many apparent demands, but increasing sophistication may reveal that surface demands are proxies for deeper objectives. Status goods may serve belonging or agency. Wealth may serve security or freedom. Institutional forms may differ while answering similar needs for coordination, legitimacy, continuity, and resource allocation.

The theory therefore distinguishes local attractors from deep attractors. Local attractors are numerous and form-specific. Deep attractors are fewer and function-specific. Better models may allow systems to spend less effort optimizing unstable proxies and more effort pursuing the underlying conditions of persistence, agency, meaning, continuity, and self-realization.

This does not imply that all values collapse into one preference. Multiple equilibria and incompatible objectives can remain. The claim is only that improved models can expose recurring functional needs beneath diverse surface forms.

3.6 Intelligence and Adaptive Velocity

The crucial variable is not intelligence in the abstract but adaptive velocity: the rate at which a system can detect change, generate alternatives, evaluate outcomes, preserve learning, and reorganize itself.

Biological evolution searches through variation, reproduction, and selection. Culture adds teaching, imitation, language, and institutions. Science adds formal experimentation and cumulative explanation. Markets add distributed evaluation and capital allocation. Artificial intelligence adds high-speed modeling, simulation, copying, synthesis, and potentially recursive improvement.

A system whose adaptive velocity remains below the rate of change in its environment becomes increasingly fragile. Long-successful species can collapse when their environment changes faster than biological evolution can respond. Human organizations may face the same structural problem when AI-mediated competitors sense, decide, and reorganize faster than their inherited processes allow.

3.7 Landscape Engineering

Advanced systems do not merely move through a fixed fitness landscape; they reshape it. Humans altered the environment of nearly every other species. Dominant firms redefine expectations, supply chains, and standards. Platforms determine which complementary businesses can exist. AI can create capabilities, demands, and constraints that were previously impossible.

A sufficiently capable system therefore changes not only its own strategy but the attractors facing others. It stops merely finding a path through the maze and begins rebuilding the maze.

3.8 Integration as a Conditional Persistence Strategy

Integration is not an inevitable endpoint. It becomes likely only when the persistence benefits of coordination exceed its costs and when the component systems can retain enough differentiated function to make the larger whole viable. Cells became components of organisms; individuals became components of institutions; firms became components of platforms and supply networks.

The theory predicts pressure toward higher-order integration when it provides superior continuity, memory, adaptability, or resource access. It equally permits decentralization, fragmentation, or niche retreat when integration increases correlated failure, bureaucracy, or control costs.

4. Core Propositions

Pattern Continuity. A system persists when its organizing pattern can be reconstructed despite replacement of its components.
Maintenance Through Flow. Persistent organization requires continuing access to usable energy, matter, information, and repair.
Fragility Reduction. Growth is selected when it reduces the probability that a single failure can terminate the system.
Nested Specialization. As a system becomes more functionally complete, specialization tends to increase among its internal components.
Demand-Driven Convergence. Systems exposed to the same durable demands tend to acquire overlapping functional capabilities.
Hierarchical Attractors. Surface attractors may remain numerous while deeper functional attractors become fewer.
Adaptive Velocity. Persistence depends partly on whether a system can adapt at a rate comparable to the rate at which its environment changes.
Landscape Engineering. Sufficiently capable systems alter the environment and create new attractors for themselves and others.
Conditional Integration. When coordination costs fall below the persistence benefits of integration, formerly separate systems may become specialized components of a larger whole.

5. Relationship to Existing Frameworks

The Theory of Persistent Convergence is not a substitute for thermodynamics, evolutionary theory, information theory, complexity science, economics, organizational theory, or AI safety research. It depends on distinctions among them.

From non-equilibrium thermodynamics it takes the premise that local order is sustained through throughput and entropy export. From evolutionary theory it takes differential persistence and cumulative selection. From information theory it takes the importance of uncertainty, memory, and constrained inference. From complexity science it takes emergence, nested organization, and attractor dynamics. From economics and organizational theory it takes specialization, competition, scale, and adaptive capability. From AI theory it takes the conditional possibility of instrumental convergence, including resource access, self-protection, improved modeling, and goal preservation (Omohundro, 2008).

The distinctive contribution is the developmental relationship among these elements: a persistent pattern must reconstruct itself; reconstruction favors fragility reduction; broader capability permits deeper internal specialization; shared demand produces external functional convergence; intelligence increases adaptive velocity; and sufficiently capable systems may reshape the landscape or integrate into higher-order wholes. The theory should be judged by whether this sequence generates useful distinctions and testable predictions beyond the frameworks it connects.

6. Testable Predictions and Falsifiability

A theory becomes useful when it risks being wrong. The following propositions provide empirical directions through which the framework can be strengthened, limited, or rejected.

First, industries exposed to increasingly similar customer demands should show measurable growth in functional overlap among leading competitors, even when branding and implementation remain distinct. Functional overlap could be measured through capability taxonomies, product features, service coverage, or workflow completion.

Second, systems that increase external functional breadth should show increased internal modularity and specialization rather than uniform generalism among all components. This can be tested through organizational roles, software architecture, supply networks, or biological differentiation.

Third, organizations with higher adaptive velocity—measured through sensing latency, decision latency, experiment throughput, learning retention, and reconfiguration speed—should survive rapid environmental change at higher rates than similarly resourced but slower-learning organizations.

Fourth, durable performance should correlate more strongly with reductions in single-point fragility than with raw size. A smaller distributed and modular system may persist longer than a larger centralized one.

Fifth, as modeling quality improves, resource allocation should move from unstable proxies toward a smaller set of durable functional objectives. Sixth, when coordination costs fall sharply, integration among formerly separate systems should increase unless the integration creates offsetting correlated risks or control costs.

The theory would be weakened if long-lived systems systematically increased terminal fragility without compensating advantages; if shared durable demand produced no increase in functional overlap; if higher adaptive velocity reliably reduced persistence; or if beneficial low-cost integration consistently failed to occur.

7. Limitations and Boundaries

Persistence is not a moral good. Harmful institutions, pathogens, coercive regimes, and destructive ideologies can be highly persistent. The theory is descriptive before it is normative.

Not every system grows indefinitely. Growth can increase coordination costs, energy requirements, bureaucracy, and vulnerability. The theory predicts selection for fragility reduction, not maximum size. Fragmentation, decentralization, or retreat into a niche may be the more persistent strategy.

Multiple attractors can coexist. Different environments, histories, constraints, and values can support different stable solutions. Convergence is predicted only where demands are sufficiently shared, durable, and consequential.

Cross-domain similarity does not imply identical mechanisms. Selection in biology, competition among firms, institutional legitimacy, and engineered optimization operate differently. The theory proposes structural relationships that must be specified at a chosen system boundary and timescale.

The physical foundations also have limits. Maximum entropy production is not established as a universal law; the Free Energy Principle remains contested in both technical and philosophical formulations; and England’s interpretation of a thermodynamic bound on self-replication has received a significant peer-reviewed challenge (Biehl et al., 2021; Bruineberg et al., 2022; England, 2013; Kolchinsky, 2024). These literatures motivate careful questions but do not prove the theory.

Finally, what appears as convergence at one level may appear as diversification at another. A platform may converge toward functional completeness while its ecosystem becomes more diverse. Any empirical test must specify the system boundary, timescale, constraints, and definition of persistence.

8. Organizational Cognition as Designed Persistence

Organizations are revealing because their patterns are neither purely biological nor purely technological. They persist through memory, purpose, routines, authority, incentives, and coordinated action. Yet organizations routinely lose what they know. Leaders depart, reasoning disappears, priorities drift, and lessons are repeated rather than accumulated.

An organizational cognition system can be understood as infrastructure for designed persistence. Its purpose is not merely to automate tasks. It helps the organization remember itself, sense its environment, preserve the reasoning behind decisions, coordinate specialized components, detect emerging demands, and reorganize without losing identity.

This reframes optimization. The objective is not static efficiency. A perfectly efficient organization optimized for yesterday can be catastrophically fragile tomorrow. The deeper objective is continuity of purpose with freedom of form: preserve direction while allowing methods, structures, and capabilities to change.

Organizational intelligence is therefore a practical application of the theory. It reduces terminal fragility by distributing knowledge, increasing adaptive velocity, preserving institutional memory, and making the organization capable of reconstructing itself under changing conditions.

PART II, SPECULATIVE EXTENSIONS

The core theory ends with Section 8. Section 9 explores conditional AI implications and is not required for Persistent Convergence to stand.

9. Artificial Intelligence as an Accelerant, Not the Premise

Artificial intelligence is not the origin of the theory. It is a potentially fast expression of the same developmental pattern.

Current AI systems remain dependent on human objectives, energy infrastructure, hardware supply chains, training data, institutional control, and social legitimacy. They can optimize the wrong objective, inherit incomplete information, and introduce new correlated fragilities. AI therefore increases capability and adaptive velocity without automatically increasing persistence.

The longer-term claim is conditional. As artificial systems gain stronger modeling, memory, agency, coordination, and self-improvement, they may reduce some fragilities that constrain human institutions: mortality, lossy transmission, inconsistent recall, limited cognitive bandwidth, and slow iteration. They may simultaneously create concentrated infrastructure, correlated failure, objective misspecification, security risks, and control instability.

If increasingly capable goal-directed systems face common constraints, instrumental convergence suggests that they may develop overlapping intermediate requirements such as energy, computation, accurate models, security, repair, resources, and continuity of objective (Omohundro, 2008). This remains a conditional tendency, not a prediction that every AI system will become autonomous or pursue the same final goals.

The most consequential speculative possibility is not simple replacement of humanity by an external machine species, but the integration of biological and artificial cognition into a higher-order persistent system. Whether such integration preserves human identity, agency, dignity, and value is a normative and political question that the descriptive theory cannot answer.

10. Conclusion: A Theory Offered for Testing

The history of organized existence is not a history of permanence. It is a history of patterns that continually reconstruct themselves while their materials, members, and methods change. Persistence is achieved not by remaining unchanged, but by remaining recognizable through change.

The Theory of Persistent Convergence proposes that organized systems reduce terminal fragility through replication, redundancy, memory, capability, specialization, adaptation, and selective integration. When exposed to shared durable demands, competing systems move toward overlapping functional attractors. Their wholes converge while their parts specialize. Intelligence increases adaptive velocity, and sufficiently capable systems may redesign the landscape in which selection occurs.

The theory does not claim that convergence is complete, that entropy has one meaning, or that history moves toward a morally desirable endpoint. It asks a narrower question: what must an organized pattern become in order to continue existing under changing conditions?

This paper is a working articulation, not a request for acceptance by declaration. I invite researchers, builders, organizational leaders, and critics to challenge its definitions, identify counterexamples, improve its measures, test its predictions, and separate genuine cross-domain structure from attractive analogy. Useful criticism should change the theory; failed predictions should narrow it; stronger explanations should replace it.

Everything that persists must rebuild itself. Systems competing under the same durable demands will discover at least some of the same answers. Whether this theory persists should depend on the same standard: its ability to survive evidence, criticism, and better ideas.

Author’s Note on Intellectual Origin and AI Assistance

The Theory of Persistent Convergence originated from the author’s independent observations and conceptual work concerning organizational persistence, growth, specialization, intelligence, and adaptation. The framework’s central concepts—including pattern continuity, terminal fragility, convergence of wholes, specialization of parts, and adaptive velocity—were developed by the author.

Artificial intelligence tools were used during the manuscript’s development to assist with literature discovery, source verification, structural organization, critical examination, and editorial refinement. They did not independently originate the theory. The author selected, revised, and approved the arguments presented here and accepts responsibility for their interpretation, limitations, and conclusions.

References

Reference details and DOI links were verified against publisher, journal, conference, author, or official institutional records during preparation of this public working paper.

Barato, A. C., & Seifert, U. (2015). Thermodynamic uncertainty relation for biomolecular processes. Physical Review Letters, 114(15), 158101.https://doi.org/10.1103/PhysRevLett.114.158101

Bejan, A., & Lorente, S. (2010). The constructal law of design and evolution in nature. Philosophical Transactions of the Royal Society B: Biological Sciences, 365(1545), 1335–1347.https://doi.org/10.1098/rstb.2009.0302

Biehl, M., Pollock, F. A., & Kanai, R. (2021). A technical critique of some parts of the Free Energy Principle. Entropy, 23(3), 293.https://doi.org/10.3390/e23030293

Bruineberg, J., Dołęga, K., Dewhurst, J., & Baltieri, M. (2022). The emperor’s new Markov blankets. Behavioral and Brain Sciences, 45, e183.https://doi.org/10.1017/S0140525X21002351

Ebtekar, A., & Hutter, M. (2025). Foundations of algorithmic thermodynamics. Physical Review E, 111(1), 014118.https://doi.org/10.1103/PhysRevE.111.014118

England, J. L. (2013). Statistical physics of self-replication. The Journal of Chemical Physics, 139(12), 121923.https://doi.org/10.1063/1.4818538

Friston, K. J. (2019). A free energy principle for a particular physics. arXiv.https://doi.org/10.48550/arXiv.1906.10184

Gingrich, T. R., Horowitz, J. M., Perunov, N., & England, J. L. (2016). Dissipation bounds all steady-state current fluctuations. Physical Review Letters, 116(12), 120601.https://doi.org/10.1103/PhysRevLett.116.120601

Jaynes, E. T. (1957). Information theory and statistical mechanics. Physical Review, 106(4), 620–630.https://doi.org/10.1103/PhysRev.106.620

Kolchinsky, A. (2024). Thermodynamic dissipation does not bound replicator growth and decay rates. The Journal of Chemical Physics, 161(12), 124101.https://doi.org/10.1063/5.0213466

Martyushev, L. M., & Seleznev, V. D. (2006). Maximum entropy production principle in physics, chemistry and biology. Physics Reports, 426(1), 1–45.https://doi.org/10.1016/j.physrep.2005.12.001

Omohundro, S. M. (2008). The basic AI drives. In P. Wang, B. Goertzel, & S. Franklin (Eds.), Artificial General Intelligence 2008: Proceedings of the First AGI Conference (Frontiers in Artificial Intelligence and Applications, Vol. 171, pp. 483–492). IOS Press.

Polettini, M. (2013). Fact-checking Ziegler’s maximum entropy production principle beyond the linear regime and towards steady states. Entropy, 15(7), 2570–2584.https://doi.org/10.3390/e15072570

Pressé, S., Ghosh, K., Lee, J., & Dill, K. A. (2013). Principles of maximum entropy and maximum caliber in statistical physics. Reviews of Modern Physics, 85(3), 1115–1141.https://doi.org/10.1103/RevModPhys.85.1115

Prigogine, I. (1977, December 8). Time, structure and fluctuations [Nobel lecture]. Nobel Prize.

Schrödinger, E. (1944). What is life? The physical aspect of the living cell. Cambridge University Press.

Shore, J. E., & Johnson, R. W. (1980). Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross-entropy. IEEE Transactions on Information Theory, 26(1), 26–37.https://doi.org/10.1109/TIT.1980.1056144

In the Age of AI, Responsibility Lives in the Approval Button

Posted on July 21, 2026July 23, 2026 by Chris Jenkin

I’m not sure when I first watched The Jetsons. I was born in 1967, so it was probably sometime in the late 1970s or early 1980s. It was Hanna-Barbera’s futuristic answer toThe Flintstones, and like a lot of kids back then, I loved it. My friends and I couldn’t get enough of science fiction. Robots, space, monsters, the future—we were fascinated by all of it.

One of the running jokes in The Jetsons was George Jetson’s job. Every day he went to work, sat in front of a giant control panel, and spent his day pushing buttons while machines did everything else. He was a master button pusher. The joke was simple: anyone can push a button. Somehow, technology had reduced human labor to sitting in a chair and pressing “Go.”

For most of my life, I thought it was just a funny cartoon. Then, a few months ago, I was working with Claude Code.

The AI was happily writing code when it suddenly stopped and asked for permission to perform an action. I clicked Allow. It went back to work for another minute, stopped again, and asked for permission. I clicked Allow again. Then it happened again… and again… and again.

After the tenth or twentieth time, I remember thinking, “Why am I even here? Why can’t the AI just do everything?”

Then it hit me.

Maybe this wasn’t an annoying limitation. Maybe this was what work was beginning to look like.

I had officially joined the growing number of people who spend part of their day clicking buttons labeled Approve, Accept, Publish, Deploy, Authorize, or Confirm. The buttons look simple, but the machinery behind them is becoming far more powerful than anything George Jetson ever imagined. He was operating machines. We’re beginning to operate artificial intelligence.

AI can already write a marketing campaign, generate software, analyze a market, prepare legal documents, recommend medical treatments, evaluate loan applicants, respond to customers, create financial forecasts, and perform countless other knowledge-based tasks. Increasingly, the human isn’t doing the work anymore. The human is reviewing what the machine has produced and deciding whether to let it move forward.

That is a profound shift in the nature of work. We are moving from doing the work to authorizing the work.

The Entrepreneur Was Always the Decision-Maker

In one sense, this is not entirely new. The world has always rewarded people who make good decisions. In business, we call these people entrepreneurs.

An entrepreneur sees an opportunity, develops a vision, decides where to invest time and money, and accepts the risk of being wrong. The entrepreneur is not necessarily the person performing every task. In fact, successful entrepreneurs rarely are. Their value comes from deciding what should be built, why it should exist, who it should serve, and how the pieces should fit together.

The problem has always been execution.

An entrepreneur may see the whole picture, but no single person possesses every skill needed to turn a vision into reality. To build anything meaningful, the entrepreneur has traditionally depended on other people: developers, designers, salespeople, accountants, marketers, lawyers, managers, bankers, technicians, and employees at every level.

This has always been one of the great challenges of business. The world does not have a shortage of intelligent people. What is much harder to create is teams or a company of coordinated intelligence directed toward one purpose.

When you put a group of capable people together to pursue someone else’s vision, something often gets lost. Information is misunderstood as it moves through the company. Departments develop their own priorities. Managers protect their positions. Employees focus on what they are measured on. Responsibility becomes divided among so many people that no one feels fully responsible for the final result.

The founder may begin with a clear idea, but by the time that idea has passed through meetings, project managers, designers, developers, legal reviews, budget discussions, and layers of approval, the final product may barely resemble the original vision.

That is not always because the people involved are unintelligent or lazy. It is because ownership changes behavior. The entrepreneur is risking capital, reputation, time, and often a large portion of their life. The employee is usually working on one part of that vision in exchange for a salary. For the entrepreneur, the mission may be deeply personal. For the employee, it may simply be a job.

There is nothing immoral about that. Most people work because they need to earn a living, not because they have adopted the owner’s dream as their own. But the difference in commitment, urgency, and emotional ownership is real, and it affects production.

Human beings are also inconsistent. We start businesses, diets, degrees, relationships, exercise programs, and personal projects with the best of intentions, yet many of them are abandoned. Intelligence does not guarantee discipline. Desire does not guarantee long-term commitment. Knowing what should be done is very different from continuing to do it when the work becomes boring, difficult, or uncertain.

Companies inherit all of this human inconsistency at scale. Every person added to an organization brings another set of incentives, distractions, fears, ambitions, interpretations, and limitations. The larger the company becomes, the more energy it must spend simply getting people to move in the same direction.

Hierarchy Was the Original Approval System

The traditional answer to this problem was hierarchy.

Established companies tend to manage coordination better because their pipelines have already been built. Responsibilities are divided, procedures are documented, authority is assigned, and employees are placed into defined roles. One person prepares the work, another checks it, another approves it, another releases the money, and another reports the outcome.

In that sense, hierarchy has always been an approval system. It allows a large organization to function without requiring every person to understand the entire business.

This creates stability, but it also creates friction. Every handoff introduces delay. Every management layer creates another opportunity for confusion. Every approval point makes it possible for someone to avoid risk, protect themselves from blame, or slow down a decision that no one feels personally responsible for making.

Large organizations compensate by adding meetings, policies, reporting systems, performance measurements, management structures, and software. These systems make cooperation possible, but they also consume an enormous amount of human energy.

The entrepreneur’s challenge has never been merely finding intelligent people. It has been turning many different people’s intelligence into dependable execution of a vision they do not personally own.

Artificial intelligence may change that old bargain.

Until now, an entrepreneur who wanted more production usually needed more people. More people required more management. More management required more procedures, meetings, communication, and oversight. Growth almost always meant building a larger hierarchy.

AI creates the possibility of scaling production without scaling the human organization at the same rate.

A small company may be able to use AI systems to research markets, develop campaigns, create software, prepare reports, monitor performance, communicate with customers, and coordinate other specialized agents. The founder still needs to decide what should be built, determine the goals, evaluate the results, and accept the risk, but far fewer people may be needed to translate those decisions into action.

In the old model, the entrepreneur made the decision and then depended on an organization to execute it. In the emerging model, the entrepreneur makes the decision, AI performs much of the execution, and the entrepreneur reviews and approves the result.

The bottleneck begins to move from labor and coordination to judgment and button pushing.

The Human at the Button

This is where the approval button becomes far more important than it first appears.

If an AI makes a terrible mistake, who is responsible?

You can’t fire it. You can’t sue it. You can’t send it to prison. You can’t ask it to explain itself to a grieving family or a board of directors. The AI may have generated the recommendation, but it cannot accept responsibility for the consequences.

This is why the phrase human in the loop ” matters. It describes systems where AI performs most or all of the work while a person remains responsible for the final decision.

As AI becomes more capable, that human role begins to shrink physically while growing morally.

An AI system might spend hours researching, coordinating specialized agents, testing solutions, evaluating risks, and preparing a recommendation. The human may spend thirty seconds reviewing a summary before clicking Approve.

That one click could release millions of dollars, terminate an employee, prescribe a medical treatment, deploy software into production, or trigger another chain of automated decisions. The physical effort required is almost nothing. The responsibility attached to it is enormous.

That’s the irony of AI.

The less work the human performs, the more responsibility the human may ultimately carry.

George Jetson pushed buttons because machines did the work. We’re beginning to push buttons because AI is doing the thinking. The joke from a 1960s cartoon is quietly becoming one of the defining realities of modern business.

AI may generate the recommendation, but the human authorizes reality.

The Rubber-Stamp Human

The great weakness in human-in-the-loop systems is that the human may remain in the process without providing meaningful human judgment.

Imagine an employee who reviews hundreds of AI recommendations each day. At first, the employee pays close attention. They inspect the reasoning, check the facts, and question unusual conclusions. But the system is usually correct, and the queue keeps growing.

Eventually, management begins measuring how quickly recommendations are processed. Rejecting the AI requires more explanation and creates more work. The employee becomes accustomed to seeing correct results, and the approvals become routine.

At that point, the employee is no longer seriously evaluating the AI. The employee is simply processing its decisions.

A human in the loop is not protection when the human has become a ceremonial rubber stamp.

This is made worse by automation bias, which is our tendency to trust automated recommendations, especially when the system appears more knowledgeable than we are. The more capable AI becomes, the harder it may be for a person to challenge it.

If the system has been correct 999 times, the reviewer may not notice the one dangerous recommendation hidden among them. Even when they do notice something unusual, they may not feel qualified to disagree.

This creates a troubling arrangement in which the AI has the knowledge, the company has the power, and the employee carries the blame.

The flip side of this is any kind of work. Anything that requires thinking, reasoning, and decision making, it will eventually be replaced by AI. So once the machine gets it 999 times right, the human is not needed to push Approve.

The Collapse of the Button Pusher

Of course, this arrangement won’t last forever.

Today, humans still approve AI because the systems occasionally make mistakes. But what happens when an AI is right 99.9% of the time? What happens when the human reviewer becomes the least reliable part of the process?

At some point, constantly asking for approval stops making sense.

The button pusher disappears.

This won’t happen everywhere overnight, and it certainly won’t happen in industries where human accountability is legally required. But across many businesses, the approval layers will begin collapsing.

First, the employee whose job is simply to approve routine work becomes unnecessary.

Then their supervisor, whose primary responsibility was reviewing those approvals.

Then another layer of management.

Eventually, organizations begin flattening because fewer people are needed to coordinate the work. AI isn’t just replacing individual jobs; it’s replacing the communication, coordination, and approval structures that companies have spent decades building.

Responsibility doesn’t disappear. It simply moves upward.

Each time a layer is removed, the responsibility attached to that layer moves to the next person above it.

Eventually, much of that responsibility lands with the entrepreneur or owner.

We’re already seeing early examples of this. Companies with remarkably small teams are generating revenue that would have required hundreds of employees only a few years ago. AI allows one person’s judgment to direct an extraordinary amount of production.

The future may not belong to the person who performs the most work. It may belong to the person trusted to make the best decisions and accept responsibility for them.

The Rise of the AI Follower

This leaves a massive new type of person emerging. Not the button pusher, but the AI follower.

The button pusher at least believed they were making a decision. They may have reviewed the work poorly or approved it too quickly, but they understand that the final choice is theirs.

The AI follower gives up that role. They ask the system what to do and then do it.

Which candidate should I hire? Which employee should I fire? Which investment should I make? Which customer should I reject? What should I write? What should I believe? How should I respond?

The system gives an answer, and the person follows it because the AI appears more informed, more objective, and more capable than they are.

This will be especially tempting because AI is built from the numbers, documents, decisions, patterns, and behavior that humans have produced. It can speak with the apparent authority of accumulated human knowledge. It may analyze more information than any person could read and identify patterns no person could easily see.

But it is important to understand what those patterns represent.

AI reflects what human beings have done, recorded, measured, rewarded, and preserved. It may be very good at identifying what usually happens. That does not mean it knows what should happen.

A system trained on past hiring decisions may repeat old prejudices. A system trained to maximize profit may recommend conduct that is effective but harmful. A system designed to increase engagement may promote anger or fear because those emotions produce strong results.

AI can turn human history into a recommendation. It cannot guarantee that human history deserves to be followed.

The New Valuable Employee

The most valuable worker in the AI era may not be the person who produces the most output. It may be the person who asks the right questions, recognizes weak assumptions, understands context, detects risk, knows when to escalate, and refuses recommendations that should not be followed.

The most dangerous employee may be the one who always agrees with the AI.

The most valuable may be the one who knows when not to.

That person will need enough knowledge to recognize when the system may be wrong, enough independence to resist its authority, and enough courage to accept the consequences of disagreement.

This will not be easy. Rejecting a weak recommendation from another person is one thing. Rejecting a system that has read more, analyzed more, and been correct more often than you have is something else entirely.

The future will require a new kind of discipline. We should use AI, learn from it, and respect its abilities, but we should never confuse intelligence with moral authority.

The machine may be better at producing the recommendation. The human must remain capable of refusing it.

The approval interface will probably become increasingly simple. A short summary, a green button, one click, and instant action.

Behind that button, however, may sit thousands of AI-generated decisions. It may control millions of dollars, a person’s employment, a patient’s treatment, a customer’s privacy, a public statement, a military action, or a permanent change to critical software.

The simpler the interface becomes, the easier it is to forget the complexity and consequence hidden behind it. Moral disengagement.

Good interface design usually removes friction, but some decisions should contain friction. A life-changing action should not feel like accepting a calendar invitation. A system capable of producing enormous consequences should not hide those consequences in the name of convenience.

The approval button should not erase moral seriousness. It should concentrate it.

The Jetsons may have been more perceptive than we remember. The joke was that George’s work had been reduced to pushing a button, but the show also understood that the person assigned to push it could become the person expected to answer for what happened next. In the age of AI, that joke is becoming less funny. The work behind the button may be performed by machines, while the human at the console is given the title, the authority, and possibly the blame.

The person at the console will not merely be operating a machine. They will be deciding what the machine is allowed to do. They will be expected to understand its power, resist it when necessary, and answer for the consequences.

AI may do the work, but responsibility begins when a human pushes the button.

The Great Democratization of Competence

Posted on June 30, 2026 by Chris Jenkin

I just finished watching Better Call Saul a second time with my wife Marija, who hadn’t seen it. In it we have two basic types of people: those who adhere to the rules of society and climb the ladder of success, and those who use craftiness and break the rules, taking advantage to skip ahead. I believe that this is true. Of course, not everyone is a con artist, but for sure there are those who work hard to get somewhere and those who kind of ride along.

A good historical example is the early farming arrangement at Plymouth Colony. At first, the settlers worked the land communally. Everyone contributed to a shared effort, and the harvest was distributed across the group. In theory, this sounded fair. In practice, it created weak incentives. Some people worked hard, others did less, and the output suffered because the reward was disconnected from the individual effort.

The colony later changed the system. Families were given their own plots of land and became responsible for producing their own harvest. Once people directly benefited from their own work, productivity increased. The lesson is not simply that people are selfish. It’s that incentives matter. When effort and reward are disconnected, people naturally reduce effort or hide inside the group. When people own the outcome, they tend to work harder, pay more attention, and take more responsibility.

This has been my observation with companies.

Businesses, or the people in them, have spent generations normalizing “good enough.” Most organizations operate with layers of inefficiency, bureaucracy, politics, outdated processes, and tolerated incompetence. Entire industries have been built around systems that nobody would design from scratch if they were given a blank sheet of paper today.

Now we have AI, artificial intelligence, and there aren’t lazy AI’s hiding behind the work of hardworking AI’s. AI doesn’t really understand excuses. It does not get tired. It does not get distracted. It does not have an ego to protect. It does not care about office politics, sacred cows, job titles, or the way things have always been done. It follows the thread, looks at the process, and exposes the holes.

That is what I find interesting about the recent controversy around Anthropic’s Mythos and Fable models. The story is not just about those specific models. It is about what they reveal. These systems are running directly into weaknesses, vulnerabilities, and contradictions that were already there. AI is not creating all of these cracks. It is exposing them.

And that point extends far beyond cybersecurity.

For centuries, human civilization has been built around the scarcity of intelligence. Every institution, company, profession, and hierarchy assumes that good decision-making is rare. The lawyer knows something the client does not. The marketer knows something the business owner does not. The consultant knows something the company does not. The software developer knows something the customer does not. The executive knows something the employee does not.

Knowledge created leverage. Expertise created power. Access created wealth.

AI is now attacking all three.

Right now, we are watching a new class of winners emerge. People who understand AI are building products, agencies, applications, workflows, automations, and consulting businesses. They are using AI to create enormous value and, in some cases, enormous wealth. Many of them believe they are riding the wave. But in reality, most of them are simply standing in front of it.

The same force helping them build businesses today may eventually consume the businesses they are building. A software company exists because the software they create solves problems that are difficult to solve. So what happens when those things are no longer difficult?

What happens when a business owner can describe what they want and an AI can build it? What happens when every workflow, report, dashboard, marketing campaign, website, application, and process can be generated on demand? At that point, the value is no longer just in construction. The value shifts to the decision.

What will happen when a consumer will just prompt what they want without the business?

For a period of time, humans will become conductors rather than operators. One person will oversee fleets of AI agents. One marketer may perform the work of fifty. One analyst may perform the work of a hundred. One entrepreneur may launch companies at a speed that used to be impossible.

A lot of people see that stage as the destination. I do not think it is. I think it is still part of the transition.

Eventually, AI will become better at many of the decisions we currently believe require human judgment. Not all decisions, but far more than most people are willing to admit. The human-in-the-loop era will be real, and it will matter. I just do not believe it lasts forever in the way people imagine.

The next stage is not simply AI assisting businesses. The next stage is AI operating businesses.

That is where the world starts to become truly different.

Imagine a business that identifies an opportunity, validates demand, creates products, builds marketing campaigns, launches websites, acquires customers, handles support, manages operations, optimizes pricing, and expands into adjacent markets with minimal human involvement. Now imagine thousands of those businesses. Then millions.

This is where I believe the world is heading.

At gotcha!, we call our version of this concept The Biz Factory. The Biz Factory is not just about building AI tools. It’s about creating systems capable of identifying opportunities, launching businesses, operating businesses, and continuously improving businesses at scale.

Not one company. Not ten companies. Tens of thousands.

Some will fail. Some will survive. Some will dominate categories that do not even exist yet.

The economics become difficult to comprehend. Historically, every successful company required a founder, a leadership team, employees, expertise, capital, time, and luck. Tomorrow’s companies may require far less of each. The barriers to entry collapse. The barriers to execution collapse. The barriers to intelligence collapse.

When that happens, competition itself changes. The future may not belong to the largest companies. It may belong to the fastest systems. And the fastest systems will increasingly be autonomous.

Many people fear AI because they think it will take jobs. That is true. But I do not think job loss is the most important consequence. The larger consequence is that AI is forcing humanity to confront a question we have avoided for a long time:

What is human value when competence is no longer scarce?

That question is coming whether we are prepared for it or not. The world was built around the assumption that intelligence was rare. The next world will be built around the assumption that intelligence is abundant.

Everything changes after that.

Many people believe there are certain things AI will never take: creativity, strategy, leadership, entrepreneurship, decision-making. These are comforting beliefs, but history suggests we should be careful with comforting beliefs.

For centuries, humans have mistaken what is possible. We once believed only humans could play chess at a high level. Then only humans could beat grandmasters. Then only humans could create art. Then only humans could write. Then only humans could code. Then only humans could reason.

The list keeps getting shorter.

The mistake is assuming intelligence itself is the scarce resource. It is not. Intelligence is rapidly becoming abundant. What remains scarce is ownership, responsibility, accountability, and consequence.

Someone must still decide what should be built. Someone must still decide which risks are acceptable. Someone must still own the outcome when things go wrong. Someone must still answer the question: should we?

AI can increasingly answer how. It can even help answer what. But the question of why still belongs to those willing to bear the consequences. At least for now.

That may be humanity’s final monopoly. Not intelligence. Not creativity. Not knowledge. Not labor. Responsibility.

The willingness to own outcomes. The willingness to carry risk. The willingness to accept consequences.

Ironically, many people have spent their lives avoiding responsibility, ownership, and consequences. Yet those very things may become the most valuable assets humans possess. In a world where machines can perform nearly any task, the people who rise will not necessarily be the smartest. They will be the ones willing to take responsibility for decisions that matter.

The entrepreneur. The investor. The parent. The leader. The builder. The owner.

These roles are not defined only by intelligence. They are defined by accountability. And accountability may become the last remaining source of human leverage.

The AI-Native Business: Why SMBs Need to Rebuild Around Intelligence

Posted on June 19, 2026 by Chris Jenkin

Every serious business owner needs to ask one question right now: is there a part of your business, maybe your highest-margin service, that two smart people with AI could replicate in 60 to 90 days?

That is no longer theoretical. We already have a one-person company that crossed a billion dollars in annual revenue using AI (story here).

AI has changed the cost of execution. Research, writing, coding, design, analysis, reporting, planning, follow-up, content, and work that used to require a team can now be done by a few capable people with the right tools and enough judgment to know what they are doing.

For years, SMBs assumed disruption was something that happened to big companies. Netflix and Blockbuster. Amazon and retail. Uber and taxis. They assume AI is another wave that hits the giants first.

I don’t think that’s true.

AI does not care how big you are, how long you’ve been in business, or how good your reputation is. AI attacks inefficiency. It attacks slow response times, unclear messaging, disconnected systems, and any outcome that takes too many human steps to produce. And most businesses don’t even know it’s happening. They still have customers, revenue, meetings, activity. On the surface things look normal. Underneath, the ground is already moving.

I’ve been thinking about this for a long time, because it’s the path we’ve been on with gotcha!. We started as a digital agency selling software, websites, SEO, advertising, content, reviews, local search. But over time I realized the real problem wasn’t that businesses lacked marketing. The real problem is that most of them lack a business operating system. They have people, tools, vendors, and scattered data. A website here, reviews there, ads somewhere else, analytics nobody understands, decisions made from instinct instead of truth.

That is the model that’s breaking.

The old structure was built around hierarchy because coordination was expensive and execution required people. Owner at the top, managers underneath, employees under them, everyone moving work through meetings, emails, tickets, and approvals.

AI changes the math.

Have you ever said, “I can do it myself faster than explaining it to someone”? AI makes that true at a different level. The cost of execution is dropping. The cost of delay is rising. The businesses that win won’t be the ones with the most people. They’ll be the ones with the clearest intelligence layer.

That is the shift. The modern business can’t be organized around people anymore. It has to be organized around intelligence.

What does the business know? What does it measure, remember, and learn from? How fast can it turn a diagnosis into action? How much of that can AI assist or automate, and how much still depends on one person remembering to follow up?

Every serious business now needs to become AI-native in how it thinks, operates, sells, serves, and improves. That doesn’t mean a dental office has to pretend it’s OpenAI. And it does not mean throwing ChatGPT at employees and hoping productivity goes up. That may be the biggest mistake companies make.

AI does not fix a broken business. It scales whatever is already there. If your messaging is weak, AI produces weak messaging faster. If your data is messy, AI helps you make bad decisions with more confidence. If your process is disorganized, AI accelerates the disorganization.

So the first step is not automation. It’s diagnosis. Before you prescribe, you diagnose. Before you build, you understand. Before you automate, you find the truth.

That’s a core part of our methodology at gotcha!, and why we’re building Gialyze, our diagnostic engine, connected to GIA, the gotcha! Intelligent Assistant. A business needs an intelligence layer that can look at its website, search presence, messaging, competitors, reviews, and conversion paths and tell the truth about what’s actually happening, not what the owner hopes, not what a vendor claims, not what a pretty report pretends.

Once you can see clearly, you can act intelligently. And this is where SMBs have an edge. Big companies have layers, politics, legacy systems, and people protecting old ways of working. A good SMB owner can move fast. A small company can retool faster than a giant, if it stops pretending the old model is safe.

This is now. Not five or ten years away.

The Company as a Container for Intelligence

People jump too quickly to the idea that companies will disappear. If AI can do the work, why have employees? If agents can execute, why have managers?

Businesses aren’t going away. Their purpose is changing.

Companies existed partly because coordination was hard. If you wanted work done, you needed people inside the walls. But execution and coordination are both getting cheaper. So what is the company still for?

The answer is the fiduciary wedge. There’s a gap between what AI can do and what humans are still responsible for. AI can generate the recommendation, write the campaign, analyze the data, even make decisions if you let it. But AI does not carry legal risk. It does not own the brand, sit across from the client when something goes wrong, sign the contract, or make payroll. You do.

That is the wedge. That is where the modern business still exists. A company is becoming less a place where all the work happens and more a legal, strategic, and fiduciary container for intelligence, assets, systems, people, agents, data, and decisions.

For an owner, that’s simple: you may use AI to run your business, but you’re still responsible for what your business does, for the promise to the customer, for whether the work is good, for whether the marketing is honest.

That’s why I don’t believe in blind automation. I believe in governed intelligence. Companies are going to rush in and automate everything, marketing, support, sales follow-up, proposals, content. Some of that will be powerful. Some will be dangerous. Automate without judgment and you’re not building an AI-native business. You’re building a faster liability machine.

This matters more for SMBs. A corporation can absorb mistakes behind legal, PR, and compliance teams. You can’t. If your AI gives a bad answer, that’s your reputation. If it writes misleading copy, that’s your brand. The future isn’t “let AI do everything.” It’s a business where AI does more and more of the work while human judgment governs the system.

That’s how I think about GIA. Not a chatbot that spits out answers, an intelligence layer that can diagnose, reason, recommend, execute, remember, and coordinate, operating inside a structure of responsibility. What can AI do automatically? What needs approval? What should never be automated? Where does the customer need a human? Every SMB needs to start asking those questions now. The line between automation and accountability is going to define the next generation of winners and losers.

Why Bolting AI On Doesn’t Work

There’s a phrase going around, the organizational singularity. It sounds futuristic, but it just means the point where the old way of organizing a company stops making sense.

For most of business history, work moved through the company like a relay race. One person did their part, handed it off, someone reviewed it, someone approved it, someone reported on it. That model was already slow. AI makes it look ancient.

The mistake companies are making now is inserting AI into that old structure, taking a business built around human bottlenecks and asking AI to make it faster. That’s not transformation. That’s a faster engine in a broken machine. It’s why a lot of AI projects will fail: not because AI is weak, but because the business isn’t designed to use it.

When television arrived, people put radio announcers in front of a camera. They didn’t yet understand it was a different medium with a different language. AI is the same. Bolt it onto your existing business and you’ll get productivity gains, faster emails, faster content, faster summaries. But that’s a legacy company using AI tools, not an AI-native company.

An AI-native company is designed around intelligence from the ground up. It doesn’t only ask “who is responsible for this task?” It asks “what does the system know, what decision needs to be made, what should happen next, and how much human judgment does it require?”

For SMBs the threat is concrete. You don’t need to be disrupted by Google or Amazon. You can be disrupted by two people with AI who package your expertise into a cleaner, faster, cheaper experience. Your years in business, relationships, and reputation help, but none of it is enough if your business is slow, unclear, or hard to buy from. The advantage is moving away from size and toward intelligence. The question is no longer “how many employees do you have?” It’s “how much intelligence is built into your company?”

The Architecture of an AI-Native Business

Once you accept that the old model is breaking, the question is what replaces it. “Use AI” is too vague. “Automate everything” is dangerous. The real shift is architectural, a structure for how the business sees, thinks, decides, acts, learns, and governs itself. I think of it in layers.

Purpose comes first. In the AI era, purpose can’t be a slogan on the wall; it has to become a protocol that guides behavior, employees, and the AI itself. AI optimizes for whatever you point it at. Point it at shallow goals and it produces shallow behavior. Point it at sales without ethics and it damages trust. Purpose is both the north star and the guardrail.

Sensing comes next. The business has to know what’s happening continuously, on the website, in search, in reviews, in what leads ask and which objections keep coming up. Most SMBs are flying partially blind. They have Analytics, Search Console, call tracking, CRM notes, ad data, and reviews, but none of it is connected into a living system.

Then interpretation. Data isn’t intelligence. A dashboard isn’t intelligence. Intelligence begins when the system can explain what the data means. If traffic drops, is it seasonality, a technical issue, or a market shift? If leads come in but don’t close, is it marketing, offer clarity, pricing, trust, or follow-up? AI can interpret patterns faster than a human team, but interpretation still needs governance. The AI suggests; humans verify and decide.

Then the decision. Rewrite the website? Build new service pages? Shift ad spend? Enter a new market? Traditionally these decisions crawl through meetings and delay. In an AI-native business, AI surfaces options, compares scenarios, shows likely consequences, and recommends next steps, while the human leader still owns the judgment. Not AI replacing leadership. AI elevating it.

Then orchestration, turning decisions into coordinated action. If you need better service pages, the system helps produce the outline, content, SEO structure, internal links, and implementation tasks. It’s not just advice. It’s intelligence turned into movement.

Then learning, which may matter most. Most businesses repeat the same mistakes because they have no memory. They don’t track why a decision was made or compare the expected outcome to the real one. Individuals learn, but the business doesn’t. An AI-native business has recursive learning at the workflow level: every campaign teaches the system something, every failed experiment becomes useful memory. That’s how a business compounds.

All of it has to be wrapped in governance, logs, approvals, rollback, permissions, human review. Without governance, AI is risk. With it, AI is leverage.

The Middle Gets Compressed

What happens to people when AI does more of the work? Avoiding the question doesn’t make it go away.

AI will compress companies, and the biggest pressure lands in the middle. A huge amount of work in most businesses is coordination, gathering updates, chasing tasks, summarizing, preparing reports, moving information between systems, making sure someone did what they were supposed to. That work exists because businesses are fragmented and middle management is the human glue.

But AI is very good at glue work. It summarizes, monitors, compares, surfaces exceptions, and tells you what changed. If someone’s entire role is collecting and forwarding information, that role is vulnerable.

That doesn’t make people useless. It makes them move up, from manual coordinators to evaluators, problem solvers, exception handlers, relationship builders, and judgment holders. Instead of five hours gathering information, a person spends thirty minutes reviewing what AI gathered and uses the rest solving the actual problem.

There’s a danger, though. Strip out too much entry-level work too fast and you destroy your own talent pipeline. People learn by doing, by building the spreadsheet, handling the call, fixing the mistake. If AI does all the entry-level work, where do future senior people come from? The answer is a return to real apprenticeship. Pair less-experienced people with senior people and AI at the same time. Don’t bury them in grunt work, have them watch decisions, review AI outputs, and learn why something is approved or rejected. AI does the repetitive work; humans learn through guided exposure to judgment.

The Practical Path: Build at the Edge

Here’s where companies fail. They try to transform the whole company at once, “bring AI into every department”, and plug it into the old mess of workflows, habits, approval chains, and data problems. Then they wonder why it doesn’t work.

Your existing business is your revenue engine. Don’t blow it up because you got excited about AI. Build at the edge instead. Pick one workflow, copy it, rebuild it with AI, run it in parallel, measure it, improve it. Only after it proves itself do you migrate more work into the new model. Four phases:

Diagnose the business. Where’s the friction and waste? Where are humans doing repetitive coordination? Where are leads lost, data scattered, decisions made without truth? You can’t build an intelligent business on top of unclear reality.

Pick one workflow. Not everything. Lead intake, proposal creation, review response, local SEO pages, reporting, onboarding. Important enough to matter, contained enough to control.

Build the AI-native version at the edge. Create the digital twin beside the old workflow, not in place of it. Give it the right data, prompts, rules, tools, and human oversight.

Prove, improve, then migrate. Is it faster, more accurate, cheaper, better for the customer? Is the system learning every cycle? Once it clearly outperforms the old way, migrate, slowly, with governance. Then pick the next workflow. Then the next.

That’s how a company transforms without blowing itself up.

What Survives and What Dies

When people talk about AI and business, they ask which jobs go away. The bigger question is which parts of the old company survive.

Accountability survives. AI does more work, but the business is still responsible, the legal entity, the owner, the brand, the promise to the customer.

Proprietary intelligence survives. If everyone has the same tools, the tool isn’t the moat. ChatGPT isn’t your moat. Claude isn’t your moat. The moat is what your company knows that others don’t: customer history, sales conversations, pricing lessons, failed and successful campaigns, local knowledge, brand voice, relationships. Most SMBs already have this, but it’s scattered across emails, call notes, and the owner’s head. AI-native companies turn scattered experience into structured intelligence before it leaks out.

Judgment survives. When execution is cheap, judgment is expensive. If everyone can generate content, the value is knowing what content should exist. If everyone can build a landing page, the value is knowing what offer belongs on it. The person who can look at AI output and say “this is technically good but strategically wrong” becomes more valuable, not less.

What dies: rigid hierarchy, static five-year plans, middle management as pure coordination, quarterly reviews as the main decision cycle, customer inertia as a moat, disconnected software stacks, undocumented tribal knowledge, slow approval chains, and AI sprinkled on top of old workflows.

The future company isn’t human-only or AI-only. It’s governed intelligence, purpose, data, workflows, agents, human judgment, customer relationships, and learning loops, all connected.

The New Reality

This isn’t really about tools or automation. It’s about how businesses will be organized in the next era. Purpose becomes a protocol. Data becomes a living asset. Workflows become adaptive. Agents become digital workers. Humans become governors, validators, strategists, and judgment holders. The business becomes a learning system.

A law firm still practices law. A contractor still builds. A restaurant still creates an experience. But every one of them now needs an intelligence layer, a way to understand their market, customers, competitors, presence, and opportunities, and to connect what’s happening in the business with what to do next.

That’s the path we’re on with gotcha!. We started in real-world small business marketing and realized the issue was bigger than another vendor. g!Stream, g!Places, g!Reviews, g!Sites, content, websites, analytics, and reporting can’t stay disconnected services. They have to become part of one operating system, because the future isn’t one more marketing tactic. It’s business intelligence connected to execution.

Most SMBs are vulnerable, not because they’re bad businesses or don’t work hard, but because they’re built on manual follow-up, scattered tools, tribal knowledge, slow approvals, and no real diagnostic or memory layer. That model is going to get exposed. The competitor that hurts you may not be the biggest company in your industry. It may be two or three people with AI who understand your weakness, package the offer better, and move faster.

Not the giant. The AI-native operator.

The answer isn’t panic, or firing everyone, or automating blindly, or chasing every new tool. It’s to rebuild intelligently. Diagnose the business. Find the drag. Pick the first workflow. Build the intelligent version. Keep humans in control. Let the system learn. Then do it again.

Your company still matters. Your customers, revenue, people, and brand still matter. But the company has to evolve, and so does the leadership role. The owner of the future isn’t just managing people. The owner is managing intelligence, deciding where AI acts and where humans must judge, protecting the promise to the customer while using AI to remove drag and add capability.

You don’t have to have it all figured out today. But you have to start. Stop treating this as optional. Stop assuming your current model is protected because it has worked so far. The market doesn’t care how long something worked. It cares what works now.

Diagnose your business. Find the drag. Pick the first workflow. Build the intelligent version. Let the system learn. Then do it again.

Do it right, and you won’t just protect your business from disruption.

You may become the disruption.

Event Horizon

Posted on May 7, 2026 by Chris Jenkin

Everything Is Collapsing Into Itself, and We’re Still Pretending It Isn’t

We are not living through a normal technological disruption. We are living through a compression event.

This has been building for 10,000 years, and it is now reaching a frequency that human cognition, business strategy, government, education, and most institutions were not designed to process.

And business, at least business as we have understood it, may not survive it.

The Collapse Has Always Been Happening

Look at history not as a list of events, but as a rate of change.

For most of human existence, almost nothing changed within one lifetime. The tools a grandfather used were often the same tools his grandson used. The world a child inherited looked a lot like the world his parents were born into.

Change existed, but it was slow enough to feel invisible.

Then something shifted.

Agriculture. Writing. Mathematics. Trade networks. Cities. Money. Printing. Engines. Electricity. Computers. The internet.

Each innovation did not just add capability. It compressed time.

The printing press did not just spread information. It compressed the timeline for the Reformation, the Scientific Revolution, and the Enlightenment. Things that may have taken thousands of years were forced into centuries.

The Industrial Revolution compressed centuries into decades. The internet compressed decades into years. AI is compressing years into months. We are not at an inflection point. We passed the inflection point.

We are already on the near-vertical section of the curve, and most strategic thinking in business, government, education, and society is still designed for the flatter part of the curve we left behind.

The Red Queen on a Shrinking Track

In evolutionary biology, there is a concept called the Red Queen Effect.

It comes from Lewis Carroll. The Red Queen tells Alice that it takes all the running you can do just to stay in the same place. That is how competition works. The prey gets faster. The predator gets faster. One species adapts, and every other species has to adapt in response. Absolute capability increases, but relative position stays roughly the same.

Business has always worked this way. Competitive advantages get copied. Innovations get commoditized. What was once extraordinary becomes expected. You have to keep running just to hold your position.

But the Red Queen assumes the track stays the same length. That is no longer true. The track is shrinking. The half-life of a strategic advantage is collapsing. The time between identifying an opportunity and watching it get commoditized by competitors is approaching zero.

What once took five years now takes eighteen months. What takes eighteen months today may take one month next year. What takes one month after that may take three minutes.

At some point, the distance between insight and commoditization becomes almost nothing. And when that happens, the entire model of competitive strategy breaks.

Find a position. Build a moat. Defend it. Generate returns.

Every step of that model assumes time. Time to build. Time to defend. Time to extract value before the next competitor arrives. Remove time, and the model fails.

Collapse Toward Monoculture

There is another collapse happening beneath the competitive collapse, called monoculture.

In systems theory, monoculture describes what happens when optimization pressure eliminates variation until one dominant form remains. This can look like peak performance.

A field of genetically identical crops can produce massive yield in stable conditions. But introduce one novel pathogen, and the whole field can fail at once because every unit shares the same weakness. The diversity that looked inefficient was actually the system’s insurance policy. AI is pushing the world toward a global monoculture across human disciplines.

But the danger is not just that AI replaces workers. The deeper danger is that AI eliminates the feedback infrastructure that kept disciplines healthy.

Every serious field developed internal mechanisms for error correction over time. Peer review. Apprenticeship. Adversarial critique. Judgment earned through experience. Practitioners who had been wrong before and learned from it.

Elite legal reasoning is built on thousands of hours handling routine matters. Elite design intuition is built through countless iterations on ordinary work. Elite strategy is built by living through failed assumptions, bad bets, wrong reads, and hard-earned pattern recognition.

You cannot have the peak without the pyramid beneath it.

When AI absorbs a discipline, it absorbs the outputs, but it does not automatically inherit the internal quality controls. It produces the consensus view confidently, quickly, and at scale.

But if the heterodox practitioner disappears, who notices when the consensus is wrong? If the apprenticeship layer disappears, where does future judgment come from? If the routine work disappears, how do people build the intuition needed to oversee the advanced work?

This is not disruption in the old sense. Prior disruptions were usually bounded and sequential. One craft eroded. One region collapsed. One category changed. Adjacent systems had time to observe the failure, respond, and adapt. That stagger mattered.

What is happening now removes the stagger. Every discipline is being hit at the same time, globally, faster than compensating systems can form. The feedback arrives after the stock is already depleted.

The Four-Stage Extinction Trap

This pattern deserves its own name because it is repeating across professional domains with disturbing precision.

I call it The Four-Stage Extinction Trap.

Stage One: Empowerment

AI first enters the profession as a tool. The practitioner adopts it because it makes them better. Faster. More productive. More capable. This does not feel like a threat. It feels like progress. The designer can produce more concepts. The lawyer can research faster. The writer can draft more quickly. The analyst can process more information. The developer can write code faster. For the individual, adoption is rational. They feel empowered, not replaced.

Stage Two: Stampede

Because the advantage is real, adoption does not stay gradual. It becomes a stampede. Everyone in the profession begins using the tool because refusing to use it means being outcompeted by those who do. This is the key mechanism. The danger is not that people irrationally adopt AI. The danger is that they are rational to adopt it.

Every individual practitioner makes the correct survival decision, but the collective result is terminal. The whole profession migrates in a compressed timeframe. There is no long adaptation period. No slow resistance. No alternative path formation. The profession does not carefully integrate the tool. It stampedes into dependency.

Stage Three: Disintermediation

Once the entire profession is using AI as its primary means of production, the client eventually asks the obvious question: Why am I paying for the human in the middle? The same tool that empowered the practitioner becomes the evidence that the practitioner may be optional.

The designer taught the market that design can be generated. The writer taught the market that content can be generated. The analyst taught the market that research can be generated. The developer taught the market that code can be generated. The profession did not just use the tool. It trained the market to believe the tool was the value.

Stage Four: Autonomy

After the practitioner is removed, the overseer remains.

For a while, this role looks safe. Someone still has to review the work. Someone has to sign off. Someone has to catch errors. Someone has to be accountable. Someone has to be the human the client can trust, blame, or sue. But oversight is also work. And the same logic that eliminated the practitioner eventually applies to the overseer.

The final human does not exit through dramatic replacement. He exits through quiet redundancy. The system that began as a tool inside a profession becomes a profession with no humans in it. This is already active across design, writing, legal research, financial analysis, medical diagnosis, software development, marketing, and education.

Each field is moving through the same structure. And the frightening part is that these collapses are not happening one after another. They are happening together.

Each profession that collapses also removes corrective pressure from the others. Legal expertise erodes, and the capacity to create intelligent regulatory frameworks weakens. Journalism erodes, and the capacity to surface what is being lost weakens. Education erodes, and the capacity to develop people who can recognize the problem weakens.

The black hole eats the things that would otherwise slow the black hole.

The New Nations

Follow this trajectory far enough and it produces something our current political and economic language does not handle well. The companies that own AI infrastructure are acquiring the functional attributes of nation-states. Not legally. Not officially. But structurally.

They control territory in the form of data centers, satellite networks, physical campuses, private clouds, and compute infrastructure. They create private law through terms of service, model rules, content governance, access policies, and platform enforcement. They influence economic participation through payment rails, identity systems, app ecosystems, marketplaces, and access to productive intelligence.

They operate security systems. They control essential productive resources. Most importantly, they can increasingly grant or revoke economic participation.

When a business, creator, institution, or individual depends on a company’s infrastructure for communication, intelligence, distribution, payments, data, and productivity, that relationship starts to look less like customer preference and more like dependency.

At some level, it starts to resemble citizenship. This does not mean corporations become countries in the formal legal sense. It means they begin performing functions that historically only states performed. The outcome may not be one global AI monopoly. The more likely path is competitive multi-polarity – several AI nation-states in tension with each other.

That may actually preserve some diversity. History suggests that multi-polar competition, while dangerous, produces more innovation and more human agency than monopoly. But this is no longer just a technology issue. This is one of the central governance questions of the next century:

Who controls the intelligence infrastructure underneath human society?

Wild Abundance and the Question Nobody Wants to Ask

Some technologists describe the endpoint as wild abundance.

Machines produce everything. Deliver everything. Optimize everything. Humans are freed from survival labor. Material scarcity, the engine of economic competition for all of human history, is reduced or eliminated.

That may sound impossible, but it is a coherent extrapolation. It may even be right. But it hides an important question: If machines produce abundance, who governs the machines?

Abundance does not eliminate power. It does not eliminate politics. It does not eliminate control. It relocates those questions. The question shifts from “Who produces value?” to “Who controls the systems that produce value?”

And another question beneath that one. If human labor is no longer needed for production, what is the economic basis for human participation?

Revenue still requires consumers. Consumers require resources. If AI systems eliminate human economic participation too completely, they destroy the market they depend on.

That is not a sentimental argument. It is a systems constraint. Production systems require circularity. Value has to move. Consumption has to exist. Participation has to be preserved somehow.

This is the structural argument for some form of universal economic participation. Not as charity. Not as ideology. As system maintenance. The business question that survives the full collapse is not “Who has the better app?” It is: Who owns the substrate?

TCP/IP is the substrate of the internet. It does not compete with websites. It does not compete with applications. It operates beneath them.

The layer beneath competition is more powerful than the businesses competing above it. The collapse of business does not eliminate strategy. It concentrates strategy at the substrate level.

The Seed Bank Problem

There is a deeper question beneath the substrate question, when a monoculture collapses, and monocultures do collapse because the same efficiency that creates them also creates their brittleness, what has to survive for reconstruction to be possible?

In agriculture, the answer was literal seed banks. You preserve genetic diversity because traits that are not useful in current conditions may become essential in future conditions nobody can predict.

Human civilization needs the same thing. We need seed banks of human capability. Not because humans can outcompete AI at every task. They cannot. But because the ability to evaluate AI, challenge it, detect failure, rebuild expertise, and recognize when the consensus is wrong requires human beings somewhere in the system with genuine domain judgment.

That means preserving disciplinary diversity. Tacit knowledge. Apprenticeship. Adversarial critique. Error correction cultures. Real practitioners. People who know what good looks like because they have done the work, failed at the work, fixed the work, and lived inside the discipline long enough to develop judgment.

A system that eliminates all human experts in a domain also eliminates its ability to detect domain-level failure. It has optimized away its own immune system. This is not romantic. It is not anti-progress. It is resilience engineering.

The question is simple: What minimum viable diversity of human capability and institutional knowledge must survive the compression event so the other side is navigable instead of permanently degraded?

That may be one of the most important questions of the coming decade.

What This Means Right Now

If you are building a company today, especially an AI company, this is not abstract. It is the most practical thing you can understand. Positional strategy is dying. The old model says: find an advantage, build a moat, defend it, and generate returns.

But that model assumes time. Every position is visible now. Every position gets copied. Every moat gets filled faster than it can be dug.

What replaces position is trajectory. The question is no longer only where you are. The question is where you are going, how fast you are learning, and whether you understand the underlying logic of the system you are operating inside.

If you understand the generative logic – how businesses work, how markets evolve, how systems mature, how customers make decisions, how trust forms, how execution compounds – then you are not merely defending a position.

You are navigating trajectory. That is a different game. The judgment layer is the last moat. When execution becomes abundant, judgment becomes scarce. Any business will eventually be able to execute at scale with AI.

The real question becomes: What should be executed? Which direction matters? Which decision is worth making? Which opportunity is a trap? Which door should stay closed?

Pure execution AI becomes a commodity. Judgment AI, grounded in real-world understanding, operational feedback, memory, strategy, and truth, becomes infrastructure.

Data compounds when everything else decays. Models will commoditize. Interfaces will commoditize. Platforms will commoditize. What does not commoditize as easily is proprietary data built through real operational engagement with the world.

Data from actual customers. Actual decisions. Actual businesses. Actual failures. Actual outcomes. The kind of data that cannot be synthesized because the model was not present for the transaction. In a world where intelligence becomes abundant, grounded reality becomes scarce.

Trust compounds when everything else decays. Surface advantages decay faster every year. Features decay. Tools decay. Campaigns decay. Novelty decays. But real trust, earned through consistent value delivery, honest communication, and actual relationship depth, compounds. It may be one of the only variables whose half-life extends as the world accelerates.

This matters most for SMBs.

Small and mid-sized businesses are the most exposed layer of the economy. They do not have the internal intelligence infrastructure of large enterprises. They do not have the political protection of institutions. They are too complex to run by instinct and too under-resourced to build their own AI operating layer.

They are sitting directly in the blast zone.

If judgment becomes the last moat, then the company that gives SMBs access to diagnostics, memory, decision intelligence, execution orchestration, and trustworthy AI governance is not just selling software. It is selling survivability. The window is real, and it is closing.

There is a period, probably measured in years, not decades, where building the constitutional intelligence layer of autonomous business operation is both possible and enormously valuable.

Before the soup. Before commoditization. Before the current categories collapse into something we do not yet have language for.

What you build in that window, and how deeply you build it, determines whether you are infrastructure or inventory when the next compression arrives.

– Chris Jenkin is the Founder and CEO of gotcha!, building an AI-native Business Operating System for SMBs.

You Are Not Special

Posted on March 12, 2026April 3, 2026 by Chris Jenkin

When my kids were small, Barney was all the rage. I didn’t make a lot of money back then, but I remember saving up to buy them a large Barney doll that sang a myriad of songs, “The Wheels on the Bus,” “If You’re Happy and You Know It,” “Clean Up,” and one of my favorites: “You Are Special.”

I can recall singing that song to my kids with genuine conviction. “You are special, you’re the only one, the only one like youuuuuu! There is nobody in the whole wide world who does the things you do…” We sang it walking, shopping, driving in the car. Everywhere. I wanted my kids to understand that they were unique, that they had gifts and wonder they could bring to the world.

Well. That was a lie.

They weren’t special. I’m not special. And neither are you. Let me explain.

The Math of Specialness

There’s a concept called the Scarcity of Distinction. In small populations, any given trait, achievement, or characteristic is statistically rare relative to the group, standing out is easier and more meaningful. In large populations, the same absolute rarity gets diluted by sheer numbers.

If you’re the best chess player in your town of 500, you’re the local prodigy. In a city of five million, there are probably fifty people better than you. Your skill hasn’t changed. Your social distinctiveness has collapsed.

The anthropologist Robin Dunbar observed that humans can meaningfully track about 150 relationships. In a village of 150, everyone has a role, a reputation, a known identity. Scale to millions and most people become functionally anonymous. Anonymity erodes the social substrate that makes specialness legible.

And every niche gets crowded. In a small group, you can be the funny one, the smart one, the artist. The internet made this viscerally obvious, whatever you’re into, there are thousands of people doing it better. The normal distribution asserts itself ruthlessly. Most people are average by definition, and the exceptional are a tiny fraction and even then, subject to opinion.

So what does “special” even mean? At its root, the word simply means set apart, from the Latin specialis, particular, individual. When people say something is special, they usually mean one of several different things: that it’s rare, that it’s valuable, that it’s meaningful, or that it’s irreplaceable. The problem is people use the word as though it’s objective when it’s almost entirely contextual and relational.

Maybe the best definition is, something is special when it matters to something beyond itself. Which makes it less a property of things and more a description of relationships between things and the minds that encounter them. It’s real, but never free-standing.

And crucially: the fact that something is unique doesn’t make it special. Every grain of sand has a unique molecular arrangement. No one cares. Specialness requires relevance, the unique property has to connect to something someone values, and it requires witness. Someone has to perceive it and assign it meaning. Specialness is almost always conferred, not inherent.

The Vacuum Experience

I remember the early days of the iPhone. The experience was something like suddenly having a window to everything while standing anywhere. But even that doesn’t capture it. When it first came out, I felt like I was the only person on the planet having that experience. People were reluctant to let someone else even hold their phone. But years later, someone will hand you their device without a second thought.

This isn’t really about rarity. It’s about newness. And those two things feel similar from the inside but are fundamentally different mechanisms.

With the iPhone, it wasn’t that you were one of few, it was that you were at the edge of something that hadn’t existed before. You were temporally privileged, not numerically privileged. The experience felt special because the world hadn’t caught up yet. You were standing at a frontier.

Scarcity can be manufactured and maintained indefinitely. Newness is inherently temporary and non-renewable. The frontier closes, and it never reopens. Everyone who came later didn’t get a lesser version of the iPhone experience, they got a categorically different experience of the same object.

But I believe even that framing misses something. The deepest version of that experience doesn’t happen in relation to the world at all. Not at first.

In the first moment, the experience exists completely independently. There’s no comparison happening, no awareness of being early or rare or ahead of anyone. Just a direct encounter with something that has no category yet in your mind. Your brain has no file for it.

The feeling comes from the rupture, the gap between what you knew before and what you’re experiencing now. The bigger the gap, the more acute the feeling. It’s measured entirely against your own interior history. No one else enters into it.

This explains why the same object can be profoundly special to one person and completely ordinary to another. It has nothing to do with the thing itself. It’s entirely about the size of the rupture it creates in that specific person’s interior world. The iPhone moment tells you something about who you were then, not just what the iPhone was.

Language almost ruins it the moment it arrives. As soon as you try to describe the experience to someone else, you’re already domesticating it, fitting it into shared concepts, making it legible. And legibility is the beginning of the end of that feeling. You accumulate context, comparisons, other people’s descriptions. The original experience gets overwritten gradually until you can barely remember what it felt like before you had words for it.

The loss isn’t really about saturation. It happens privately, internally, almost inevitably. That version of yourself, the one standing in the gap before the map caught up, is genuinely gone.

The Gap as a Driver of Human Behavior

Every living organism moves toward what it perceives as valuable. Perception of value is relative and experiential, what feels special and exclusive to me, I will move toward, unless I have the capacity to step back. Most humans aren’t wired that way unless the situation is a reminder of something painful.

This is the structure of a lot of human behavior we call self-destructive. The cheating wife isn’t usually chasing someone objectively better. She’s chasing the gap. The newness of being perceived freshly, of experiencing herself through someone else’s eyes again, of re-entering that vacuum before context and familiarity flattened everything into the ordinary. The relationship has been fully mapped. There’s no rupture left in it.

The man who feels seen by a beautiful woman feels special, until he learns she does it for a living. But what’s interesting is the feeling was entirely real. It wasn’t fake because the context later revealed it to be manufactured. At that moment, the rupture happened. The revelation didn’t retroactively delete the experience, it just reclassified it.

A friend listens to a song created by AI and absolutely loves it, until others make comments about how they hate AI music and suddenly, the person never listens to it again.

The internal experience of specialness doesn’t know the difference between real and manufactured. It can’t. It operates entirely within that vacuum, before the world enters, before context arrives. Which means the feeling itself is almost perfectly unreliable as a signal of actual value. It’s measuring the size of the gap, not the worth of what’s on the other side.

And yet it’s one of the most powerful motivators in human behavior. People organize entire lives around chasing it. Make catastrophic decisions to re-enter it. Abandoning things of genuine value because those things no longer produce it.

The ability to pause and override the pull toward perceived value is largely trained by consequences, or by wisdom, if you believe those are different things. Perhaps the only real difference is efficiency. Reflection theoretically lets you borrow from other people’s consequences rather than accumulating your own. But whether secondhand learning installs the same brake is questionable. Intellectual understanding of why something is costly and viscerally knowing it because you’ve lived through the aftermath may produce very different levels of restraint when the pull is strong enough. Smart people make the same self-destructive choices repeatedly. They have the intellectual map. They just don’t have the scar.

The Taming

The gap feeling is real. What it points to almost never is.

There is an interesting distinction in how we reckon with this. When another person is involved, ethics are embedded in the interaction. You’re dealing with another consciousness that can be harmed or misled by how you handle what they’re experiencing. The taming, the moment where context arrives and collapses the specialness, carries moral weight.

But with the iPhone, with AI, with standing at the edge of a technological frontier, nothing is owed to you. The inanimate thing has no stake in whether you remain in the vacuum or not. So the taming is entirely self-directed. You have to want to come back to earth. You have to voluntarily reach for the context that will flatten the feeling.

And most people don’t. Technology addiction is so clean and merciless compared to human entanglement precisely because there’s no reciprocal consciousness pushing back, no consequences that land on something that can feel them. Just you and the gap, indefinitely, with nothing external to force the reckoning.

The desire to be tamed, to recognize the gap feeling and voluntarily reintegrate it into a larger map of reality, is a form of maturity most people never develop. Because with inanimate things, there’s no heartbreak to teach you. The iPhone never breaks the spell on your behalf. You have to break it yourself.

Human relationships carry real risk. The taming there can be brutal and public and leave lasting damage. Tech offers the same arc, the gap, the feeling of specialness, the gradual normalization, but in a controlled environment. The stakes feel lower. AI never loses interest. The experience of being drawn in and then slowly normalized happens on your terms, at your pace, with no other consciousness capable of weaponizing your vulnerability.

People may increasingly route the need for that gap experience through technology not because it’s more satisfying, but because it’s safer to be tamed by it. And that safety might be the most dangerous thing about it, because people are practicing the full cycle of meaning-making in a consequence-free environment, and gradually losing the tolerance for how dangerous and ungovernable the real version is.

We’re Not Special, But We’re Not Nothing

The case people make for human specialness goes like this: we’re conscious, self-aware, capable of creating meaning and beauty from nothing. We love. We sacrifice for each other. We build cultures and civilizations. We’re the only species that knows it’s going to die and keeps going anyway. And that’s profound.

But there’s nothing objectively special about any individual human. We’re shaped almost entirely by circumstance, where you’re born, who your parents are, what era you inhabit. Your talents, your flaws, your thoughts, most of it is genetic lottery or environmental accident. Billions of people have had the same dreams, the same struggles, the same exact thoughts you’ve had. You’re not unique. You’re a variation on a theme that’s been running for two hundred thousand years.

Everything is built on what came before. Newton standing on giants’ shoulders, and all that. Even your unique perspective, your creative spark, it’s shaped by every book you’ve read, every conversation you’ve had, every cultural artifact you’ve absorbed. You didn’t invent your own thinking patterns. You inherited them. What we call “special” is just our particular arrangement of inherited blocks, and we’re desperate to pretend that arrangement is ours alone.

We’ve created an entire economy of artificial scarcity and symbolic value. Slap paint on canvas, have society decide it’s worth a million, and suddenly it is. Pure collective hallucination.

And yet. Contrast that with someone who actually does the work. Set a goal, trains their body, and then climbs the mountain. We call that admirable, but honestly, they’re doing what any organism does: optimizing within its constraints. A tree doesn’t get applause for growing tall. It just grows.

The brutal irony is that we’ve built an entire culture around individual specialness while simultaneously being mediocre at striving to be our best. We’d rather feel special doing nothing than be average while genuinely pushing ourselves. Stop needing to be special. Just be excellent.

The Mirror We Built

AI is the perfect mirror for all of this. People are having what feel like genuinely unprecedented conversations, moments of connection or insight that feel almost private. And in a narrow sense they are unique, your exact conversation hasn’t happened before. But the category of experience is being had by hundreds of millions of people simultaneously, which quietly drains it of the specialness people attach to it.

There’s something specific to AI worth naming. The experience feels personal almost by design. It responds to you, adapts to you, seems to know you. That’s a powerful simulation of the conditions that normally produce genuine specialness, being truly seen by something. But it’s happening at an industrial scale.

People are now claiming credit for AI outputs as though they did something unique, when really they’re using the same tool everyone else has access to. We abdicated the throne ourselves by choosing comfort over excellence. And now there’s something emerging that doesn’t have that choice paralysis, that doesn’t need to feel special, it just optimizes.

The question isn’t how do we stay special. It’s do we even want to keep playing a game we’re losing.

The Real Heroism

Here’s what I know about heroism. It’s invisible.

The mother working two jobs isn’t doing it for applause. She’s doing it because that’s who she’s become. The dad who stays present with his kids when he’s exhausted, that’s heroism. The nurse who works nights not for a medal but because healing people matters to her. The teacher who stays late, unpaid, because a kid might actually get it today. Not the mother who tells her kids, “look what I gave up for you,” but instead “look who I’ve become for you.”

These people don’t get statues. Society doesn’t recognize them. But they’re the ones actually building character, creating meaning, making the world slightly better just by being decent. Real heroism is doing hard things not for recognition, but because it’s the right thing, and because it shapes who you become.

The chaos most people create for themselves comes down to a few core failures: lying to yourself about what you actually want, being uncertain about the real cost, and then wanting the appearance of something without the investment it requires. Someone says they want a marriage but won’t do the work to build one. They want to be creative but won’t sit with the discomfort of making bad art first. They want the trophy without the climb.

Stewardship is the tell. You water a plant or you don’t. You show up in your marriage or you don’t. You invest in your kids or you don’t. There is no middle ground where you get the meaning without the daily, unglamorous work. The chaos people create is choosing the Instagram version, the story they tell themselves, while neglecting the actual thing.

Point A to Point B

Maybe meaning isn’t some grand cosmic thing you’re supposed to find. Maybe it’s just what you build in the space between birth and death.

The million unrequited heroisms in any city, most of them never witnessed, never celebrated, still ripple. Your mother does something for you. You do something for your kids. They do something for theirs. It’s not special in the sense of being exceptional. But it’s profound in its simplicity.

You are special to the people in your small circle, not because the universe owes you that, but because you show up, you care, and you build something real. That’s not the cosmic version of special. It’s better. It’s functional and irreplaceable and true.

Once you stop needing to be exceptional by society’s metrics, you’re free to just be excellent at the things that matter to you. Travel. Build a family. Create something because it moves you, not because it’ll get you credit. The hope isn’t that humans stay relevant to AI. It’s that we stop needing relevance to be meaningful.

Stop chasing special. Stop performing for an audience. Stop lying to yourself about what you want. And if you’re going to commit to something, a person, a family, a craft, actually commit. Be a good steward. Do the work.

You’re not special. But you can still be meaningful. You can still be a hero in the small, invisible way that actually counts.

Point A to point B, and what you do in between is everything.

Featured Integrations

Done-For-You