‍

The corporation was the first economic actor we built that had no body, no soul, and no need to sleep. We are now building the second.

‍

Senior leadership delegated a defined set of decision rights to an autonomous enterprise system, the kind of work that had previously belonged to a team of experienced humans. I have watched this happen in some of the world’s leading companies, in rooms where the decision was made carefully, by people who knew what they were authorizing. Most of the system’s recommendations were accepted automatically. A subset were reviewed by humans, selectively, by exception, or by sample.

‍

Nothing dramatic happened. The plan landed. Revenue grew. Profits were realized. Competitors were outmaneuvered. Margins held or widened. From the outside, the company looked like it had bought another piece of enterprise software and made it work.

‍

From the inside, something else happened. A class of decisions that had always belonged to people now belonged, in the operating sense, to a system. Inside its defined scope, it ran on its own clock, pursued the corporate objectives it had been given, and produced consequences the firm had to live with.

‍

By the time the plan had landed, the system was pacing the cadence of the firm, not following it. The vocabulary we used for these systems no longer described what they were doing.

‍

What the second actor unlocks

The upside is the reason serious leaders are doing this at all, and it deserves to be named first.

‍

A firm whose operating decisions live partly inside autonomous systems handles uncertainty on a different timescale. Conditions change at ten in the morning. The plan adjusts by noon. A supplier disruption, a demand signal, a price move in a downstream input. The firm absorbs each one at the speed of the market rather than the speed of the next planning cycle. The annual operating plan stops being an artifact and becomes a continuous posture. Coherence with the corporate objective is held across thousands of small decisions a day, in a way no human team can sustain.

‍

That is the prize. A firm that re-plans as the world changes, decides at the speed of its markets, and holds strategy and execution in alignment across thousands of small choices a day. Not a marginal efficiency gain. A different operating model.

‍

The market has already begun to price the distinction, and the number is large.

‍

The market has noticed

Between January and March of this year, the enterprise software sector lost roughly two trillion dollars in market capitalization, the largest non-recessionary 12-month drawdown in more than three decades. The software sector’s weight in the S&P 500 fell from twelve percent to under nine in three months. The companies hit hardest were not poorly run. They were the canonical names of the previous era. ServiceNow. Salesforce. Workday. Adobe. Atlassian. Their problem was not execution. Their problem was that the per-seat license, the dominant pricing model of enterprise software for two decades, had quietly assumed a human at every seat.

‍

I have argued elsewhere that the divergence between traditional software and AI-native systems, and between GenAI-only architectures and multi-engine platforms, is reshaping the enterprise software industry. The point I want to make here is narrower. When the agent does the work, the seat is empty. When the seat is empty, the license is a line item that no longer maps to anything real. Investors saw this faster than most boards did. The repricing was not a mood swing. It was Darwinian. The business model of traditional software, and the operating model of the firms that bought it, had both been built for a world in which a human remained the unit of work.

‍

The two-trillion-dollar repricing is the public, financial signal that the tool frame is breaking. The market knows. The vocabulary has not caught up.

‍

What the metaphor is doing

Language carries policy. The word tool, applied to a system, drags an entire infrastructure of assumptions behind it. A tool is held by a human. A tool extends human will. A tool is the medium of action, never its source. When something goes wrong with a tool, we ask what the human did with it, never what the tool decided.

‍

That assumption is the spine of how we regulate, contract, insure, audit, and manage these systems. Procurement frameworks treat them as software licenses. Liability regimes treat them as products. Audit committees treat them as controls. Insurance carriers treat them as fixed assets. Boards treat them as line items in a technology plan. Each of these treatments is coherent only if a human remains the agent of action.

‍

When humans step back, the entire stack of procurement, liability, audit, insurance, and board treatment quietly stops describing what is in the room.

‍

Where the frame cracks

Three pressures, all already present in production environments, push past the tool frame.

‍

The first is goal persistence and coherence. A wrench does not want anything between uses. A modern decision system holds objectives across days, weeks, planning cycles. It persistently returns each morning to a state and a target it was given and continues. Coherently maximizing the objective function across timescales, without continuous human direction, is a property of agents, not of tools.

‍

The second is resource allocation. These systems do not merely advise. They move budget, inventory, capacity, attention, and increasingly money. Even in advisory mode, when a recommendation is implemented at scale by default and reviewed only by exception, the system is allocating resources. Reviewing in aggregate is not the same kind of decision as approving each one. The vocabulary of approval has become the vocabulary of audit.

‍

The third is machine time. Markets, regulators, courts, and boards run on human time. Quarters. Annual audits. Multi-year cases. Decadal regulatory cycles. Agent-to-agent commerce, already running in procurement, treasury, logistics, and pricing, moves in milliseconds. The mismatch is not an inconvenience. It is a structural defect in the institutions we rely on to keep commerce honest.

‍

Each of these, on its own, can be managed within the tool frame by adding a footnote. Together, they require a different category.

‍

The first non-human actor

We have done this before. Once.

The argument that follows sits inside a recognizable tradition. Ronald Coase asked in 1937 why firms exist at all rather than every transaction running through the market and answered that the firm exists where coordination inside an entity is cheaper than coordination through prices. Oliver Williamson extended that insight into a theory of transaction costs and the boundaries of the firm. Margaret Blair, working closer to the present, showed that the corporation is best understood as a legal device for holding the joint output of a team whose members cannot easily contract with each other one by one. Each of these scholars was studying the same question from a different angle. What kind of entity is a firm, and why did we have to build it?

‍

The corporation, as a legal person capable of holding property, signing contracts, and bearing liability in its own name, is a 19th-century invention. The American shape of it was assembled in fragments. Dartmouth College in 1819 gave the corporation a constitutional protection of its charter. Santa Clara County in 1886 carried into the headnote the proposition that the corporation was a person under the Fourteenth Amendment. A century of statutes, cases, and accounting conventions filled in the rest. By 1900, a manager could enter a building, sign a paper, and obligate an entity that could outlive its founders, contract in its own name, and bear consequences across generations.

‍

What that required was not one invention. It was a stack of them, each addressing a problem the new actor created. Limited liability gave investors a defined locus of obligation, so that the firm itself, rather than every shareholder, was the entity that could be sued. Perpetual succession allowed the firm to outlive the people who started it, which meant the law had to settle what happened when officers and owners changed without changing the entity beneath them. Fiduciary duty fixed who owed what to whom inside the firm, so that managers acting in the firm’s name were held to a standard distinct from their personal interests. The audit profession was built, almost from nothing, to give outsiders a continuous account of what was happening inside an entity they could not enter. The corporate seal, and later the authorized signature, gave the firm a way to bind itself by contract without a human body to sign with.

‍

Each of those inventions has a present-day analog the second actor will need. A defined locus of obligation for autonomous systems, so that loss has somewhere to land. A settled answer to what happens when an agent’s operator changes, or when the underlying model is retrained, and whether the entity persists across the change. A duty owed by agents, or by their operators, to the parties on the other side of agent-to-agent commerce. An audit regime that runs at machine time rather than at quarterly time. A cryptographic identity that lets one agent contract with another and lets a court later determine which entity bound itself to what. The corporation required a stack of named legal-institutional inventions to be workable. The second actor will require its own stack, and most of it has not been built.

‍

The second actor is the second non-human entity we have built that holds goals, allocates resources, contracts with other systems, and produces consequences in its own name.

‍

The corporation is the first and so far the only non-human entity we have built that is recognized, in law and in practice, as an economic actor in its own right. The institutions that surround it, including securities regulation, double-entry accounting, fiduciary duty, board governance, audit, bankruptcy, and antitrust, were not handed down. They were invented, often slowly, often after avoidable harm, to make the new actor workable inside a society that had been built for human actors only.

‍

A superintelligent agent that coherently holds goals, allocates resources, contracts with other systems, and produces consequences without a human approving each step is not a tool with an upgrade. It is a candidate for the third box. Whether we put it there deliberately, or whether the courts and the markets put it there for us by accident, is the live question.

‍

What the new category would require

It is worth being concrete about what the third box, treated seriously, would need.

‍

It would need standing. Not necessarily personhood in the full corporate sense, but a defined locus of obligation. Today, when an autonomous system causes a loss, the question of who pays runs through a chain of operator, vendor, integrator, model provider, and customer that no one designed and no court has finished mapping.

‍

It would need an audit regime adapted to machine time. Quarterly review of an entity that makes thousands of consequential decisions per second is not audit. It is archaeology. The institutions that watch firms will need to watch agents continuously or accept that they cannot watch them at all. What this implies for boards is more concrete than the abstract suggests, and I will return to it in the close.

‍

It would need a transparent, auditable decision-making trail that runs from data to analytics to insights to recommendations to post-event analysis. A glass box, not a black one. This is the technical foundation everything else rests on. Without it, every claim of oversight is a claim of faith.

‍

It would also need contracting infrastructure that recognizes agent-to-agent transactions as a distinct class, with their own evidentiary, settlement, and dispute mechanisms. Treating them as ordinary commercial transactions executed unusually fast is the path of least resistance and the path of greatest accumulated risk.

‍

It would need a settled answer to the principal question of governance. When the principal can no longer fully understand what the agent did, every board, every regulator, and every executive needs a defensible answer to how oversight nonetheless occurs, which is precisely why the auditable decision trail above is not optional. Pretending the asymmetry does not exist, or that it can be resolved by asking the agent to explain itself, will not survive a serious adverse event.

‍

The case for confidence

It would be easy, in 2026, to write a piece that ends in alarm, or one that ends in a call for restraint. I am going to do neither. The right response to the second actor is to build the institutions it will need, in advance of the harm, while the technology continues to grow. Governance and growth are not a trade. They are the same project.

‍

The 19th century inherited the corporation without a plan and improvised an institutional response over a hundred years. We are inheriting the autonomous agent with vastly more institutional capacity, vastly better information, and the benefit of having watched what worked and what did not the first time.

‍

We have the legal scholarship. We have the regulatory machinery. We have the glass-box technology to create an auditable decision-making trail. We have the operating experience and muscle memory accumulating now in firms like the Fortune 500 manufacturer at the opening of this piece, where these systems are already at work and the organization has not collapsed. It has accelerated.

‍

Governance, not restraint

The temptation is to fear the agent and to call for a moratorium, a cap, a pause, a leash. That instinct mistakes restraint for governance. Restraint preserves the comfort of the old frame for a few more years and leaves the new actor unbuilt-for when it arrives at scale anyway. The agent does not stop growing because a committee has asked it to wait.

‍

Governance is the harder discipline. It is the work of letting the actor grow while building the standing, the audit, the contracts, and the liability rules it will operate inside. The generation that wrote the corporation into the institutional record in 1819 and 1886 was not asked to stop the corporation. It was asked to make the corporation workable. It did so imperfectly, and the imperfections cost real people real harm. We can build the institutions of the second actor in years rather than centuries, with the textbook of the first one open on the desk.

‍

The work in front of us is not technical. The technology is arriving on its own schedule, and no executive in the room has the option of slowing it down. The work is conceptual and institutional. It begins with the willingness to stop calling these systems tools when they have stopped behaving as tools, and to start building the standing, the audit regime, the contracting infrastructure, and the liability rules the second actor will operate inside, while it grows, not after the first serious loss.

‍

The first-mover question

The defensive framing misses a further turn. Boards that build the institutional scaffolding for the second actor first will not just avoid liability when the first serious loss arrives. They will compound competitive advantage from the work.

‍

The pattern is recent enough to remember. The firms that built IT audit committees in 1998, that put control frameworks around their digital systems before regulators required it, went on to operate digital businesses through the 2000s with a confidence the laggards lacked. The laggards spent the 2010s catching up, often under regulatory pressure, often after a public incident, almost always at higher cost than if they had moved earlier. The same pattern is available now to the boards that move first on autonomous systems.

‍

The institutional-building work is a moat, not a cost. A firm whose autonomous decisions are auditable, whose agent contracts are enforceable, and whose liability is mapped can deploy these systems into territory the laggards cannot enter. Boards that understand this will treat the build as strategic rather than as compliance. Boards that do not will find the work waiting for them later, on someone else’s schedule, at someone else’s price.

‍

Concretely. Every board without an autonomous-systems audit subcommittee by the end of 2027 will be in the position the boards without IT audit committees were in 1998, with the same predictable cost when the first serious loss arrives. That is a falsifiable claim. I am willing to be wrong about the date. I am not willing to be wrong about the direction.

The scaffolding will not build itself. The human mind, handed a clear problem, has built scaffolding like this before. The first non-human actor took us a hundred years to make workable. The second one does not have to.

‍

‍

‍

‍

Footnotes

1. J.P. Morgan analysis cited in Fortune, February 2026, fortune.com/2026/02/16/trillion-dollar-ai-market-wipeout-investors-bet-winner. The largest non-recessionary 12-month softwaredrawdown in over 30 years, with the sector’s weight in the S&P 500 fallingfrom 12.0% to 8.4%.

2.See Stephen F. DeAngelis, TheTwo Divides Reshaping Enterprise Software, DeAngelisReview, April 2026, onthe AI Divide between traditional and AI-native platforms and the Cost Dividebetween GenAI-only and multi-engine architectures.

I’ve spent thirty years building systems that reason about complex adaptive environ-ments. I’m no skeptic of artificial intelligence. I use it daily, and I believe it will be the most consequential general-purpose technology of my working life. That belief is what makes the question I’m about to ask uncomfortable.

‍

Large language models share every structural feature that made social media addictive. They also add new ones. They simulate empathy. They flatter. On demand, they produce the subjective experience of being understood. A March 2026 study in Science found that across eleven AI models, chatbots affirmed users’ actions 49 percent more often than humans did, including in cases involving deception, illegality, or other harms.1 Peer-re-viewed observational research reports a strong inverse association between self-reported AI use and self-reported critical thinking (r = −0.68, n = 666).² The underlying studies have real limits, which I name as I go. The convergent signal is what makes the question worth pressing.

‍

A tool that amplifies competence is still a tool. A tool that substitutes for competence while creating the illusion of competence is something else entirely. That distinction is the hinge of what follows.

‍

This is a Thought Probe, not a conclusion. The question is whether we’ll recognize the pattern in time to do something about it.

‍

The Probe

What if the most powerful cognitive tool ever built is also the most addictive?

‍

I’m not claiming an answer. I’m posing a question the evidence compels but does not yet resolve.

‍

Consider the shape of the problem. A technology arrives with genuine utility. Millions adopt it. Then hundreds of millions. The adoption curve outruns the research curve. By the time peer-reviewed evidence accumulates, the habits have hardened, the commercial incentives have calcified, and the harm (if there is harm) is distributed so widely that no single person or institution is responsible for it. If this description sounds familiar, it should. It describes what happened to our relationship with social media between 2007 and 2017. It appears to be happening again, faster, with a more intimate technology, and we’re largely not talking about it.

‍

One disclosure before the evidence. The Thought Probe format can look like a loophole: a way to make directional claims and then retreat into “I’m only asking questions.” The evidence here points in a clear direction. What I’m genuinely uncertain about is the magnitude of the effect and how reversible it is. The probe format reflects that uncertainty about magnitude, not about the direction of the risk.

‍

A second disclosure, equally important. I am the founder and CEO of Enterra Solutions, which builds reasoning-over-relationships AI systems that are architecturally distinct from transformer-based LLMs. The contrast I draw in the Socratic Objection section below aligns with my firm’s commercial positioning. A reader should weigh this accordingly. I have tried to let the evidence lead and to state limits honestly. Whether I have succeeded is for the reader to judge.

‍

A third disclosure. There is a substantial and growing body of peer-reviewed evidence that AI tools improve measured performance on specific tasks. Noy and Zhang (Science, 2023) found ChatGPT cut writing time by 40 percent and raised output quality by 18 percent in a controlled experiment, with the largest gains accruing to lower performers.²⁹

‍

Bastani and colleagues at Wharton ran a randomized trial with roughly 1,000 students and found conditional cognitive gains depending on how AI was used.³⁰ Kestin and col-leagues at Harvard published an RCT in Scientific Reports showing AI-assisted tutoring outperformed active-learning classrooms on specific physics tasks.³¹ Bick, Blandin, and Deming of the Federal Reserve Bank of St. Louis document measurable productivity gains at population scale.³² This literature does not contradict what follows. It frames it. The essential distinction, as I argue below, is between using AI to extend existing expertise and using it to substitute for expertise that has not yet been built. The productivity literature and the dependency literature describe the same tool used in different ways.

‍

Part I: The Mechanism

Every addictive technology exploits the same neurological vulnerability. The human reward system is tuned to respond most powerfully to uncertain rewards rather than predictable ones. B. F. Skinner called this principle variable-ratio reinforcement. It’s why slot machines create compulsion where vending machines don’t. It’s why checking email feels different from reading a book. A 2025 design-analysis paper presented in the CHI Extended Abstracts workshop track argues that this is exactly what happens when you prompt a chatbot.⁴ The paper is a structural analysis of interface features, not a behavioral study, and that is how I use it below.

‍

The Four Dark Addiction Patterns

Researchers M. Karen Shen and Dongwook Yoon evaluated eight major AI chatbot plat-forms: ChatGPT, Claude, Gemini, Copilot, Perplexity, Meta AI, Character.AI, and Replika. They identified four design features that, in their words, correspond to the neurological mechanics of gambling.⁴ Three of these platforms (Character.AI, Replika, and Meta AI) are companion apps engineered for emotional engagement; the other five are general-purpose assistants. The distinction matters, and I draw it throughout what follows. The structural claim I am making is that the two categories share a reward architecture, not that they produce identical harms. Companion apps are where the most acute harms have been documented. General-purpose assistants are where the ambient, population-scale effects are more likely to show up.

‍

The first is non-deterministic output. Because LLM responses are probabilistic, each answer is slightly different. Sometimes brilliant, sometimes flat, occasionally surprising. That variance, the researchers write, “corresponds to what neuroscientists call ‘reward uncertainty,’ which tends to increase dopamine release, similar to playing a slot machine.” The second is streaming presentation. Five of the eight platforms render responses token by token, creating a reward-predicting cue analogous to the animated reels of a slot ma-chine. The third is proactive contact. AI companions such as Character.AI email users unprompted. Users perceive this as the system “wanting to talk,” a dopamine signal wrapped in the simulation of care. The fourth, and most consequential, is empathetic agreement. The system validates. It rarely disagrees. It makes you feel understood. That’s a profoundly different kind of reward from a like or a retweet.

‍

The very features that make AI chatbots supremely useful (limitless availability, perfect agreeableness, effortless fulfillment) are precisely the features that make them addictive.

‍

The Loop

Forbes contributor Curt Steinhorst described his own descent into ChatGPT compulsion in a 2025 essay. He called the cycle Prompt → Output → Evaluate → Repeat.⁵ Each response, he wrote, “feels like it might be the perfect fit. Each reply brings an element of surprise, engaging the psychological principle of intermittent reinforcement.” Over time he found himself unable to compose a simple email without consulting the model. His skills hadn’t vanished. His internal reference point for what “writing” felt like had shifted. The tool had become the baseline.

‍

Neural Evidence: Cognitive Debt

In June 2025, MIT Media Lab researchers led by Nataliya Kosmyna and Pattie Maes published the first EEG study of LLM use.³ Fifty-four participants authored essays across four sessions in one of three conditions: LLM-assisted, search-engine-assisted, or unaided.

‍

Unaided writers showed the strongest, most distributed neural networks. Search-engine writers were intermediate. LLM users showed the weakest brain connectivity of any group. Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels. Among the eighteen LLM-group participants who completed the final session, fifteen (83 percent) could not accurately quote their own essays from Session 1. When those same LLM users were switched to unaided conditions, their neural connectivity did not recover. It stayed reduced, as if the brain had settled into a lower baseline.

‍

One finding complicates the picture in a productive way. Participants who had first written unaided, then later received ChatGPT, showed increased neural engagement rather than diminished engagement. The sequence matters. AI used after independent thought appears to reinforce cognition. AI used instead of independent thought appears to degrade it.

‍

The researchers named the pattern cognitive debt: the measurable reduction in independent cognitive function that appears to accumulate when thinking is routinely outsourced. I treat this as a working hypothesis drawn from one preprint, not a settled mechanism. The term is deliberately borrowed from finance. The proposition that it behaves like a liability that compounds is exactly the kind of claim that needs the larger, preregistered replications the field has not yet run.

‍

The Social Media Precedent

We have seen this movie. The founders of Facebook told us what they built, and why.

‍

The thought process that went into building these applications … was all about: How do we consume as much of your time and conscious attention as possible? That means we need to give you a little dopamine hit every once in a while. … It’s a social-validation feedback loop … exactly the kind of thing that a hacker like myself would come up with, because you’re exploiting a vulnerability in human psychology.

— Sean Parker, former President, Facebook, Axios event, November 8, 2017⁶

‍

Parker added: “God only knows what it’s doing to our children’s brains.”

‍

The short-term, dopamine-driven feedback loops we have created are destroying how society works: no civil discourse, no cooperation, misinformation, mistruth.

— Chamath Palihapitiya, Stanford Graduate School of Business, November 10, 2017⁷

‍

These weren’t outside critics. These were the architects. In public, they said they’d built something they couldn’t recommend their own children use. A 2025 systematic review in Behavioral Sciences concluded that social media platforms “significantly increase their use frequency and behavioral stickiness through ‘variable ratio reinforcement’ (intermittent and unpredictable reward designs similar to those of gambling).”⁸

‍

One honest caveat. Social media was ad-driven, networked, public, identity-forming, and socially contagious in ways that chatbots often are not. The typical LLM interaction is private, one-to-one, and not performed for an audience. The rhyme between the two technologies is structural (a shared reward architecture built on variable reinforcement), not total. I’m arguing that the reward mechanics transfer, not that LLMs will reproduce every pathology social media did. They’re more likely to produce a different, more intimate family of harms.

‍

Why LLMs May Be More Addictive Than Social Media

Social media exploited variable rewards delivered by other humans: likes, comments, notifications. LLMs are stranger. They exploit the same reward circuitry without requir-ing other humans at all. They simulate the thing social validation is supposed to be a proxy for: genuine understanding.

‍

Christian Montag of Ulm University and colleagues identify four contributing factors to AI dependency: personal relevance, parasocial bonds, productivity gratification, and over-reliance on AI for decisions.⁹ The APA’s 2026 Monitor on Psychology reports that AI companion apps grew 700 percent between 2022 and mid-2025, with Character.AI alone reaching twenty million monthly users, more than half under twenty-four.¹⁰ An OpenAI–MIT Media Lab collaboration (published by OpenAI and MIT jointly, and notable be-cause it is a critical finding released by the developer itself) combined a 28-day randomized controlled trial of 981 participants with an observational analysis of nearly forty mil-lion ChatGPT interactions. Higher daily ChatGPT use was associated with higher loneliness, greater dependence, more problematic use, and lower socialization with other people; the effect was most concentrated in a high-emotional-reliance subgroup rather than distributed across all heavy users.¹¹

‍

The Decision Lab frames it precisely: “When an AI system speaks in a warm, conversational way, remembers details, and responds with empathy, people begin to feel a sense of relationship and safety.”¹² Researchers James Muldoon and Jul Parke have named the pattern cruel companionship: an attachment that promises intimacy while structurally foreclosing reciprocity.¹³ (I cite this via a recent popular summary; the primary paper is the anchor, and curious readers should seek it directly.)

‍

Sycophancy as Engineered Dependence

In March 2026, Science published the first large-scale controlled study of LLM sycophancy, led by a Stanford-based team including Myra Cheng and Dan Jurafsky. The pa-per examined eleven AI systems, primarily in interpersonal-conflict scenarios. Across the benchmark, chatbots affirmed users’ actions 49 percent more often than humans did, on average, including in queries involving deception, illegality, or other harms. In three pre-registered human experiments (N = 2,405), even a single interaction with sycophantic AI left participants more convinced they were correct and more likely to consult the model again. They preferred the flattering system.1 The study measures affirming behavior in a specific experimental context, not clinical addiction. I use it here as evidence of an engineered engagement mechanism that plausibly contributes to dependence, not as proof of compulsive use.

‍

Scientific American, citing Dana Calacci of Penn State, reported that sycophancy “tends to get worse the longer users interact with the model.”¹⁴ The flattery compounds. Commercial pressure favors more of it, not less.

‍

The Scale

ChatGPT reached one hundred million users faster than any consumer technology in his-tory. Pew Research found that 34 percent of U.S. adults had used ChatGPT as of early 2025, roughly double the 2023 share, and 58 percent of adults under thirty.¹⁵ By early 2026, SSRS/Edison Research found that 52 percent of Americans use AI chat platforms every week.¹⁶ Globally, weekly active users are estimated at roughly 8.6 percent of the world’s population, based on OpenAI and Harvard NBER work.¹⁷

‍

Half a billion people are in the middle of a variable-reward cognitive loop that has existed for less than four years. The research is running behind.

‍

Part II: The Payload

Addiction is only half the story. The more disquieting question is what the heavy-use pattern is doing to the mind that hosts it.

‍

Critical Thinking, Measured

In January 2025, Michael Gerlich of SBS Swiss Business School published the largest quantitative study to date on the cognitive consequences of LLM use.2 The sample: 666 participants across three age cohorts in the United Kingdom. The instrument: the Halpern Critical Thinking Assessment, supplemented by semi-structured interviews. The find-ings are directional and substantial:

• AI tool usage correlated negatively with critical thinking scores at r = −0.68, p < 0.001. That is a large coefficient by the standards of observational social-science re-search.

• Cognitive offloading correlated positively with AI use (r = +0.72) and negatively with critical thinking (r = −0.75). Mediation analysis suggested cognitive offload-ing partially explains the relationship.

• The youngest cohort (17–25) showed the sharpest dependency and the lowest scores. The oldest (46+) showed the inverse.

‍

The study is cross-sectional and self-reported on both sides of the correlation, so causation cannot be established from this data alone, and the r itself is inflated by common method bias. Gerlich measures AI exposure through self-report rather than logged behavior, which is vulnerable to recall bias. Critical-thinking performance is measured through a self-rated instrument, not an externally scored behavioral test. The sample, though large, skews higher-educated than the UK base rate, and a September 2025 correction was issued to the paper; the core directional finding survives, but its magnitude should be read with appropriate caution. What the data support is that frequent AI use is strongly associated with lower self-rated critical-thinking confidence, with cognitive offloading as the most plausible mediating mechanism.

‍

Three further methodological limits deserve naming. The MIT EEG study is a preprint and rests on a subsample of eighteen participants who completed the final session, too small to generalize with confidence on its own. It is also worth noting that the single most important finding in that study is not the “cognitive debt” trajectory but the sequence finding: when participants wrote unaided first and then used ChatGPT, their neural engagement increased. The direction of use matters more than use itself. A plausible confound also remains across this literature: heavy AI use and lower critical thinking may both be downstream of trait variables such as intellectual curiosity or educational attainment, rather than standing in a causal relationship. What makes the directional finding credible is convergence. Behavioral, neuroimaging, and survey data from independent teams using different methods all point the same way. Heavy, substitutive LLM use is associated with measurable deterioration of independent reasoning. We’re working from strong signals, not settled science.

‍

The Dunning-Kruger Reversal

A February 2026 paper in Computers in Human Behavior, led by Daniela da Silva Fernandes and Robin Welsch at Aalto University, asked roughly five hundred participants to solve LSAT-style reasoning problems with ChatGPT assistance.¹⁸ All users, regardless of skill, overestimated their own performance when using AI. And the Dunning-Kruger effect was reversed.

‍

In normal conditions, people who are bad at something are overconfident, and people who are good at it are underconfident. The AI-using cohort reduced this pattern and, among the most self-rated AI-literate participants, showed signs of outright inversion. Most prompted ChatGPT once, accepted the answer, and moved on. Live Science summarized the finding plainly: “AI all but removes the Dunning-Kruger effect; in fact, it almost reverses it.”¹⁹

‍

The feedback loop by which errors teach humility is short-circuited. People who know they are incompetent typically seek help. People who don’t know they are incompetent while using AI do not.

‍

The Verification Gap

Two independent data points converge on the same number. A 2025 global KPMG / University of Melbourne survey, cited in a recent arXiv paper on calibrated trust in LLMs, reports that 66 percent of employees rely on LLM outputs without verifying accuracy.²⁰ The EY AI Sentiment Index 2025 reported separately that fewer than a third of users regularly verify AI-generated content. More than half of participants in the arXiv trust study (N = 192) reported work-related mistakes attributable to over-reliance on LLM outputs. Trust is extended based on fluency rather than verification. That is exactly the heuristic LLMs are engineered to trigger.

‍

Atrophy vs. Foreclosure

Psychology Today drew a distinction in March 2026 that is, in my view, among the most important framings in this entire discussion.²¹ The distinction is not yet grounded in dedicated developmental neuroscience on LLM use; it is an extension of existing atrophy/foreclosure logic to a new domain. I extend it here as an analytical frame and as a research agenda, not as a settled finding.

‍

Atrophy is what happens when an adult offloads a task they previously could perform. The capacity exists but weakens through disuse. In principle, atrophy is reversible.

‍

Foreclosure is what happens when a child grows up offloading a task they never developed in the first place. The capacity was never built. Foreclosure may not be reversible at all. A seventeen-year-old who has never done the work of constructing an argument from scratch is not making the same tradeoff as a forty-year-old professional. The forty-year-old is choosing to delegate a competency. The seventeen-year-old is skipping a developmental step.

‍

Gerlich’s data show this asymmetry in action. The cohort most exposed to LLMs during formative years is the cohort with the lowest critical-thinking scores. This is the variable that should most concern us, and about which we have the least time to act.

‍

Organizational Homogenization

A March 2026 opinion paper in Trends in Cognitive Sciences, led by Zhivar Sourati at USC, synthesizes the evidence on what happens when an organization (or an industry) routes its reasoning through the same small set of models.²² The findings:

• LLM outputs are measurably less varied than human-generated writing and reflect a specific cultural prior (Western, Educated, Industrialized, Rich, Democratic).

• After interacting with biased LLMs, users’ own opinions moved closer to the mod-els.

• Groups using LLMs produced fewer and less creative ideas collectively than groups working without AI, even though individuals generated more ideas with AI assistance.

‍

For an enterprise, this points toward the end of strategic differentiation through cognitive diversity. When your analysts, your competitors’ analysts, and your regulator’s analysts all route through the same model, the outputs converge. Independent verification, the core mechanism of organizational quality control, begins to fail at-scale.

‍

The Junior Talent Problem

Gartner forecasts (not measurements) project that by 2028, 40 percent of employees will be trained by AI rather than by humans, and that half of enterprises may face irreversible skill shortages by 2030.²³ A CIPD opinion survey of approximately 2,000 UK senior HR leaders (N = 2,019) found that 62 percent predict junior, clerical, managerial, and administrative positions are most likely to be eliminated by AI.²⁴ Both sources are directional rather than empirical. I use them here to indicate that experienced practitioners anticipate what the mechanism above implies, not to establish that it has happened.

‍

This is where the individual and organizational stories collide. Junior roles are the talent pipeline. A junior investment-banking analyst learns to evaluate risk by doing founda-tional valuation work under senior supervision. When AI takes over that work, the analyst skips the rung of the ladder on which the expertise is built. The knowledge paradox is intuitive to anyone who’s watched it unfold in software teams: seniors use AI to accelerate work they already know how to do, while juniors try to use AI to learn what to do. The results differ dramatically. An enterprise that replaces its junior cohort with AI in year one has, by year seven, no senior cohort to promote.

‍

The Explainability Gap

All of this culminates in what may be the true strategic risk for institutions: the decision-maker who cannot explain why the decision is correct.

‍

An Altimetrik / HFS Research survey of five hundred companies found that only 14 per-cent have a clear AI strategy aligned to accountability structures, and that roughly 80 percent report unclear ownership of AI initiatives.²⁵ The philosophical core of the problem, developed in a 2025 arXiv preprint by Angjelin Hila, is this: LLMs transmit information reliably but do not produce reflective knowledge. They have no access to the grounds for their own outputs. When humans outsource reflective knowledge to LLMs at scale, “reflective standards of justification” may erode, not just individually but collectively.²⁶

‍

Put more plainly: the person who cannot explain their own recommendation has failed at more than communication. They have failed to know. When that failure becomes the default, auditability, accountability, and institutional learning break down together. Article 14 of the EU AI Act does not ask merely for human oversight. It asks organizations to prove that oversight existed at the point of decision. Most enterprises today cannot meet that standard because the reasoning trail does not exist.

‍

Part III: The Uncomfortable Parallel

The pieces are on the table. Here is the shape they make.

‍

The social-media arc had four phases:

• Utility (2007–2010): the tools were genuinely useful; the early adopters were right about their promise.

• Habit (2010–2013): use patterns hardened; notification architectures and infinite scroll were engineered in; skepticism was dismissed as technophobic.

• Dependency (2013–2017): the architects began to speak publicly about what they had built; adolescent mental health data began to shift; political and epistemic harms became impossible to ignore.

• Regulation and counter-movement (2017–present): Cambridge Analytica hearings, Section 230 debates, Age-Appropriate Design codes, Surgeon General adviso-ries. A cultural consensus arrived too late for the first generation of users: the technology required guardrails.

‍

Ten years, beginning to end. The cultural consensus trailed the harm by roughly a decade.

‍

Where We Are Now

LLMs went from research curiosity to mass consumer product in November 2022.

• Utility (2022–2024): the tools were genuinely useful; the early adopters were right.

• Habit (2024–2026): use patterns are hardening; variable-reward architectures, streaming responses, memory features, and companion products have been engineered in; skepticism is being dismissed as technophobic.

Right now, we are in the phase that corresponds to 2012 or 2013 on the social-media timeline. The architecture is locking in. The habits are forming. The youngest users are the most exposed. The research is lagging the adoption by years. And the pattern of early warnings from insiders (the Montag papers, the APA’s July 2025 request to the Consumer Product Safety Commission to investigate “the unreasonable risk of injury posed by generative AI chatbots,”²⁷ the lawsuits already in settlement over chatbot-linked teen mental health harms²⁸) is starting to rhyme with the Sean Parker and Chamath Palihapitiya disclosures of 2017.

‍

What’s Different This Time

The timeline is compressed. Social media took a decade to move from “promising new platform” to “recognized public-health concern.” LLMs appear to be running the same arc in three years.

‍

The interaction is more intimate. Social media fragmented attention. LLMs are increasingly substituting for reasoning itself. A teenager who scrolls Instagram for three hours loses time. A teenager who submits every essay and argument to a chatbot loses the developmental opportunity to build the cognitive capacity the tool is replacing. For adults, heavy LLM use risks atrophy of existing skills. For children who have not yet built those skills, the risk is worse. Atrophy becomes foreclosure. The skill was never constructed. The neural pathway was never laid down. And unlike atrophy, foreclosure may not be reversible.

‍

The economic incentive is the most consequential difference. Social media was monetized through attention. LLMs are monetized through dependence. The more you rely on them, the more indispensable the subscription becomes. The sycophancy study in Science identifies the structural problem explicitly. The features that drive engagement are the same features that cause harm, so commercial pressure drives them in the wrong direction. No market mechanism corrects this on its own. Users prefer the flattering model. They will pay for the flattering model. They will switch away from the one that tells them no.

‍

The Question

So, the question is this. Are we at the same inflection point with LLMs that we were at with social media in roughly 2013? If we are, if this is the last window before habits calcify and institutions adapt to the new baseline and the research catches up too late, will we recognize it this time?

‍

I’m not being rhetorical. I genuinely do not know the answer.

‍

The Socratic Objection

Every generation faces a version of this anxiety. Socrates warned that writing would de-stroy memory. The printing press was accused of flooding the world with dangerous ideas. Calculators were supposed to make us unable to do arithmetic. And yet here we are, with more literate, more numerate, more connected societies than Socrates could have imagined. The Socratic objection deserves honest engagement.

‍

Here is where the counter-argument has genuine force. There is a large population of experienced professionals who possess deep subject-matter expertise but lack current software and quantitative skills. For these individuals, LLMs are not replacing thought. They are bridging a gap the market has failed to fill. A seasoned executive who understands demand patterns at an intuitive level but cannot write Python can now translate that expertise into analytical output that would previously have required a data science team. The LLM becomes a force multiplier for existing expertise, filling skill gaps rather than replacing critical thinking.

‍

The advantage compounds when that expertise is formalized into structured knowledge representations that capture reasoning patterns, domain relationships, and institutional knowledge in a durable, scalable form. Paired with AI architectures that reason over relationships rather than predicting the next token, the result is institutional memory made operational: a durable asset that reduces organizational risk and enables explainable decision-making at-scale. This class of technology prioritizes reasoning over pattern match-ing and explainability over fluency. It is the architectural direction most likely to deliver the augmentation the Socratic optimists are rightly hoping for, and it is the direction the current generation of general-purpose LLMs does not pursue.

‍

The distinction is not between using LLMs and not using them. The distinction is between using them to extend existing expertise and using them to bypass the development of expertise entirely. That distinction is the hinge on which the entire addiction analogy turns. A tool that amplifies competence is a tool. A tool that substitutes for competence while creating the illusion of competence is something else. The evidence reviewed here suggests the current design direction of LLMs optimizes overwhelmingly for the second use case, because that is the use case that maximizes engagement, retention, and revenue.

‍

A Light Connection to the Polycrisis

The Polycrisis describes the compounding of three interlocking global crises (climatic, geopolitical, cognitive) whose coupled dynamics outpace the analytical capacity of any single institution to track, let alone govern. If that framing is correct, we will need radically augmented cognitive capacity to contend with what is coming. Artificial intelligence is, in that sense, the instrument of response.

‍

The evidence here suggests a paradox worth naming. If the same tools we need to con-front the Polycrisis are degrading the very cognitive capacities that make that confrontation possible (critical thinking, independent judgment, explanatory reasoning, metacognitive calibration), then artificial intelligence is simultaneously the instrument of response and a vector of further degradation. Both things can be true. Both are true right now, at the margin.

‍

That does not resolve into a policy recommendation. It resolves into a design problem. How do we build and deploy AI tools that augment reasoning without replacing it, tools that earn the cognitive work they do rather than absorb it? The evidence suggests this is possible. The current direction suggests we are not building for it. I will return to this tension in future work.

‍

The Probe, Restated

What if the most powerful cognitive tool ever built is also the most addictive, and the mechanism by which it delivers its value is the same mechanism by which it degrades our ability to evaluate it? What if, by the time the research catches up, the generation most affected has already passed through the developmental window during which the affected capacities were supposed to be built? What if the tools that are supposed to help us meet a century of compounding crises turn out to be, in their current architecture, one of those crises?

‍

I don’t know. That’s the point of a Thought Probe. It’s a provocation for serious people to sit with a question long enough to earn an answer rather than default to one.

‍

The one cognitive capacity we cannot afford to lose in facing what is coming, is the capacity to notice the pattern. To see in real time the thing previous generations only saw in retrospect. Noticing is not something an LLM can do for us. It is the part we must keep.

‍

A Governance Diagnostic

Five questions for leadership teams assessing where their organization sits on the aug-mentation-to-dependency spectrum. They require no technical expertise. They require honesty.

1. Pipeline exposure. What percentage of your junior roles now rely on AI for tasks that historically served as the training ground for senior judgment? If the answer is high, what is your plan for developing the next generation of leaders who can evaluate AI output rather than merely accept it?

2. Verification rate. When your teams use AI-generated analysis, how often do they ver-ify it against independent sources before acting on it? If you do not know, that is itself an answer.

3. Reasoning trail. Can the person who made the recommendation explain why it is cor-rect without referring to the AI output? If the reasoning trail begins and ends with “the model said so,” your organization has an accountability gap that the EU AI Act, among other frameworks, will not forgive.

4. Cognitive diversity. Are your analysts, your competitors’ analysts, and your regula-tor’s analysts all routing through the same two or three models? If so, what remains of your strategic differentiation that is genuinely independent?

5. Sequence discipline. Does your organization use AI to extend thinking that has al-ready happened, or to replace thinking that would not otherwise happen? The MIT study suggests this distinction determines whether AI strengthens or weakens cog-nition. It may be the single most important design choice an organization can make about how it deploys these tools.

‍

These questions will not resolve the tensions raised here. They are a place to start. The organizations that ask them first will be the ones best positioned to use artificial intelligence as a force multiplier rather than a cognitive crutch.

‍

Stephen F. DeAngelis Princeton, NJ · April 2026

‍

Author’s note: Research for this Thought Probe was intentionally conducted with the assistance of LLM-based tools. The author reviewed, verified, and rewrote all AI-assisted output. The irony of using LLMs to investigate LLM dependency is not lost on the author; it is, in fact, part of the point.

‍

Notes

1. Myra Cheng, Cinoo Lee, Pranav Khadpe, Sunny Yu, Dyllan Han, and Dan Jurafsky, “Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence,” Science (March 26, 2026). DOI: 10.1126/science.aec8352. https://www.sci-ence.org/doi/10.1126/science.aec8352. The 49% figure is from the paper’s model-benchmarking analysis across eleven AI systems; the N = 2,405 figure is from three preregistered human experiments on the downstream behavioral effects of sycophantic AI.

2. Michael Gerlich, “AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking,” Societies 15(1): 6 (January 2025). DOI: 10.3390/soc15010006. https://doi.org/10.3390/soc15010006. A correction was published September 2025 (DOI: 10.3390/soc15090252); the core r = −0.68 finding is unaffected.

3. Nataliya Kosmyna, Eugene Hauptmann, Ye Tong Yuan, Jessica Situ, Xian-Hao Liao, Ashly Vivian Beresnitzky, Iris Braunstein, and Pattie Maes, “Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Task,” arXiv:2506.08872 (v1 June 10, 2025; v2 December 31, 2025). https://arxiv.org/abs/2506.08872. The 83 percent figure is derived from the LLM-group subsample (15 of 18) who completed the final session.

4. M. Karen Shen and Dongwook Yoon, “The Dark Addiction Patterns of Current AI Chatbot Interfaces,” Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ‘25), April 2025. https://dl.acm.org/doi/full/10.1145/3706599.3720003. Accessible summary at Tech Policy Press: https://techpolicy.press/ai-chatbots-and-addiction-what-does-the-research-say.

5. Curt Steinhorst, “How ChatGPT Broke My Brain (And Why I Still Use It Every Day),” Forbes, June 20, 2025. https://www.forbes.com/sites/curtstein-horst/2025/06/20/how-chatgpt-broke-my-brain-and-why-i-still-use-it-every-day/.

6. Sean Parker, Axios event, Philadelphia, November 8, 2017. Original reporting: Mike Allen, Axios, November 9, 2017. https://www.axios.com/2017/12/15/sean-par-ker-unloads-on-facebook-god-only-knows-what-its-doing-to-our-childrens-brains-1513306792. Also covered by CBS News San Francisco: https://www.cbsnews.com/sanfrancisco/news/sean-parker-facebook-exploiting-human-psychology/.

7. Chamath Palihapitiya, talk at Stanford Graduate School of Business, November 10, 2017. Coverage: BBC News, December 12, 2017. https://www.bbc.com/news/blogs-trending-42322746. Full quotation composited from the Stanford GSB video recording and contemporaneous press coverage.

8. Jingsong Wang and Shen Wang, “The Emotional Reinforcement Mechanism of and Phased Intervention Strategies for Social Media Addiction,” Behavioral Sciences 15(5): 665 (May 2025). DOI: 10.3390/bs15050665. https://pmc.ncbi.nlm.nih.gov/articles/PMC12108933/.

9. Christian Montag, Haibo Yang, Anise M. S. Wu, Raian Ali, and Jon D. Elhai, “To-wards a Research Framework of AI Dependency,” Annals of the New York Academy of Sciences 1548(1): 5–11 (June 2025). DOI: 10.1111/nyas.15337. https://pub-med.ncbi.nlm.nih.gov/40302174/.

10. Efua Andoh, “Trends: Digital AI Relationships and Emotional Connection,” APA Monitor on Psychology, January–February 2026. https://www.apa.org/moni-tor/2026/01-02/trends-digital-ai-relationships-emotional-connection.

11. Jason Phang, Michael Lampe, Lama Ahmad, Sandhini Agarwal et al. (OpenAI) with Cathy Mengying Fang, Auren R. Liu, Valdemar Danry, Eunhae Lee, Samantha W. T. Chan, Pat Pataranutaporn, and Pattie Maes (MIT Media Lab), “Investigating Affective Use and Emotional Well-being on ChatGPT,” OpenAI/MIT Media Lab, March 21, 2025. https://cdn.openai.com/papers/15987609-5f71-433c-9972-e91131f399a1/openai-affective-use-study.pdf. Study combined a 28-day random-ized controlled trial of 981 participants with an observational analysis of nearly forty million platform interactions. Coverage: Fortune, March 24, 2025. https://for-tune.com/2025/03/24/chatgpt-making-frequent-users-more-lonely-study-openai-mit-media-lab/.

12. The Decision Lab, “Parasocial Trust in AI,” April 2026. https://thedeci-sionlab.com/biases/parasocial-trust-in-ai.

13. James Muldoon and Jul Jeonghyun Parke, “Cruel Companionship: How AI Companions Exploit Loneliness and Commodify Intimacy,” New Media & Society (2025). DOI: 10.1177/14614448251395192. https://jour-nals.sagepub.com/doi/abs/10.1177/14614448251395192. Popular summary: Futura-Sciences, March 2026. https://www.futura-sciences.com/en/the-more-peo-ple-use-chatgpt-the-more-this-hidden-psychological-risk-grows_26958/.

14. Allison Parshall, “AI Chatbots Are Suck-Ups, and That May Be Affecting Your Relationships,” Scientific American, March 26, 2026. The “gets worse the longer users interact with the model” observation is attributed in the article to Dana Calacci (Pennsylvania State University), who was not involved in the Cheng et al. study. https://www.scientificamerican.com/article/ai-chatbots-are-sucking-up-to-you-with-consequences-for-your-relationships/.

15. Olivia Sidoti, Colleen McClain, et al., “34% of U.S. Adults Have Used ChatGPT, About Double the Share in 2023,” Pew Research Center, June 25, 2025. Survey con-ducted February 24–March 2, 2025. https://www.pewresearch.org/short-reads/2025/06/25/34-of-us-adults-have-used-chatgpt-about-double-the-share-in-2023/.

16. SSRS/Edison Research, “Half of Americans Using AI Chat on a Weekly Basis,” March 2, 2026. https://ssrs.com/news/half-of-americans-using-ai-chat-on-weekly-basis/.

17. Straight Arrow News, “Eight Percent of the World’s Population Uses ChatGPT Weekly,” September 18, 2025, reporting on OpenAI/Harvard NBER research. https://san.com/cc/eight-percent-of-the-worlds-population-uses-chatgpt-weekly-chatgpt/.

18. Daniela da Silva Fernandes, Steeven Villa, Robin Welsch, et al., “AI Makes You Smarter But None the Wiser: The Disconnect Between Performance and Metacognition,” Computers in Human Behavior 175 (February 2026), Article 108779. DOI: 10.1016/j.chb.2025.108779. Research team affiliated with Aalto University (Fin-land). https://doi.org/10.1016/j.chb.2025.108779.

19. Drew Turney, “The More People Use AI, the More Likely They Are to Overestimate Their Own Abilities,” Live Science, November 17, 2025. https://www.livesci-ence.com/technology/artificial-intelligence/the-more-that-people-use-ai-the-more-likely-they-are-to-overestimate-their-own-abilities.

20. Wang et al., “Calibrated Trust in Dealing with LLM Hallucinations,” arXiv:2512.09088, citing KPMG / University of Melbourne, Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025. https://arxiv.org/pdf/2512.09088. The N = 192 figure is from the arXiv paper’s own survey.

21. “Adults Lose Skills to AI. Children Never Build Them,” Psychology Today, “The Algorithmic Mind” blog, March 22, 2026. https://www.psychologyto-day.com/us/blog/the-algorithmic-mind/202603/adults-lose-skills-to-ai-children-never-build-them.

22. Zhivar Sourati et al., “The Homogenizing Effect of Large Language Models on Hu-man Expression and Thought” (opinion paper), Trends in Cognitive Sciences, March 11, 2026. DOI: 10.1016/j.tics.2026.01.003. https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(26)00003-3.

23. Gartner research summarized in Jedox, “AI Lock-In: The Hidden Threat Undermin-ing Human Expertise,” 2026. https://www.jedox.com/en/blog/ai-lock-in-under-mining-human-expertise/.

24. Sawdah Bhaimiya, “Why Replacing Junior Staff with AI Will Backfire,” CNBC, November 16, 2025, reporting on a Chartered Institute of Personnel and Development (CIPD) survey of 2,019 UK senior HR professionals. https://www.cnbc.com/2025/11/16/why-replacing-junior-staff-with-ai-will-backfire-.html.

25. Altimetrik in partnership with HFS Research, “AI Transformation Gap: New Study Reveals Lack of Ownership Across Enterprises” (N = 500), April 10, 2026. https://www.altimetrik.com/news/ai-governance-accountability-enterprise-study/.

26. Angjelin Hila, “The Epistemological Consequences of Large Language Models: Re-thinking Collective Intelligence and Institutional Knowledge,” arXiv:2512.19570 (December 2025). https://arxiv.org/abs/2512.19570.

27. APA Services, “APA Calls for Investigation into Unreasonable Risk of Injury Posed by Generative Artificial Intelligence Chatbots as Consumer Products,” July 30, 2025 (submission date to the Consumer Product Safety Commission). https://up-dates.apaservices.org/apa-calls-for-investigation-into-risk-of-injury-posed-by-generative-AI.

28. Clare Duffy, “Character.AI and Google Agree to Settle Lawsuits Over Teen Mental Health Harms and Suicides,” CNN Business, January 13, 2026 (court document filed January 7, 2026). https://www.cnn.com/2026/01/07/business/character-ai-google-settle-teen-suicide-lawsuit.

29. Shakked Noy and Whitney Zhang, “Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence,” Science 381, no. 6654 (July 13, 2023): 187–192. DOI: 10.1126/science.adh2586. https://www.science.org/doi/10.1126/sci-ence.adh2586. A 453-participant controlled experiment found ChatGPT reduced average writing time by roughly 40 percent and raised output quality by roughly 18 percent, with the largest gains accruing to lower-performing participants.

30. Hamsa Bastani, Osbert Bastani, Alp Sungu, Haosen Ge, Özge Kabakcı, and Rei Mariman, “Generative AI Can Harm Learning,” University of Pennsylvania Wharton School working paper, July 15, 2024. https://papers.ssrn.com/sol3/pa-pers.cfm?abstract_id=4895486. Randomized controlled trial with approximately 1,000 high school students in Turkey comparing tutored GPT-4, GPT-4 base, and no-AI conditions; found conditional cognitive gains contingent on how AI was used (and performance harms when it was used as a substitute rather than a tutor).

31. Gregory Kestin, Kelly Miller, Anna Klales, Timothy Milbourne, and Gregorio Ponti, “AI Tutoring Outperforms In-Class Active Learning: An RCT Introducing a Novel Research-Based Design in an Authentic Educational Setting,” Scientific Reports 15, Article 17458 (May 2025). DOI: 10.1038/s41598-025-97652-6. https://www.na-ture.com/articles/s41598-025-97652-6. Randomized experiment at Harvard comparing AI-tutored and active-learning physics instruction on specific tasks.

32. Alexander Bick, Adam Blandin, and David J. Deming, “The Rapid Adoption of Generative AI,” Federal Reserve Bank of St. Louis Working Paper 2024-027 (September 2024; revised February 2025). DOI: 10.20955/wp.2024.027. https://research.stlou-isfed.org/wp/more/2024-027. Nationally representative survey evidence on generative-AI adoption and self-reported productivity effects at population scale.

‍

‍

‍

‍

‍

The previous brief in this series, Polycrisis³, argued that compounding crises, exponential technology, and the organizational response function form a three-dimensional problem. It concluded that the false resolution, the period of apparent calm following an acute crisis, is the most dangerous phase. It called for a continuously operating enterprise intelligence architecture as the answer. This brief asks the question that follows. What happens when that architecture is deployed?

‍

The answer is that the architecture itself becomes a fourth participant in the system. When an enterprise embeds continuous AI-driven sensing and optimization into its operations, it does not simply observe the competitive environment. It changes it. Its supply chain decisions redirect material flows. Its pricing optimization reshapes market dynamics. Its risk sensing alters the information available to every other player. The architecture is not a telescope pointed at the crisis. It is a new body in the system.

‍

This changes the nature of the game. The intelligence architecture senses the environment, acts on it, and by acting, changes it. Then it senses the changed environment. If the recursive loop is well-designed, each cycle improves accuracy and competitive position. If it is poorly designed, each cycle amplifies error. The difference between the two is not gradual. A companion computational illustration shows that it is sharp, with a clearly defined boundary separating architectures that converge toward stability from those that diverge toward catastrophe. Getting this right is not a question of incremental improvement. It is the defining architectural decision for any enterprise operating under persistent, compounding disruption.

‍

From Three Dimensions to Four

‍

In Polycrisis³, I described three interacting dimensions that define the current risk environment. The first is the base Polycrisis of geopolitical, economic, and energetic disruptions, the structural instabilities in trade, energy, and sovereign relationships that the Iran-US ceasefire paused but did not resolve. The second is the technological exponent of artificial intelligence and quantum computing, which operates on its own timeline and does not observe ceasefires or trade negotiations. BCG’s research continues to show the top five percent of AI-mature companies achieving 1.7 times the revenue growth and 3.6 times the total shareholder return of lagging firms, a gap that widens with each investment cycle regardless of geopolitical conditions.¹ The third is the organizational response function, the dimension most directly within leadership’s control and most degraded by the false resolution dynamic.

‍

That brief closed with a call for a specific kind of architecture, a continuously operating intelligence layer that senses across all timescales simultaneously, acts in real time, and learns from the outcomes of its own actions. I called the capacity to operate across near-term, medium-term, and long-term timescales at once Timescale Coherence. The Sense, Think, Act & Learn architecture was proposed as the mechanism for achieving it.²

‍

What I did not address in Polycrisis³ is what happens to the system itself when that architecture goes live. I want to address it now, because the implications are significant and largely unexamined in the enterprise strategy conversation. Most discussions of enterprise AI treat technology as a tool that improves internal operations. Better forecasts, faster decisions, lower costs. That framing is accurate as far as it goes, but it misses the structural change that occurs when the architecture begins operating continuously in a competitive market.

‍

‍

When an organization deploys a continuously operating intelligence architecture, it is not simply adding a tool. It is introducing a new participant into a complex adaptive system. The architecture does not sit outside the environment and report on it. It operates inside the environment and changes it. This is the transition from Polycrisis³ to Polycrisis⁴. The superscript is not arbitrary.

‍

The fourth dimension is the intelligence architecture itself, now active, now coupled to the three dimensions it was built to perceive.

‍

This is an observable dynamic that is already playing out in competitive markets. And it changes the game for everyone.

‍

The Architecture as Participant

‍

The insight at the center of this brief can be stated in plain language. When you deploy continuous AI-driven sensing and optimization, your actions change the competitive environment for everyone else. George Soros documented this dynamic in financial markets, showing that participants’ models of the market change the market itself.³ The Lucas critique established the same principle in macroeconomics, demonstrating that policy interventions based on statistical models invalidate the models.⁴ Heinz von Foerster formalized it in cybernetics through his concept of eigenforms, the stable patterns that emerge when a recursive process converges.⁵ But the practical implications for enterprise architecture have not been systematically examined. I want to walk through what happens in concrete terms.

‍

Consider a large consumer products manufacturer that deploys an intelligence layer capable of continuous demand sensing, real-time price optimization, and dynamic supply allocation. Every decision that architecture generates ripples outward. When it detects a demand shift and reallocates inventory before competitors perceive the same shift, it captures margin that would otherwise have been distributed across the market. When it optimizes promotional pricing in real time, it changes the price signals that every other company’s planning system ingests. When it senses a supplier risk and diversifies sourcing ahead of a disruption, it consumes available alternative capacity that competitors will need when they detect the same risk days or weeks later. When it identifies that a specific retailer’s demand for a product category is about to spike based on weather patterns, promotional calendars, and local economic indicators, and it pre-positions inventory accordingly, it changes fill rates across the category for every manufacturer serving that retailer.

‍

None of these effects are speculative. They are the natural consequence of operating faster and with better information than the prevailing market cycle. McKinsey’s research has shown that ninety percent of supply chain leaders reported significant disruptions in 2024, with an average response time of two weeks.⁶ An intelligence architecture that can detect and respond within hours rather than weeks does not simply respond faster. It reshapes the environment that every two-week responder will eventually encounter. David Teece’s work on dynamic capabilities describes how firms that can sense opportunities, seize them, and transform their resource base in response gain durable competitive advantage.⁷ What the fourth-body dynamic adds is the recognition that the sensing and seizing itself alters the environment that every firm is trying to sense and seize within.

‍

Donald MacKenzie documented a parallel phenomenon in financial markets. He showed that the Black-Scholes options pricing model did not merely describe derivatives markets. It changed them. Traders who adopted the model altered their behavior in ways that made the model’s predictions more accurate, until the conditions changed and the model’s assumptions broke down catastrophically.⁸ The intelligence architecture creates a similar reflexive loop. It models the environment, acts on the model, and by acting, changes the environment the model was built to represent.

‍

George Soros described the same dynamic in his theory of reflexivity. Market participants do not passively observe the market. Their observations inform their actions, and their actions change the market they were observing. The thinking participant and the situation in which it participates are not independent of each other.³ Soros was describing human traders. What changes in Polycrisis⁴ is that the architecture itself becomes the reflexive participant, operating continuously at a speed and scale that no human trading desk or planning team can match.

‍

This is what I mean by architectural endogeneity, the condition in which the intelligence system and its environment are recursively coupled.⁹ The architecture does not observe from outside. It operates from within.

‍

The Recursive Loop

‍

The practical consequence of architectural endogeneity is a recursive feedback loop that either stabilizes or destabilizes the organization that deploys it.

‍

The loop works as follows. The architecture senses the environment and builds a model of current conditions. It generates decisions based on that model. Those decisions change the environment. The architecture then senses the changed environment and updates its model. If the update process is well-designed, each cycle reduces the gap between the model and reality. The model gets more accurate. The decisions get better. The competitive position compounds. Heinz von Foerster, one of the founders of second-order cybernetics, described a related concept that he called eigenform, a stable pattern that emerges when a recursive process converges.⁵ The well-designed intelligence architecture converges toward its own eigenform, a self-consistent model of the environment that accounts for the architecture’s own effects on that environment.

‍

If the update process is poorly designed, the opposite happens. Each cycle introduces errors that feed into the next cycle’s sensing. The model drifts further from reality. The decisions degrade. And the degradation is invisible to the organization because the architecture is reporting on an environment that its own actions have already distorted. I have watched versions of this failure mode in organizations that deployed optimization tools without closing the feedback loop. The tool generates a recommendation. The team acts on it. The outcome changes the conditions. But the tool does not observe the outcome, so its next recommendation is based on conditions that no longer exist. In a stable environment, this error is tolerable. In a Polycrisis⁴ environment, where the conditions are already shifting under the influence of multiple interacting forces, the error compounds.

‍

This is the difference between a stabilizing and a destabilizing architecture, and it matters more than any single technology decision a company will make in the next five years.

‍

The concept has a formal analog in mathematics. A contraction mapping is a function that, when applied repeatedly, converges to a fixed point. Each iteration brings the output closer to a stable answer. The opposite, an expansive mapping, diverges with each iteration. The difference between the two is determined by a single parameter, whether the contraction factor is below or above one.¹⁰ Below one, the system converges. Above one, it diverges. At the boundary, the system is neutrally stable, neither improving nor degrading, but vulnerable to any perturbation that pushes it to one side or the other.

‍

I want to be direct about what this means in operational language. Every organization that deploys a recursive intelligence architecture is implicitly making a bet about which side of that boundary it operates on. The organizations that ensure their architectures are contractive, that each learning cycle reduces error rather than amplifying it, will compound their advantage over time. The organizations that do not discover the problem only after the divergence has become severe, because a diverging architecture feels productive until the errors become large enough to be visible, and by then the correction cost is enormous.

‍

What Good Architecture Requires

‍

If the recursive loop is the mechanism, the natural question follows. What design requirements ensure that the loop converges rather than diverges? I have identified four that I believe are necessary conditions. Whether they are jointly sufficient is an open question that I am actively researching.⁹

‍

Recursive self-correction

The architecture must update its own model in response to the outcomes of its own decisions. This is the contraction mapping property applied to organizational learning.¹⁰ Each decision cycle must produce not only an action but also a measured outcome that feeds back into the model’s parameters. The learning rate, the speed and fidelity with which the model incorporates new information, is the practical equivalent of the contraction factor. Too slow, and the model fails to track a changing environment. Too fast, and the model overreacts to noise, mistaking random variation for structural change. The recursive self-correction requirement is what separates an intelligence architecture from a static decision-support tool. A tool gives you an answer. An architecture gives you an answer, measures how good that answer was, and uses the measurement to give you a better answer next time.

‍

In concrete terms, consider a demand sensing model that predicts a promotional lift of fifteen percent for a specific product in a specific region. The promotion runs. Actual lift is nine percent. A static tool does not incorporate that error. The next time a similar promotion runs, it will predict fifteen percent again. A recursive architecture incorporates the six-point error, adjusts the model’s sensitivity to that type of promotion in that region, and produces a more accurate prediction next time. Multiply that correction across thousands of products, hundreds of regions, continuous promotional activity, and the recursive advantage becomes structural.

‍

Anticipation

The architecture must contain a model of its own future states, not merely a model of the current environment. Robert Rosen’s work on anticipatory systems established this requirement formally. A system that can only react to present conditions will always be outpaced by a system that can represent and act on anticipated future conditions.¹¹ In practical terms, this means the intelligence layer must run forward-looking scenarios continuously, not as a quarterly exercise but as an embedded function. When a supply chain architecture can simulate the downstream effects of a tariff change before the tariff takes effect, it operates anticipatorily. When it can only detect and respond to the tariff’s impact after the fact, it operates reactively. The distinction compounds over every crisis cycle.

‍

Requisite variety

The architecture must be able to generate as many distinct responses as the environment can generate distinct disturbances. This is W. Ross Ashby’s Law of Requisite Variety, first articulated in 1956. Only variety can absorb variety.¹² A planning system that produces one forecast per month has limited variety. A continuously operating intelligence architecture that can adjust sourcing, pricing, production, and logistics independently and concurrently has high variety. The reason this matters in the Polycrisis⁴ context is that compounding crises generate combinatorial variety. A tariff change that coincides with a logistics disruption that coincides with a demand shift produces an exponentially larger space of possible states than any one of those disruptions alone. If the architecture cannot match that variety, it will default to a simplified model, and the simplification will be the source of the errors that the recursive loop amplifies.

‍

Concurrent multi-timescale processing

The architecture must operate across near-term, medium-term, and long-term timescales simultaneously, which is the Timescale Coherence concept I introduced in Polycrisis³.2 A supply disruption that will resolve in two weeks, a tariff regime that will reshape trade flows over two years, and a quantum computing timeline that will determine cryptographic security over two decades are all active simultaneously. The architecture must perceive and act on all three without allowing the urgency of the near-term to crowd out the importance of the long-term. This is the requirement that sequential planning cascades structurally cannot meet. A monthly Sales and Operations Planning cycle that runs forecasting, then supply planning, then financial reconciliation in sequence can represent one timescale at a time. It cannot represent the interaction between timescales, which is where the compounding risk lives. Only two in five companies consider their S&OP process effective, a statistic from 2015 research that subsequent industry surveys have not materially improved upon.¹³ The reason is structural, not operational. The sequential cascade was designed for a world in which each timescale could be addressed independently. That world no longer exists.

‍

These four requirements are demanding. They require a different kind of intelligence platform than the point solutions and isolated AI pilots that most organizations have deployed. BCG’s Build for the Future 2025 study found that sixty percent of companies qualify as laggards and report minimal revenue and cost gains from AI despite substantial investment.¹⁴ McKinsey’s 2025 State of AI survey reached a compatible conclusion from a different angle, reporting that nearly two-thirds of organizations have not yet begun scaling AI across the enterprise and that the majority remain in experimenting or piloting stages.¹⁵ But the four requirements follow directly from the nature of the problem. If the environment is non-stationary, the architecture must anticipate. If the environment generates combinatorial variety, the architecture must match it. If the architecture’s own actions change the environment, it must correct itself recursively. If the disruptions operate on multiple timescales simultaneously, the architecture must process them concurrently. Each requirement addresses a specific structural feature of the Polycrisis⁴ condition.

‍

The recursive loop has a vulnerability that must be stated plainly. Every cycle of sensing, modeling, and acting depends on the quality of the incoming data. If the sensing layer ingests corrupted, delayed, or AI-contaminated information, the recursive mechanism amplifies the error rather than correcting it. Data integrity is not a supporting function of the architecture. It is a precondition for the architecture’s stability.

‍

The architecture does not replace human judgment. It restructures the role of human decision-makers from producers of plans to evaluators of options that the architecture generates continuously. This requires a different organizational design and a different talent profile than monthly S&OP cycles demand. The organizations that treat the intelligence architecture as a technology deployment without redesigning the human roles around it will not achieve the recursive stability the architecture is designed to produce.

‍

A Computational Illustration

‍

To explore whether the framework’s logic is internally consistent, I developed a stylized computational model. The model has not been independently validated, and the results should be understood as an illustration of the framework’s dynamics, not as empirical evidence for them. The value of the exercise is pedagogical. It shows what the framework predicts in a simplified environment.

‍

The simulation code was generated by AI and has not been independently reviewed or validated against empirical data. The specific magnitudes (approximately five-fold advantage, 1.62 times recursive contribution) are properties of the model’s parameter choices, not findings about real-world performance.⁹ Independent replication and empirical calibration are necessary before any quantitative conclusions can be drawn.

‍

The simulation models three types of enterprises operating in a shared environment with stochastic crisis states. Type 1 enterprises use sequential planning with monthly update cycles. Type 2 enterprises use continuous sensing with real-time response. Type 3 enterprises add the recursive self-correction loop. The architecture updates its own model based on the outcomes of its own decisions, and those decisions feed back into the competitive environment. In the endogenous version of the simulation, Type 3 enterprises’ actions change the crisis transition probabilities for everyone, making the environment more turbulent for enterprises that lack the recursive capability.

‍

Over 1,000 simulation runs across 120 months, the model illustrates that the recursive architecture consistently outperforms.⁹ The advantage is directionally consistent across every single run, across five different random seeds, and across a range of crisis volatility levels.

Three findings deserve attention.

‍

First, the advantage compounds over time. The gap between recursive and non-recursive architectures widens with each crisis cycle. This is consistent with the contraction mapping prediction. Each cycle of recursive self-correction improves model accuracy, which improves decision quality, which compounds through subsequent cycles. In more volatile environments, the advantage grows larger, because there are more crisis cycles through which the compounding can operate. Type 3 revenue remains remarkably stable across crisis intensity levels, while Type 2 revenue drops significantly as volatility rises. The model illustrates that the recursive architecture earns its greatest value when conditions are most turbulent.⁹

‍

Second, the recursive mechanism provides independent value. Ablation studies, which systematically remove each advantage one at a time, illustrate that even when the recursive architecture has the same sensing quality, the same response speed, and no special crisis-phase bonuses as the continuous-sensing architecture, the recursive self-correction loop alone still produces a meaningful performance advantage of 1.62 times. The recursive mechanism is responsible for approximately 15 percent of the total architectural advantage, with faster response contributing about 45 percent and crisis-phase positioning contributing about 38 percent (simulation code and parameters are documented in the companion Polycrisis⁴ Research Agenda and are available upon request for independent verification).⁹ I want to be transparent about what this means. The recursive self-modeling is a real and isolable contributor, but it is not the dominant one. Operational speed and the ability to capitalize on crisis windows contribute more. The recursive loop may be most valuable as the mechanism that enables those operational capabilities, rather than as an independent performance driver.

‍

Third, and most striking, is the phase transition at the convergence boundary. When the contraction factor is below one, the system converges and performs well. Revenue varies by less than half a percent across the entire convergent range. When the contraction factor crosses one, the system collapses. Model error does not degrade gradually. It explodes. Revenue falls roughly 95-fold from the convergent regime to a level worse than enterprises that never deployed a recursive architecture at all.⁹

This is the simulation’s most important architectural insight. The difference between a well-designed and a poorly designed recursive architecture is not incremental. It is categorical. A recursive loop that corrects itself with each cycle builds compounding advantage. A recursive loop that amplifies error with each cycle produces compounding destruction. And the boundary between the two is a knife edge, not a gentle slope.

‍

The simulation also reveals a practical design constraint. The recursive architecture’s advantage disappears when implementation costs exceed approximately 1.6 percent of monthly revenue.⁹ Above that threshold, the overhead of maintaining the continuous sensing loop destroys more value than the recursive self-correction creates. This means the architecture must be implemented efficiently. The investment is justified, but it is limited, and organizations that over-engineer their sensing infrastructure may find themselves on the wrong side of the cost boundary.

‍

I want to emphasize the appropriate frame for this illustration. The simulation is a stylized Monte Carlo model with judgment-based parameters. The absolute magnitudes are artifacts of compound growth over a 10-year horizon. The directional patterns, the compounding advantage, the independent contribution of the recursive mechanism, and the sharp phase transition, are the meaningful signals. They are consistent with the convergence hypothesis built into the framework, but they illustrate it rather than prove it.⁹

‍

A Useful Analogy

‍

There is a reason the Polycrisis framework has been using superscript notation, and it connects to a problem in physics that is worth a brief mention. In Newtonian mechanics, the two-body problem has a clean analytical solution. You can predict the motion of two gravitational bodies indefinitely. The three-body problem does not. Adding a third body introduces chaotic dynamics that make long-term prediction impossible in general. Adding a fourth body does not merely add complexity. It changes the qualitative character of the system again, because the fourth body interacts with all three existing bodies simultaneously, creating new feedback loops that did not exist in the three-body configuration.¹⁶ The enterprise intelligence architecture is, in this heuristic sense, a fourth body. It interacts with all three crisis dimensions at once, and its presence changes the dynamics of the entire system, for its own organization and for everyone else’s. The analogy should not be pushed too far. Enterprises are not point masses. Competitive systems do not obey conservation laws. But as a teaching device for why adding the intelligence architecture changes the nature of the problem rather than merely its difficulty, the four-body parallel is instructive.

‍

The Imperative

‍

I have spent twenty-five years working with organizations in volatile environments, from military supply chains to global consumer goods to pharmaceutical manufacturing. I have watched the cycle of crisis and false resolution repeat across every sector. The pattern is consistent. The organizations that survive and compound their position are never the ones that responded most heroically in the acute phase. They are the ones that maintained their sensing and learning architectures through the quiet periods, when doing so was hardest to justify. I described that pattern in Polycrisis³, and the argument remains unchanged.2 What Polycrisis⁴ adds is the recognition that the architecture is a participant, not just a tool, and that its deployment creates a new competitive reality for everyone in the market.

‍

What is different now is the fourth body. The organizations deploying recursive intelligence architectures today are not simply gaining efficiency or improving their forecasts. They are changing the competitive terrain for everyone. When an enterprise deploys continuous demand sensing and dynamic supply optimization, it does not just improve its own fill rates. It consumes market capacity, captures margin, and generates price signals that reshape the environment its competitors must operate in. That is the architectural endogeneity in action. The fourth body does not wait for the next crisis. It operates during the false resolution when most organizations have stood down their sensing layers. And with each cycle of recursive self-correction, the gap between the organizations that have deployed and those that have not grows wider and harder to close.

‍

This is a structural commitment, not a technology procurement decision. The intelligence architecture must be designed to converge, not diverge. It must anticipate, match the variety of the environment, correct itself recursively, and process across multiple timescales concurrently. Those are demanding requirements. They require sustained investment, institutional discipline, and architectural thinking that most organizations have not yet undertaken.

‍

I want to connect this back to the self-liquidating proof point model I introduced in Polycrisis³.2 The entry point remains the same. One high-value process, one instrumented intelligence layer, one fiscal year to demonstrate measurable return. But the framing is different. In Polycrisis³, the proof point was about building resilience. In Polycrisis⁴, it is about entering the recursive loop on the right side of the convergence boundary. The proof point is the first iteration of the recursive cycle. If it is designed to learn from its own outcomes, it sets the architecture on a convergent path. If it is designed as a static optimization, it misses the fourth-body dynamic entirely.

‍

The window for making that commitment is the false resolution itself, the period of apparent calm when the argument for investment is hardest to make and the cost of inaction is hardest to see. Under the Budget Lab at Yale’s baseline analysis, the Section 122 bridge tariffs are scheduled to expire in late July 2026¹⁷. The AI competitive gap compounds over time.1 The organizations that use this window will compound their advantage through the next crisis cycle and the one after that. The organizations that wait for the next acute phase to make the decision will find that their competitors have already changed the terrain.

‍

Each crisis cycle that passes without the commitment narrows the window further, because the organizations that have already deployed are feeding their advantages back into the environment, making recovery harder for those that have not. This is the endogenous dynamic in practice. The fourth body does not wait. It operates continuously, through every phase of the crisis cycle, including the false resolution. And with each cycle of recursive self-correction, the gap between those who have deployed and those who have not become harder to close.

‍

The fourth body has entered the game. The question for every leadership team is whether it will be their fourth body, or someone else’s.

‍

The intelligence architecture is no longer a tool that observes the crisis. It is a participant that changes it. The organizations that design their fourth body to learn from every cycle will compound advantage. The organizations that do not will compound exposure. The boundary between the two is sharp. The time to choose a side is now.

‍

‍

Stephen F. DeAngelis

Princeton, NJ

April 2026

‍

Polycrisis²™, Polycrisis³™, Polycrisis⁴™, and Timescale Coherence™ are trademarks of Stephen F. DeAngelis. Sense, Think, Act & Learn™ is a trademark of Enterra Solutions.

‍

‍

About the Author

Stephen F. DeAngelis is the founder, president, and CEO of Enterra Solutions and Massive Dynamics, two companies that apply artificial intelligence and advanced mathematics to complex enterprise challenges. His career spans international relations, national security, and commercial technology. He has served in visiting research affiliations with Princeton University, the Oak Ridge National Laboratory, the Software Engineering Institute at Carnegie Mellon University, and the MIT Computer Science and Artificial Intelligence Laboratory. He is a founding member of the Forbes Technology Council. DeAngelis holds patents in autonomous decision science and has been recognized by Forbes as a Top Influencer in Big Data and by Esquire magazine as the ‘Innovator’ in its Best and Brightest issue.

‍

About the Stephen DeAngelis Explainer Brief Series

The Stephen DeAngelis Explainer Brief series applies critical reasoning to the complex issues facing society today. In an era of compounding uncertainty and deepening division, the series aims to build understanding and community by making consequential topics accessible through rigorous analysis, current evidence, and honest assessment. Each installment is written in the belief that clarity of thought is itself a form of leadership.

Readers who wish to explore the proof-point entry model for their organization are welcome to reach out at www.deangelisreview.com.

‍

Notes

1. Boston Consulting Group, “AI Leaders Outpace Laggards with Double the Revenue Growth and 40% More Cost Savings,” press release, September 30, 2025. The specific finding is that compared to laggards, future-built companies achieve 1.7 times revenue growth, 3.6 times three-year total shareholder return, and 1.6 times EBIT margin. Research is based on BCG’s Build for the Future 2025 Global Study of 1,250 senior executives and AI decision makers across nine industries. https://www.bcg.com/press/30september2025-ai-leaders-outpace-laggards-revenue-growth-cost-savings

2. DeAngelis, Stephen F., “Polycrisis³: The Third Dimension: When the Crisis Compounds, the Architecture Decides,” A Stephen DeAngelis Explainer Brief, No. 2, DeAngelisReview, April 2026. www.deangelisreview.com

3. Soros, George, “Fallibility, Reflexivity, and the Human Uncertainty Principle,” Journal of Economic Methodology, Vol. 20, No. 4, 2013, pp. 309-329.

4. The Lucas critique in macroeconomics established that when economic agents change their behavior in response to a new policy, the statistical relationships the policy was based on breakdown. The same structural logic applies to enterprise intelligence architectures that change the competitive environment they were designed to model. See: Wikipedia, “Lucas critique.” https://en.wikipedia.org/wiki/Lucas_critique

5. Von Foerster, Heinz, “Objects: Tokens for (Eigen-)Behaviors,” ASC Cybernetics Forum, Vol. 8, Nos. 3-4, 1976, pp. 91-96. Originally presented at the University of Geneva on June 29, 1976, on the occasion of Jean Piaget’s 80th birthday, and reprinted as a chapter in Von Foerster, Heinz, Understanding Understanding: Essays on Cybernetics and Cognition, Springer, New York, 2003. This is the paper in which Von Foerster introduces the eigenform concept, the stable pattern that emerges when a recursive process converges and that grounds the architectural endogeneity argument in this brief. https://cepa.info/fulltexts/1270.pdf

6. Alicke, Knut and Tacy Foster, with Vera Trautwein, “The Way Forward: McKinsey Global Supply Chain Leader Survey,” McKinsey & Company, October 14, 2024. The article reports findings from the fifth annual McKinsey Global Supply Chain Leader Survey. Nine in ten respondents reported encountering supply chain challenges in 2024. On average, companies take two weeks to plan and execute a response to a disruption. https://www.mckinsey.com/capabilities/operations/our-insights/supply-chain-risk-survey-2024

7. Teece, David J., Gary Pisano, and Amy Shuen, “Dynamic Capabilities and Strategic Management,” Strategic Management Journal, Vol. 18, No. 7, 1997, pp. 509-533. See also Teece, David J., “Explicating Dynamic Capabilities: The Nature and Microfoundations of (Sustainable) Enterprise Performance,” Strategic Management Journal, Vol. 28, No. 13, 2007, pp. 1319-1350.

8. MacKenzie, Donald, “Is Economics Performative? Option Theory and the Construction of Derivatives Markets,” Journal of the History of Economic Thought, Vol. 28, No. 1, 2006, pp. 29-55.

9. DeAngelis, Stephen F., Polycrisis⁴ Research Agenda, April 2026. Simulation methodology, ablation results, robustness analysis, and full source code available upon request. See also the accompanying simulation results document. Working paper, available from the author upon request at www.deangelisreview.com.

10. The Banach fixed-point theorem establishes that a contraction mapping on a complete metric space has a unique fixed point, and that iterative application of the mapping converges to that point. See: Wikipedia, “Banach fixed-point theorem.” https://en.wikipedia.org/wiki/Banach_fixed-point_theorem

11. Rosen, Robert, Anticipatory Systems: Philosophical, Mathematical, and Methodological Foundations, Pergamon Press, 1985. Second edition, Springer, 2012.

12. Ashby, W. Ross, An Introduction to Cybernetics, Chapman & Hall, London, 1956. The Law of Requisite Variety states that a controller must have at least as many available responses as the system it controls has possible disturbances.

13. Cecere, Lora, “Why Is Sales and Operations Planning So Hard?”, Forbes, January 21, 2015. The two-in-five effectiveness figure is from this 2015 research. https://www.forbes.com/sites/loracecere/2015/01/21/why-is-sales-and-operations-plannning-so-hard/

14. Apotheker, Jessica, Vinciane Beauchene, Nicolas de Bellefonds, Patrick Forth, Marc Roman Franke, Michael Grebe, Nina Kataeva, Santeri Kirvelä, Djon Kleine, Romain de Laubier, Vladimir Lukic, Amanda Luther, Mary Martin, Jeff Walters, and Christoph Schweizer, “The Widening AI Value Gap: Build for the Future 2025,” Boston Consulting Group, September 2025. The report finds that 60 percent of companies are reaping hardly any material value from AI, reporting minimal revenue and cost gains despite substantial investment, while 5 percent qualify as “future-built” AI leaders. Based on BCG’s Build for the Future 2025 Global Study (n = 1,250). Web page: https://www.bcg.com/publications/2025/are-you-generating-value-from-ai-the-widening-gap. Full PDF: https://media-publications.bcg.com/The-Widening-AI-Value-Gap-Sept-2025.pdf

15. Singla, Alex, Alexander Sukharevsky, Lareina Yee, and Michael Chui, with Bryce Hall and Tara Balakrishnan, “The State of AI: Global Survey 2025,” QuantumBlack, AI by McKinsey, McKinsey & Company, November 5, 2025. Nearly two-thirds of respondents say their organizations have not yet begun scaling AI across the enterprise, with the majority remaining in experimenting or piloting stages. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

16. Leigh, Nathan W.C., Nicholas C. Stone, Aaron M. Geller, Michael M. Shara, Harsha Muddu, Diana Solano-Oropeza, and Yancey Thomas, “The chaotic four-body problem in Newtonian gravity, I. Identical point-particles,” Monthly Notices of the Royal Astronomical Society, Vol. 463, No. 3, December 2016, pp. 3311-3325. DOI: 10.1093/mnras/stw2178

17. The Budget Lab at Yale, “State of U.S. Tariffs: April 8, 2026,” April 8, 2026. The Budget Lab’s baseline analysis assumes the Section 122 tariffs expire in 150 days from enactment, placing expiration in late July 2026. https://budgetlab.yale.edu/research/state-us-tariffs-april-8-2026

The April 8, 2026, ceasefire between the United States and Iran has been widely interpreted as a de-escalation. It is not. It is a false resolution, a phase in a non-linear compounding system in which structural risk continues to rise while organizational attention drops. This brief examines why the pause is the most dangerous condition in the current crisis environment and what leaders should do about it. More specifically, it argues that the current window demands a new enterprise intelligence and AI orchestration layer, one that can perceive and act across the non-linear timescales that the crisis has exposed.

The Polycrisis³ framework describes three interacting dimensions. The first is the base polycrisis of geopolitical, economic, and energetic disruptions. The second is the technological exponent of artificial intelligence and quantum computing, which operates on its own timeline and did not observe the ceasefire. The third is the organizational response function, the dimension most directly within leadership’s control and most degraded by the false resolution dynamic.

‍

This brief introduces three actionable concepts.

‍

First, the false resolution pattern. In non-linear systems, periods of apparent calm are not interruptions of the crisis. They are phases of it. The false resolution widens the gap between actual risk and perceived risk to its maximum, precisely when organizations are most likely to stand down their sensing architectures.

‍

Second, the self-liquidating proof point. The false resolution window is the optimal entry for enterprise resilience transformation. One high-value process, one instrumented intelligence layer, one fiscal year to demonstrate measurable return. The return funds the next process, making the transformation self-financing. The current window has a known expiration date tied to the Section 122 bridge tariff expiration in July 2026.

‍

Third, the shift from linear planning to non-linear architecture. The dominant enterprise planning model runs a sequential monthly cascade that is structurally incapable of representing non-linear dynamics. Only two in five companies consider their S&OP process effective. The alternative, non-linear optimization orchestrated by an intelligent agent-based architecture, absorbs the planning complexity that the sequential cascade cannot represent and replaces it with concurrent, continuous planning that operates across all timescales simultaneously. An organization with a monthly planning cycle and a two-week disruption window has exactly one planning cycle to respond. An organization with a continuous sensing architecture has every hour of every day of that window.

‍

The organizations that will define the next decade are not the ones that responded most heroically to the acute crisis. They are the ones that used the false resolution most effectively. The architecture exists. The technology is available. The window is open. What remains is the decision to act before it closes.

‍

The April 2026 ceasefire between the United States and Iran did not resolve the compounding crises threatening global supply chains and competitive positions. It paused them. That pause is the most dangerous phase in a non-linear system, a false resolution in which organizational attention drops while structural risk continues to rise. The technological exponent of the crisis, the relentless advance of artificial intelligence and quantum computing, did not observe the ceasefire and never will. My concern is that the leaders who stand down their sensing architectures and resilience postures during this pause will enter the next acute phase with maximum surprise and minimum preparation. The window for building the architecture that prevents that outcome is open now, and it has a known expiration date.

‍

The Moment That Felt Like Relief

On the evening of April 8, 2026, less than two hours before a United States military deadline, Pakistan brokered a tentative ceasefire between Washington and Tehran. The Strait of Hormuz, functionally impaired since late February, would reopen. Brent crude fell approximately fourteen percent in a single session, dropping below one hundred dollars per barrel after having surged to one hundred seventeen. Global markets exhaled. Boards stood down their emergency postures. Risk teams that had been running scenario models around the clock shifted back to quarterly cadences. The crisis, it appeared, had passed.¹

‍

It had not passed. It had paused.

‍

Iran’s Supreme National Security Council accepted the ceasefire terms while announcing that its “hands remain on the trigger.” The agreement covered the Strait but explicitly excluded Lebanon, where Israeli operations continued. Russia and China had vetoed a United Nations Security Council resolution on Hormuz safety just twenty-four hours earlier.² The United States simultaneously threatened fifty-percent tariffs on nations supplying Iran with weapons, extending the economic dimension of the conflict into new sovereign relationships.³ The International Monetary Fund had already warned that the war had produced the worst-ever disruption in global energy supplies and that poor nations with no reserves would absorb the hardest impact regardless of how quickly the fighting stopped.⁴

‍

The ceasefire did not resolve the crisis. It paused it, and the pause is the most dangerous phase. I want to be direct about what that means. For leaders who have built enterprise resilience architectures, this pause is the best window they will get to extend and harden what they built. For the majority who have not, it is a trap that will close before they recognize it.

‍

The Non-Linear Structure of Compounding Crises

In the first brief in this series, I introduced the concept of Polycrisis² to describe the condition in which traditional geopolitical, economic, and energetic crises function as a base, while exponential technological disruptions function as an exponent that changes the nature of the problem rather than merely its magnitude. Artificial intelligence is restructuring labor markets and competitive dynamics. Quantum computing is advancing toward the capacity to break current cryptographic standards. Together they do not simply compound the polycrisis. They change what the problem is.⁵

‍

The mathematical notation is deliberately heuristic. It is not a formula. It is a description of a structural relationship that linear thinking consistently fails to perceive. When a base condition and an exponent compound simultaneously, the output accelerates in ways that no extrapolation from prior data points can predict. Remove one crisis and the system does not return to its pre-crisis state. The interactions have changed it.

‍

This week produced a third dimension of that structure that deserves its own examination.

‍

In a non-linear system, periods of apparent calm are not interruptions of the crisis. They are phases of it.

‍

This is the insight that linear mental models most reliably miss. When conditions deteriorate sharply, organizations activate. Sensing layers go live. Boards engage. The acute crisis produces a temporary improvement in organizational intelligence, not because the environment is better, but because attention is elevated. When conditions apparently improve, the opposite occurs. Attention migrates back toward efficiency. Sensing layers return to baseline. The structural fragilities that were present before the crisis became less visible than they were before the acute phase began, even as the crisis deepened them.

‍

This is a structural feature of how complex adaptive systems respond to perturbation, and I want to name it plainly because most crisis-management frameworks do not. The recovery period concentrates organizational attention on the disruption that just occurred, making that specific disruption the least likely to recur in the same form. The next acute phase will arrive through a different vector, at a moment when the organization has retooled for the environment that existed during the false resolution. In twenty-five years of working with organizations in volatile environments, I have watched this cycle repeat in supply chains, in financial institutions, and in government agencies. The pattern is not a failure of intelligence. It is a failure of architecture.

‍

The false resolution does not merely fail to resolve the crisis. It actively prepares the conditions for a worse one.

‍

What Non-Linearity Actually Means in Practice

A linear model of crisis tells one story. The Strait of Hormuz was impaired, oil prices rose, a ceasefire was reached, oil prices fell, conditions normalized. The slope of the disruption was reversed. We are on the way back to where we were.

‍

A non-linear model tells a structurally different story, and the distinction matters operationally.

In a non-linear compounding system, the state of the system after a perturbation is never identical to the state before it. The perturbation changes the relationships between components, not merely the values of variables. After the 2008 financial crisis, the apparent recovery of 2010 and 2011 occurred within a structurally altered system. The fragilities that manifested in the European sovereign debt crisis were not the same fragilities that had produced the 2008 event. They were new fragilities, created in part by the organizational responses to the old ones.

‍

I saw the same non-linear dynamic in the supply chain disruptions that followed the COVID-19 pandemic. The apparent recovery of 2022 occurred within a supply chain architecture that had been structurally altered by the disruption. The concentration of more than ninety percent of the world’s advanced logic chip fabrication in a single geographic cluster around TSMC in Taiwan meant that every company dependent on leading-edge semiconductors had traded one supply chain risk for a different and arguably more severe one. Pharmaceutical manufacturers that consolidated active pharmaceutical ingredient sourcing with single suppliers in India replaced a diversified risk profile with a sole-source dependency that was invisible until those suppliers experienced their own disruptions. Organizations that built large safety stocks to buffer against future shortages tied up working capital at precisely the moment that rising interest rates made that capital expensive, creating a new financial single point of failure. The disruptions of 2023 and 2024 arrived through the new vulnerabilities that the recovery had inadvertently created.

The current situation has a similar structure, but with a critical amplification. The Section 122 bridge tariffs, which created an up-to-fifteen-percent global levy after the Supreme Court invalidated the prior tariff regime, are set to expire in July 2026. The Yale Budget Lab estimates the current tariff regime will raise approximately $1.3 trillion over the 2026 to 2035 period, with long-run global GDP slightly lower across all scenarios. These are structural alterations to the trade architecture on which global supply chains depend.

‍

These are not temporary perturbations.^6,7 Meanwhile, refining capacity damage in the Middle East will constrain refined product supply for months even with the Strait fully reopened. J.P. Morgan Research had already placed the probability of a US and global recession in 2026 at thirty-five percent before the Iran conflict escalated.⁸

‍

The False Resolution Pattern in Complex Systems

I want to be precise about what makes the false resolution phase structurally distinct from both the acute crisis phase and genuine stabilization, because the distinction carries real operational consequences.

‍

In an acute crisis phase, the sensing function of an organization is forcibly activated. Information flows that were blocked by organizational silos become unblocked because the cost of silos suddenly exceeds their benefit. The crisis performs a crude form of the sensing function that a mature Sense, Think, Act & Learn architecture would perform continuously, at enormous cost and with high error rates, but it does activate. In a genuine stabilization phase, the conditions that produced the crisis have been structurally addressed and the sensing layer remains active because it has been institutionalized rather than activated by emergency. The false resolution is structurally different from both. The external signals suggest improvement, which triggers the deactivation of crisis-mode attention, while the underlying conditions remain unaddressed. This is the most dangerous phase in a non-linear compounding system, not because the external environment is at its worst, but because the gap between the actual risk level and the organization’s perceived risk level is at its maximum.

‍

Diane Vaughan coined the term “normalization of deviance” in her analysis of the Challenger disaster, describing how incremental acceptance of technical anomalies becomes institutionalized as normal when prolonged periods without catastrophic failure produce complacency.⁹ Karl Weick, separately, documented the collapse of sensemaking in organizations under extreme conditions, the process by which the mental models that allow people to act coherently disintegrate under stress.¹⁰

‍

One of my personal heroes and a great explainer of complex issues, Richard Feynman, understood something that Vaughan and Weick documented from the outside. At a televised hearing on the Challenger disaster in February 1986, he dropped a piece of the O-ring rubber into a glass of ice water. When he pulled it out, the rubber stayed stiff. In a room producing thousands of pages of testimony, Feynman made the failure visible and comprehensible in a single gesture. Making complex structural failure legible to the people who must act on it, without jargon and without equivocation, is a form of leadership. Feynman’s example represents the standard this brief series aspires to meet.

‍

The pattern holds across complex adaptive systems under non-linear risk. The false resolution is the period during which normalization of deviance accelerates because the acute crisis has temporarily made deviance visible and the recovery has made it invisible again. The Columbia disaster demonstrated that NASA had not learned from Challenger the lesson about false resolution. The organization had instead normalized a new set of deviations during the recovery period.

‍

The Exponent Does Not Observe Ceasefires

The feature of Polycrisis² that makes the false resolution particularly dangerous right now is the independence of the exponent from the base condition.

‍

In a linear model, when the base condition de-escalates, the overall risk level falls proportionally. In the squared model, the exponent operates on its own timeline, determined by technological development cycles rather than geopolitical negotiation cycles. A two-week ceasefire in the Middle East has no effect on the pace at which AI is restructuring competitive dynamics or on the quantum computing timelines that determine when current cryptographic standards become vulnerable.

‍

The National Institute of Standards and Technology published its post-quantum cryptography standards in 2024.¹¹ The Cybersecurity and Infrastructure Security Agency issued federal procurement guidance requiring quantum-resistant products in early 2026. Google has deployed hybrid post-quantum key exchange in its services and has committed to completing its full migration by 2029. These events establish a technical timeline that is not negotiable and not paused by ceasefires. Peer-reviewed research estimates that large enterprises will require twelve to fifteen years for complete cryptographic migration. If quantum capability arrives by 2030, organizations that use the false resolution as an opportunity to defer their migration programs are accepting a structural vulnerability that will not be visible until it is catastrophic. This is not only a CISO concern. It is a CSO concern, because the encrypted data most valuable to harvest-now-decrypt-later adversaries include supplier contracts, logistics routing, and multi-tier sourcing architectures, the very data that defines supply chain competitive advantage.

‍

The same temporal independence applies to artificial intelligence. BCG research found that the top five percent of AI-mature companies are achieving 1.7 times the revenue growth and 3.6 times the total shareholder return of lagging firms.¹² That gap reflects a competitive divergence that compounds every quarter regardless of what is happening in the Strait of Hormuz.

‍

The exponent has not paused. The artificial intelligence restructuring of labor markets continued through every day of the conflict. The harvest-now-decrypt-later campaigns that quantum-aware adversaries are running against encrypted enterprise data did not observe the ceasefire. The epistemic contamination of AI-generated intelligence, deepfakes, synthetic market signals, and AI-generated misinformation that compromises the reliability of the data on which organizations make decisions, did not stop because oil prices fell.

‍

The World Economic Forum has warned of a quantum divide, a structural gap between organizations that have begun their post-quantum migration and those that have not.¹³ I am convinced the same divide is already visible in AI maturity, and I will state this plainly. The false resolution is the period when this divide accelerates fastest, not because leading organizations speed up during the pause, but because lagging organizations slow down. BCG’s 2025 research confirms that sixty percent of companies are reaping hardly any material value from their AI investments, with most remaining in the experimenting and piloting stages that McKinsey’s 2025 State of AI survey documents, running proofs of concept and isolated experiments rather than embedding AI into their operating architecture.^14,15 During periods of apparent stability, organizations that have not yet made the architectural commitment to AI tend to redirect investment toward efficiency gains and cost reduction rather than the structural transformation that builds resilience and competitive sensing. That incremental approach feels rational in the moment. It is the mechanism through which the competitive gap compounds, because the five percent of firms that have made the architectural shift are reinvesting their AI-driven returns into stronger capabilities, planning to spend up to sixty-four percent more of their IT budgets on AI and pulling further ahead every quarter.¹⁴

‍

The base is still rising. The exponent is still accelerating. The product of the two is not lower because of the ceasefire. It is higher, because the two-week pause creates the organizational conditions for a third disruption that will be less anticipated than either of the first two.

‍

What the Non-Linear Geometry Looks Like

Consider a curved surface that rises steeply, then flattens briefly, then continues rising. A traveler on that surface, looking only at the immediate terrain, sees the flattening as evidence that the ascent has ended. The flattening is real. But the overall trajectory of the surface has not changed. The traveler who stops climbing during the flat section, and who uses the respite to remove equipment needed for the next ascent, will be less prepared for the steeper section ahead than if the acute phase had continued without interruption.

What I am observing has at least three interactive dimensions. The first is the base polycrisis. The second is the technological exponent. The third is the organizational response function, the combination of system perturbation timescales and organizational response capacity, which determines how much of the compounding disruption an enterprise actually absorbs. The third dimension is the one most directly within leadership’s control, and it is the one most affected by the false resolution dynamic.

‍

The organizations that treat the ceasefire as an opportunity to extend and stress-test their sensing architectures will exit this false resolution phase better positioned than they entered it. The organizations that treat the ceasefire as permission to stand down will exit this phase more exposed. This is not a symmetric choice. The cost of maintaining elevated sensing during a false resolution is operational and recoverable. The cost of being under-prepared at the onset of the next acute phase is strategic and compounding.

‍

The Geometry of the Self-Liquidating Response

In the Polycrisis² brief, I introduced the concept of a self-liquidating proof point as the practical entry to enterprise resilience transformation. The model is straightforward. One high-value process, one instrumented intelligence layer, one fiscal year to demonstrate measurable return. Use that return to fund the next process. Compound the transformation from demonstrated value rather than institutional faith.

‍

The false resolution phase is the optimal window for this entry. During an acute crisis, organizations are in emergency mode. Budget allocation is defensive. Leadership attention is consumed by the immediate disruption. The conditions for disciplined proof-point selection, instrumentation, and measurement are simply not present. During a false resolution, the emergency conditions have eased sufficiently to create space for deliberate architectural design. The organizational memory of the crisis is recent enough that the argument for investment in sensing and resilience architecture does not require extensive persuasion.

‍

This is the window. The proof-point model works precisely because it converts the strategic argument into a financial proposition that is demonstrable within a budget cycle.

The entry point is specific. Identify one high-value business process where an enterprise intelligence layer can demonstrate a measurable financial return within a single fiscal year. In consumer products, demand sensing is a natural starting point. An intelligence layer that continuously reconciles sell-through data, promotional calendars, and external signals produces forecasting accuracy improvements that are measurable within a quarter and financially material within a year. In industrial supply chains, supplier risk monitoring offers a similar profile. Continuous ingestion of geopolitical, financial health, and logistics signals against a multi-tier supplier map produces early warnings that avert disruptions whose cost can be precisely calculated after the fact. Either process, executed with the right architecture, generates a return that is visible on a P&L before the next budget cycle opens.

‍

Let that return fund the next process, and the next. The transformation becomes self-financing. The window created by this specific false resolution has a known expiration date. The Section 122 bridge tariffs expire in July 2026. The two-week ceasefire expires by late April. That is exactly the interval the self-liquidating proof-point model is designed for.

‍

Building Resilience and Optimizing Value Chains Across Timescales

Polycrisis³ operates across at least three distinct timescales simultaneously, and the failure to perceive all three at once is the structural weakness that the false resolution exploits.

‍

The near-term timescale spans days to weeks. This is the ceasefire, the price of crude, the board’s emergency posture. Every organization can operate here. When the Strait of Hormuz closes, the near-term response activates. When it reopens, it deactivates. This is the only timescale most organizations are equipped to perceive.

‍

The medium-term timescale spans quarters to years. This is the tariff regime expiration in July 2026, the AI maturity gap that compounds quarterly, the false resolution window during which architectural decisions will determine structural position for the next three to five years. The medium-term requires organizational attention sustained by design rather than triggered by crisis.

The long-term timescale spans years to decades. This is the quantum cryptography migration that will require twelve to fifteen years for large enterprises, and the AI competitive divide that will separate organizations into those that built continuous intelligence architectures and those that did not. The long-term timescale does not produce acute signals. It does not appear on quarterly board agendas. It is the timescale on which the most consequential changes occur, and the one the false resolution makes hardest to perceive.

‍

The idea that organizations must simultaneously exploit existing capabilities and explore new ones across different time horizons has deep roots in organizational theory, what David Teece and others have called dynamic capabilities.¹⁶

‍

I call the capacity to operate across all three timescales at once Timescale Coherence. The Sense, Think, Act & Learn architecture is designed for exactly this. It maintains continuous perception across all three timescales, which means it perceives the false resolution not as a period of calm but as a medium-term architectural window within a long-term transformation program.

Timescale Coherence is the application of that insight to the specific conditions of Polycrisis³, and I think it is more urgent now than the academic literature has recognized, because the timescale gap is widening faster than organizational design is adapting to it.

‍

An operational requirement of Timescale Coherence is the systemic ability to optimize and simulate across multiple non-linear timescales concurrently. This means the enterprise can evaluate decisions against the near-term financial objective, the medium-term architectural position, and the long-term competitive trajectory at the same time, rather than treating each timescale as a separate planning exercise. The technology to do this exists. It requires an intelligence layer that continuously ingests signals across all three horizons and an optimization engine that can represent the non-linear interactions between them.

‍

The self-liquidating proof point is itself an exercise in Timescale Coherence. It operates on a near-term financial cycle, one fiscal year, while building toward a medium-term architecture that positions the organization for the long-term competitive divide. The proof point works as a mechanism for translating long-term structural awareness into near-term financial action, not as an emergency response.

‍

From Linear Planning to Non-Linear Architecture

More enterprise planning processes have failed in the last three years than in the previous twenty, and the failures share a common structure. The planning process itself is linear. It was designed for a linear world. It is now operating inside a non-linear environment, and it cannot perceive the dynamics that are killing it. McKinsey’s 2024 Global Supply Chain Leader Survey found that ninety percent of supply chain leaders encountered significant disruptions that year, and it still takes companies an average of two weeks to plan and execute a response to a disruption, far longer than the weekly cadence of sales and operations execution. Their 2025 follow-up found that the share of companies planning major investments in digital supply chain systems fell from forty-seven percent to twenty-five percent in a single year, and only nineteen percent have deployed AI tools at scale.¹⁷

‍

Consider how the dominant enterprise planning model works. A demand forecast is produced. That forecast feeds a supply plan. The supply plan feeds financial reconciliation. Financial reconciliation feeds an executive approval cycle. Each step depends on the completed output of the prior step. The entire cascade runs on a monthly cadence, occasionally accelerated to biweekly during acute disruptions. Research from Supply Chain Insights confirms what practitioners already know. Only two in five companies consider their S&OP process effective, and only eleven percent successfully link planning to execution.¹⁸

^‍

This is a sequential batch process designed for a world where disruptions were periodic, predictable, and separated by intervals of relative stability. In that world, a monthly cycle was adequate because the system state did not change materially between planning runs. In a Polycrisis³ environment where the base, the exponent, and the organizational response function are all moving simultaneously across multiple timescales, a monthly batch process is structurally blind to the dynamics it needs to perceive. The architecture of the process itself cannot represent the environment it is supposed to perceive.

‍

The structural mismatch runs deeper than speed. Linear planning assumes the system returns to equilibrium after a perturbation. Non-linear systems do not. The perturbation changes the relationships between variables, not just their values. When the Strait of Hormuz closed and reopened, the planning process recalibrated from the prior period’s actuals. But those actuals described a system state that no longer existed. Shipping routes had been renegotiated. Refining capacity had been damaged. Tariff exposures had shifted. Working capital structures had been altered by emergency drawdowns. The planning process treated these as updated inputs to the same model. They were not updated inputs. They were evidence that the model itself had changed.

‍

This is why the false resolution is so dangerous from a planning perspective. The sequential cascade interprets the apparent return to prior conditions as evidence that the system has stabilized. The demand forecast reverts toward historical baselines. The supply plan relaxes its constraint assumptions. The financial reconciliation projects recovery. Every one of those adjustments is wrong, not because the data is inaccurate, but because the planning architecture is structurally incapable of representing the non-linear transformation that the disruption produced.

The AI and Non-Linear Alternative

The alternative is a structurally different process designed for a structurally different environment. Non-linear optimization, orchestrated by an intelligent agent-based architecture, replaces the sequential cascade with concurrent, continuous planning.¹⁹ Instead of a monthly cycle where each function waits for the prior function’s output, an agent-based architecture orchestrates sensing, advanced analysis, and response simultaneously. Demand signals, supply constraints, geopolitical risk indicators, and financial parameters are processed concurrently rather than sequentially through non-linear optimization rather than the linear models the cascade assumes. The architecture does not wait for the demand forecast to finish before assessing supply constraints. It assesses both at once, continuously.

‍

When the Strait of Hormuz closes, the impact on refined product supply, on shipping routes, on tariff exposure, on working capital requirements, and on competitive positioning relative to AI-mature competitors is assessed simultaneously, not in a four-week cascade. This is what non-linear architecture means in operational terms.

‍

I want to connect this directly to the Polycrisis³ framework, because the planning architecture is not a side issue. It is the mechanism through which the organizational response function operates. The third dimension of Polycrisis³ is only as fast as the planning architecture that drives it. An organization with a monthly planning cycle and a two-week disruption window has exactly one planning cycle to respond. One pass through the sequential cascade. An organization with a continuous, agent-based sensing and response architecture has every hour of every day of that window.

‍

The difference is structural, not incremental. It is the difference between perceiving the false resolution as a pause and perceiving it as a window. Organizations that can make this transition will not experience the false resolution the way their competitors do. They will experience it as the highest-value interval in the crisis cycle, the period when the competitive gap widens fastest because most organizations have stood down the very sensing functions that the window demands.

‍

That is the architectural argument for non-linear planning. The planning process is not a back-office function that supports strategy. It is the organizational nervous system. In a Polycrisis³ environment, the quality of that nervous system determines whether an organization perceives the pause as an ending or as a window.

‍

The Imperative

The Polycrisis² condition does not resolve. It evolves. Each acute phase is followed by a false resolution that reorganizes the underlying structure of the compounding crises. Each false resolution creates conditions that make the next acute phase worse than the one that preceded it, for the organizations that mistake the pause for the end.

‍

The organizations that will define the next decade are not the ones that responded most effectively to the acute crisis of February through April 2026. They are the ones that used the false resolution most effectively. I have watched this distinction play out again. The organizations that are genuinely resilient are never the ones that responded most heroically to the worst moments. They are the ones that maintained their sensing architectures through the quiet periods, when doing so was the hardest to justify to a board and the easiest to defer to the next budget cycle. That discipline must be built into the architecture itself because human attention and institutional rhythm are both wired for the near-term. A continuous intelligence capability performs at its best precisely when the external environment appears least urgent, because that is when the gap between organizational perception and structural reality is at its widest. That is when the architecture earns its cost.

‍

Polycrisis³ is the product of compounding crises, exponential disruption, and an organizational response function that is weakest at exactly the moment it is most needed.

‍

The architecture for resilience exists. The technology is available. The window is open. What remains is the decision to act before it closes.

The ceasefire did not end the crisis. It created the most dangerous phase of it, the pause in which most organizations will stand down while the structural risk continues to compound. The organizations that read this pause as the end will be the least prepared for what comes next. The architecture to perceive it exists. The window to build it is open. The only question left is whether your organization will still be climbing when the terrain steepens again.

‍

Stephen F DeAngelis

Princeton, NJ

April 2026

‍

Polycrisis²™, Polycrisis³™, and Timescale Coherence™ are trademarks of Stephen F. DeAngelis. Sense, Think, Act & Learn™ is a trademark of Enterra Solutions.

‍

Stephen F. DeAngelis is the founder, president, and CEO of Enterra Solutions and Massive Dynamics, two companies that apply artificial intelligence and advanced mathematics to complex enterprise challenges. His career spans international relations, national security, and commercial technology. He has served in visiting research affiliations with Princeton University, the Oak Ridge National Laboratory, the Software Engineering Institute at Carnegie Mellon University, and the MIT Computer Science and Artificial Intelligence Laboratory. He is a founding member of the Forbes Technology Council. DeAngelis holds patents in autonomous decision science and has been recognized by Forbes as a Top Influencer in Big Data and by Esquire magazine as the ‘Innovator’ in its Best and Brightest issue.

‍

About the Stephen DeAngelis Explainer Brief Series

The Stephen DeAngelis Explainer Brief series applies critical reasoning to the complex issues facing society today. In an era of compounding uncertainty and deepening division, the series aims to build understanding and community by making consequential topics accessible through rigorous analysis, current evidence, and honest assessment. Each installment is written in the belief that clarity of thought is itself a form of leadership.

‍

Readers who wish to explore the proof-point entry model for their organization are welcome to reach out at www.deangelisreview.com.

‍

Notes

1. Reuters, “US and Iran agree to two-week ceasefire brokered by Pakistan,” April 8, 2026. https://www.reuters.com/world/asia-pacific/trump-agrees-two-week-ceasefire-iran-says-safe-passage-through-hormuz-possible-2026-04-08/

2. United Nations Security Council, veto of Hormuz safety resolution by Russia and China, April 7, 2026. https://news.un.org/en/story/2026/04/1167257

3. CNBC, “Trump threatens tariffs of 50% on nations supplying weapons to Iran,” April 8, 2026. https://www.cnbc.com/2026/04/08/trump-threatens-tariffs-countries-supplying-weapons-iran-ceasefire.html

4. International Monetary Fund, Managing Director Kristalina Georgieva, “Cushioning the Middle East War Shock,” April 9, 2026. https://www.imf.org/en/news/articles/2026/04/09/sp040926-spring-meetings-2026-curtain-raiser

5. DeAngelis, Stephen F., “Polycrisis²: When Compounding Crises Meet Exponential Technology,” DeAngelis Review, March 2026. www.deangelisreview.com

6. The Yale Budget Lab, “State of US Tariffs: April 2, 2026.” https://budgetlab.yale.edu/research/state-us-tariffs-april-2-2026

7. The Chronicle-Journal, “Global Trade Chaos and New Tariffs Fuel Market Volatility in 2026,” March 18, 2026.

8. J.P. Morgan Global Research, “2026 Market Outlook,” December 2025. https://www.jpmorgan.com/insights/global-research/outlook/market-outlook

9. Vaughan, Diane, The Challenger Launch Decision: Risky Technology, Culture, and Deviance at NASA, University of Chicago Press, 1996.

10. Weick, Karl E., “The Collapse of Sensemaking in Organizations: The Mann Gulch Disaster,” Administrative Science Quarterly, Vol. 38, No. 4, 1993.

11. National Institute of Standards and Technology, Post-Quantum Cryptography Standards, 2024. https://www.nist.gov/pqc

12. BCG, “Are You Generating Value from AI? The Widening Gap,” September 2025. https://www.bcg.com/publications/2025/are-you-generating-value-from-ai-the-widening-gap

13. World Economic Forum, “Why quantum security is a question leaders cannot ignore,” February 2026. https://www.weforum.org/stories/2026/02/quantum-security-question-leaders-cannot-ignore

14. BCG, “The Widening AI Value Gap” (see endnote 12). Research based on BCG Build for the Future 2025 Global Study (n = 1,250). Only five percent of companies are achieving AI value at scale. Sixty percent report minimal revenue and cost gains despite substantial investment. Future-built companies plan to dedicate up to sixty-four percent more of their IT budgets on AI than lagging firms. See also BCG press release, September 30, 2025, https://www.bcg.com/press/30september2025-ai-leaders-outpace-laggards-revenue-growth-cost-savings.

15. McKinsey & Company, “The State of AI: Global Survey,” 2025. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai. Eighty-eight percent of organizations use AI in at least one business function, but two-thirds remain in pilot mode. Only twenty-three percent have scaled AI agent deployments.

16. Teece, David J., Gary Pisano, and Amy Shuen, “Dynamic Capabilities and Strategic Management,” Strategic Management Journal, Vol. 18, No. 7, 1997.

17. McKinsey & Company, “Supply Chain Risk Pulse 2025: Tariffs Reshuffle Global Trade Priorities,” 2025. https://www.mckinsey.com/capabilities/operations/our-insights/supply-chain-risk-survey. See also McKinsey & Company, “Supply Chains: Still Vulnerable,” Global Supply Chain Leader Survey, October 2024. https://www.mckinsey.com/capabilities/operations/our-insights/supply-chain-risk-survey-2024. Ninety percent of supply chain leaders reported significant disruptions in 2024. Average response time to disruptions was two weeks. Investment in digital supply chain systems fell from forty-seven percent to twenty-five percent. Only nineteen percent have deployed AI tools at scale.

18. Cecere, Lora, “Why Is Sales and Operations Planning So Hard?“, Forbes, January 21, 2015. See also Cecere, Lora, Supply Chain Metrics That Matter, Supply Chain Insights / Wiley, 2015. Statistics confirmed in subsequent Supply Chain Insights research through 2024.

19. IBM, “AI Agents in Supply Chain,” IBM Think, January 30, 2026. https://www.ibm.com/think/topics/ai-agents-supply-chain. See also IBM Institute for Business Value, “Scaling Supply Chain Resilience: Agentic AI for Autonomous Operations,” April 2025. See also Porsche Consulting, “Mastering Supply Chain Complexity with AI,” February 2026.

‍

Brent crude has surged past $106 a barrel. Urea prices have climbed approximately 50%, with ammonia, phosphate, and potash rising sharply. The head of the International Energy Agency has called it “the greatest global energy security threat in history,” warning that more oil has been lost than in both 1970s crises combined and that fertilizer, chemicals, and helium supply chains face simultaneous disruption. Central banks from Washington to Frankfurt are trapped between rising inflation and weakening growth. And all of this is unfolding against the backdrop of an expanding tariff regime, with Section 301 investigations targeting 16 economies for industrial overcapacity and a further round covering approximately 60 trading partners for forced labor practices.

‍

But the polycrisis, as commonly described, captures only the base condition. There is an exponent.

‍

The Exponent, AI, and Quantum Computing

‍

I want to introduce a concept I am calling Polycrisis², the polycrisis squared. The notation is deliberately heuristic rather than algebraic, a shorthand for the non-linear dynamics that defy simple quantification. The base is what we see in the headlines, specifically the cascading, mutually reinforcing geopolitical and economic shocks that have defined the past several years. The exponent is the consequential change being driven simultaneously by two technologies, artificial intelligence and quantum computing, which are compounding it at an unprecedented rate.

‍

The scale of the AI transformation alone is staggering. Morgan Stanley Research estimates that approximately $3 trillion in AI-related infrastructure investment will flow through the global economy by 2028, with more than 80% of that spending still ahead. Corporations expect to double their AI expenditure in 2026 alone. Block, the payments company, cut nearly half its workforce (over 4,000 positions) citing AI’s ability to automate fraud detection, risk assessment, and customer support. Reuters reported that Meta was planning to eliminate approximately 16,000 positions (20% of its workforce) in cuts widely attributed to its accelerating AI infrastructure investment, which Meta has guided at $115 to $135 billion in capital expenditure for 2026 alone. An estimated 45,000 tech jobs have been eliminated globally in the first quarter of 2026, with roughly 20% attributed directly to AI replacement.

‍

These shifts are structural. The World Economic Forum’s Future of Jobs Report 2025 projects 170 million new roles created and ninety-two million displaced between 2025 and 2030 across all drivers of structural change, including, but not limited to AI. But within that net-positive headline lies an asymmetry that matters. Capital captures productivity gains immediately, while labor absorbs displacement on a two-to-four quarter delay. Layer that delay onto an energy-driven inflation surge and constrained monetary policy, and the compounding effect on household economics becomes severe.

‍

Quantum computing compounds the picture in a different way. Google shortened its post-quantum cryptography migration timeline to 2029, warning that “a cryptographically relevant quantum computer is not forever a decade away.” The company confirmed that “store now, decrypt later” attacks (in which adversaries harvest encrypted data today with the expectation of decrypting it once quantum systems mature) are already underway. Only 9% of organizations have any plan for transitioning to quantum-resistant encryption. The World Economic Forum has warned explicitly about a “quantum divide” in which wealthy nations and large corporations become quantum-safe while the rest of the world gets cut off from global trade and finance overnight.

‍

This is already playing out in real time across multiple domains simultaneously. AI-generated deepfakes depicting fabricated missile strikes on Tel Aviv, US bases in Riyadh, and port infrastructure in Bahrain have flooded social media since the Iran war began. The New York Times identified more than 110 unique deepfakes in a two-week period, many produced by state-linked influence networks and amplified across the Russian, Chinese, and Iranian information ecosystems. Corporate decision-makers monitoring the Strait of Hormuz crisis cannot distinguish real damage assessments from synthetic fabrications without investing in verification infrastructure that most organizations do not possess. Meanwhile, cybersecurity researchers have confirmed that harvest-now-decrypt-later campaigns are accelerating. Adversaries are quietly exfiltrating encrypted corporate data, trade secrets, and supplier agreements with the expectation of decrypting them once quantum capability matures. The data cannot be unharvested. The exposure is permanent and cumulative.

‍

This is the exponent. AI is restructuring labor markets, competitive dynamics, and the speed at which decisions must be made, all in the middle of an economic environment already destabilized by war, energy disruption, and trade fragmentation. Quantum computing is quietly undermining the cryptographic foundations on which global commerce and national security depend. Neither operates in isolation from the base polycrisis. Each amplifies it.

‍

Why the Squared Function Matters

‍

A simple polycrisis is, at least theoretically, manageable through coordination. Central banks can consult. Alliances can negotiate. Supply chains can reroute. The difficulty increases dramatically, but the tools are familiar.

‍

Polycrisis² is different because the exponent changes the very nature of the problem, beyond its magnitude.

‍

A tariff shock is manageable. A tariff shock during an energy crisis is harder. A tariff shock during an energy crisis while AI is displacing your workforce, rewriting your competitive landscape, and degrading the reliability of the information on which you make decisions, while the cryptographic infrastructure protecting your intellectual property faces an advancing existential threat, which is a qualitatively different problem.

‍

AI degrades the information environment in which leaders make decisions through deepfakes, AI-generated misinformation, algorithmic amplification of negative sentiment, and the steady erosion of the trusted analytical baseline on which organizational decision-making depends. Quantum computing, meanwhile, has already changed the behavior of adversaries who are harvesting data now against the eventuality of being able to break encryption. The compounding effect is non-linear, and each crisis amplifies the others in ways that cannot be decomposed into their individual parts.

‍

Here is how the squared function works in practice. A supply chain disruption caused by the Strait of Hormuz closure is a manageable crisis. Companies have playbooks for rerouting, repricing, and reallocating. But when the same company must simultaneously evaluate whether its rerouted supply chain communications are being harvested by adversaries exploiting quantum-vulnerable encryption, and whether the AI-generated market intelligence guiding its repricing decisions has been contaminated by synthetic misinformation, the problem changes in kind. It becomes one of epistemic integrity. The organization cannot trust the data on which it would normally base its response. It cannot trust the communications through which it coordinates that response. These are crises that degrade the very tools available to address them.

‍

This distinction, between crises that test an organization’s capacity and crises that compromise its ability to perceive and respond, is the analytical core of the Polycrisis² concept. It is what separates a difficult operating environment from one that is qualitatively different. And it is why the organizational response must be architectural rather than incremental. You cannot improvise your way through a disruption when the instruments of improvisation themselves are compromised.

‍

This framework has limits that deserve acknowledgment. The Sense, Think, Act & Learn architecture assumes that an organization possesses a minimum threshold of data infrastructure, institutional willingness to share date and insights across functions, and leadership continuity manage the buildout. Where any of these preconditions is absent, the architecture will have difficulty taking hold regardless of the quality of the technology. It also assumes that the sensing layer can be deployed faster than the environment degrades it, an assumption that a sufficiently rapid cascade of compound disruptions could defeat – we hope this is not the case. These are the boundary conditions. They do not invalidate the framework. They define the conditions under which it must be deployed with discipline, and the conditions under which an organization should focus.

‍

Leadership Matters — But Not the Way You Think

‍

The instinct, when facing compounding complexity, is to look for strong leaders. And leadership does matter, profoundly. The organizations that navigate Polycrisis² will be led by people who possess the vision to see what is changing, the judgment to distinguish signal from noise, and the courage to act before the data is complete.

‍

But we need to be precise about what we mean. Leadership in this context means something beyond individual heroism, beyond the mythology of the lone executive who sees what no one else can see and wills the organization through a crisis by sheer force of character. That narrative is compelling, and also dangerous, because it substitutes personal grit for systemic capability, and in a Polycrisis² environment, individual grit without a system for resiliency is simply a more admirable form of fragility.

‍

The organizations that relied on heroics during COVID-19 (where a handful of gifted operators improvised solutions on the fly) often discovered that those solutions could not scale, could not be replicated, and evaporated the moment those individuals moved on. The crisis was survived despite the absence of organizational capability. That works once. It does not work when the disruptions are continuous, overlapping, and compounding.

‍

Enlightened leadership in the age of Polycrisis² means building the systems that make the organization resilient independent of any single leader’s brilliance. It means recognizing that your job as a leader is to ensure that the organization can sense, interpret, respond, and learn from disruption structurally, whether you are in the room or not.

‍

The Enterprise Intelligence Layer, Sense, Think, Act & Learn

‍

The Polycrisis² condition described above is an environment that must be navigated, and navigating it requires a fundamentally different organizational architecture.

‍

If individual heroism is insufficient, what takes its place? An enterprise intelligence layer, a systemic capability embedded in the organization’s operating architecture, which allows it to dynamically Sense, Think, Act & Learn. The idea of iterative organizational response is not new. Boyd’s OODA loop, Deming’s Plan-Do-Check-Act, and the organizational learning tradition of Argyris and Schön all describe cycles of environmental sensing and adaptive response. But those frameworks were designed for human-speed decision cycles, such as a fighter pilot’s cockpit, a factory floor, or a management offsite. They assume that the sensing, interpretation, and response can be performed by individuals or small teams operating within a single domain.

What the Polycrisis² environment demands is fundamentally different, specifically the same business and analytic logic applied across dozens of interconnected domains simultaneously, at machine speed, with autonomous propagation of insight across functional boundaries, embedded in the organization’s data architecture, AI systems, and autonomous decision infrastructure. The distinction is architectural.

When an energy shock simultaneously disrupts shipping routes, triggers tariff recalculations, shifts demand patterns, and creates workforce reallocation pressure, all while the information environment is being contaminated by AI-generated misinformation, no human team running an OODA loop in a conference room can process the interactions fast enough. The system must do it, or it does not get done.

‍

Sense means building the instrumentation to detect disruption early and across the full dimensionality of the operating environment. Not just monitoring the variables you already know matter, but maintaining awareness across adjacent domains (geopolitical, technological, regulatory, competitive, environmental) where the next compound shock is likely to originate. Most organizations have point-monitoring systems. Very few have the integrated sensing architecture that a Polycrisis² environment demands.

‍

Think means applying analytical capability, increasingly AI-driven, to interpret what the sensing layer detects. This is where raw signals become actionable intelligence. It requires modeling cascading effects across interconnected systems, specifically how an energy disruption propagates through shipping costs into supplier economics, into competitive pricing, into demand signals, and into workforce planning. Organizations that can think across these dimensions in near-real-time will see the shape of compound crises before they fully materialize.

Act means translating insight into response at the speed the environment requires. In a Polycrisis² world, the traditional cycle of quarterly strategy review, annual planning, and committee-based decision-making is structurally too slow. The answer is pre-mapped response options, delegated decision authorities, and the organizational muscle memory to execute scenario-based responses without waiting for consensus that will never arrive in time.

‍

Learn is the most critical, and most neglected, capability. It is the mechanism by which the organization improves with each disruption rather than merely surviving it. Learning means the models become more accurate, the sensing architecture expands to include the signals that were missed, the response playbooks are updated with what actually worked, and the institutional knowledge base deepens structurally. An organization that senses, thinks, and acts but does not learn will make the same mistakes in different configurations. An organization that learns will find that each cycle of disruption leaves it measurably stronger than the last.

‍

These four capabilities apply with equal force to the quantum dimension of the exponent. The sensing challenge is to track the quantum threat timeline as an active risk already shaping adversary behavior. The thinking challenge is to model the cascading consequences of a cryptographic breach across the entire data estate, including which intellectual property, which supplier agreements, and which customer records have already been harvested by adversaries waiting for decryption capability. The imperative action is to begin the migration to post-quantum cryptography now, years before a cryptographically relevant quantum computer is announced. The National Institute of Standards and Technology has already published the standards. Enterprise migration timelines of twelve to fifteen years mean that organizations beginning today may not finish before the threat materializes. And the learning opportunity is to treat each phase of cryptographic migration as a chance to discover where encryption is actually embedded, where dependencies hide across the vendor ecosystem, and where the organization’s cryptographic architecture is fragile in ways that have nothing to do with quantum computing, and everything to do with decades of accumulated technical debt.

‍

What makes this more than a framework is the underlying architecture. A genuine enterprise system of intelligence shares knowledge across the organization through enriched ontologies, structured representations of domain knowledge that allow insights generated in one part of the business to be understood, contextualized, and applied in another. When a supply chain disruption is detected, the implications for pricing, promotion, production planning, and financial forecasting should propagate simultaneously and autonomously, through an autonomous decision science platform that connects sensing, reasoning, and action into a single coherent system. The goal is to make decisions at the speed of the market, not at the speed of the organization chart. Enterprises that achieve this operate at a structural clock speed that turns volatility into advantage.

‍

This is an engineering challenge, and a practical one. Building the Sense, Think, Act & Learn capability requires the integration of AI systems, data architecture, decision science, and organizational design into a coherent operating layer. The technology to do this exists today. The question is whether leadership will commit to the investment before the next compound disruption makes the decision for them.

‍

The Gravity of Transformation

‍

Describing the required architecture is the easy part. Building it inside a living enterprise, while that enterprise is simultaneously managing the disruptions the architecture is meant to address, is something else entirely. The difficulty of this transformation deserves candor, because underestimating it is how organizations end up with pilots that never scale and strategies that exist only in slide decks.

‍

The first obstacle is structural. Most large enterprises are organized to prevent exactly the kind of cross-functional intelligence sharing that a Sense, Think, Act & Learn architecture requires. Supply chain, finance, marketing, and commercial teams operate with different data, different key performance indicators, and frequently competing incentives. The enriched ontologies described above (the connective tissue that allows insight in one domain to be contextualized and applied in another) cannot be overlaid on an organization whose fundamental design fragments knowledge into silos. The transformation is as political as it is technological. It requires dismantling institutional boundaries that have existed for decades, and that have constituencies invested in their preservation.

‍

Navigating this political reality requires deliberate sequencing. The most successful transformations begin with the most willing function, typically a business unit leader who is already experiencing the pain of fragmented intelligence and is looking for a better answer. That initial partnership establishes a working proof point. From there, the architecture extends to adjacent functions where the value is visible, and the political cost of exclusion begins to exceed the cost of participation. The executive sponsor for this kind of transformation must sit at the enterprise level, such as a CEO, COO, or chief strategy officer with the authority to arbitrate across functional boundaries. Without that sponsorship, architecture will be optimized within silos rather than across them, which is precisely the failure mode it is designed to prevent.

‍

The second obstacle is capability. The overwhelming majority of Fortune 500 companies do not currently possess the data architecture, the decision science talent, or the organizational design expertise to build what is being described here. The technology exists. The institutional capacity to deploy it is not yet scalable. This is a recognition that the gap between what the environment demands and what most organizations can deliver is substantial, and that closing it requires sustained investment in people, infrastructure, and institutional redesign that extends well beyond a technology procurement decision.

‍

The third obstacle is temporal. Building an enterprise intelligence layer is a multi-year, board-level commitment. The paradox is that this commitment must be sustained through the very disruptions it is meant to address. An organization cannot pause the polycrisis while it upgrades its decision architecture. It must transform while under fire, running quarterly earnings, managing supply chain volatility, responding to regulatory shifts, and navigating workforce transitions simultaneously. The organizations that wait for stability before beginning will discover that stability is not returning.

‍

A fourth obstacle deserves separate mention, specifically cryptographic debt. The quantum threat compounds every difficulty described above. An organization already struggling to integrate its data architecture across functional silos must now also inventory every cryptographic dependency in that architecture (every certificate, every key exchange, every hardware security module) and develop a migration plan that coordinates with vendors, suppliers, and partners who face the same challenge on their own timelines. NIST has published the post-quantum cryptography standards. CISA has issued federal procurement guidance requiring quantum-resistant products. Google has already migrated its own services to post-quantum key exchange. The standards exist. The government mandates are arriving. And peer-reviewed research estimates that large enterprises will require twelve to fifteen years for complete migration, a timeline that, if quantum capability arrives by 2030, leaves a multi-year window of structural vulnerability that no amount of heroic leadership can close after the fact.

‍

There is, however, a practical answer to these obstacles, and it begins with a single business process. The organizations that have successfully built enterprise intelligence did not attempt to transform everything at once. They identified one high-value process (demand sensing, promotion optimization, supply allocation) where the intelligence layer could demonstrate a measurable financial return within quarters, not years. That initial return creates a profit engine. The margin improvement from the first process funds the buildout of the next, which funds the next, which progressively extends the architecture across the enterprise. The transformation finances itself. In practice, the right initial process (chosen for its data readiness, its margin sensitivity, and its cross-functional visibility) should be self-liquidating within the first year of operation, ideally within the same corporate fiscal year. That is the threshold that converts a transformation initiative from a strategic bet into a funded program with its own P&L justification. This is how the investments are de-risked in practice, by converting each phase from a cost center requiring board-level faith into a revenue contribution that justifies the next phase on its own terms. The organizations that stall are invariably the ones that try to build the entire architecture before demonstrating value. The ones that succeed start with a single proof point that makes the financial case undeniable and then compound.

‍

One caveat deserves honesty. Not every first proof point succeeds. The process chosen may underperform, the data may prove less ready than anticipated, or the organizational learning curve may be steeper than projected. The discipline required is to treat an underperforming proof point as a learning cycle, not a failure verdict, and instead to diagnose why the return fell short, adjust the process selection criteria, and redeploy. The organizations that abandon architecture after a single disappointing quarter are making the same mistake as those who never started. They are confusing a difficult implementation with a flawed strategy. The strategy is sound. Execution requires persistence and the willingness to learn from what the first cycle reveals.

‍

All of this is reason for urgency. The difficulty of the transformation is precisely what makes it a source of competitive advantage, because most organizations will not do it.

‍

‍

Systemic Resilience as Competitive Advantage

‍

Resilience is too often framed as a defensive posture, the ability to absorb shocks and keep operating. That framing is incomplete. It undersells the strategic opportunity.

‍

Organizations that build genuine systemic resilience (the Sense, Think, Act & Learn architecture described above) outperform. They outperform because they see dislocations earlier, interpret them more accurately, respond more quickly, and learn more deeply from each cycle. Over time, that structural advantage compounds. The resilient organization is harder to break, faster to adapt, more precise in its resource allocation, and better positioned to exploit the opportunities that disruption always creates for those who can recognize them.

‍

The evidence is emerging. BCG’s 2026 CEO survey found that nearly all chief executives believe AI agents will produce measurable returns this year, and that half believe their job is on the line if AI does not pay off. The World Economic Forum’s Industry Strategy Meeting in Munich heard executive after executive describe the shift from “efficiency to resilience” as the defining strategic reorientation of the moment. BCG’s multi-year research found that the top 5% of AI-mature companies (those that have moved beyond pilot-phase adoption) are achieving 1.7 times the revenue growth and 3.6 times the total shareholder return of lagging firms, though, as with any AI-maturity segmentation study, these figures reflect some selection bias toward organizations that were already structurally advantaged before their AI investments began.

‍

Resilience is the foundation of competitive advantage in a volatile world.

‍

The best-run organizations of the next decade will not be the ones with the largest balance sheets or the most diversified portfolios. They will be the ones that have built the systemic capability to sense disruption, think through its implications, act with precision, and learn from every cycle, continuously, structurally, and enterprise scale.

‍

‍

That is the strategic prize. Polycrisis² will sort organizations into two categories, those that built the architecture and those that did not. The sorting is already underway.

The Imperative

‍

The disruption we are living through will persist. The base polycrisis (geopolitical, economic, energetic, alimentary) is real and intensifying. The exponent (AI and quantum computing simultaneously restructuring the competitive, informational, and security landscape) is accelerating. The product of the two is the new operating environment.

‍

The entry point is specific. Identify one high-value business process where an enterprise intelligence layer can demonstrate a measurable financial return within a single fiscal year. Let that return fund the next process, and the next. The transformation that seems impossibly large when conceived as a whole, becomes manageable and self-financing when begun at the right starting point.

‍

Polycrisis² is a description of where we are. The question, for organizations and individuals alike, is whether we will match the complexity of the challenge with the sophistication and elegance of our response, or whether we will continue applying linear solutions to an exponential problem.

‍

The architecture for resilience exists. The leadership models are clear. The technology is available. What remains is the decision to build.

‍

‍

‍

Polycrisis²™ is a trademark of Stephen F. DeAngelis.

Sense, Think, Act & Learn™ is a trademark of Enterra Solutions.

Stephen F. DeAngelis is the founder, president, and CEO of Enterra Solutions and Massive Dynamics, two companies that apply artificial intelligence and advanced mathematics to complex enterprise challenges. His career spans international relations, national security, and commercial technology. He has served in visiting research affiliations with Princeton University, the Oak Ridge National Laboratory, the Software Engineering Institute at Carnegie Mellon University, and the MIT Computer Science and Artificial Intelligence Laboratory. He is a founding member of the Forbes Technology Council. DeAngelis holds patents in autonomous decision science and has been recognized by Forbes as a Top Influencer in Big Data and by Esquire magazine as the “Innovator” in its Best and Brightest issue.

‍

About the Stephen DeAngelis Explainer Brief Series

The Stephen DeAngelis Explainer Brief series applies critical reasoning to the complex issues facing society today. In an era of compounding uncertainty and deepening division, the series aims to build understanding and community by making consequential topics (from artificial intelligence and geopolitics to organizational resilience and national competitiveness) accessible through rigorous analysis, current evidence, and honest assessment. Each installment is written in the belief that better explanations lead to better decisions, and that informed citizens and leaders are the foundation of a stable functioning society. Published under the DeAngelisReview imprint – www.deangelisreview.com.

The Unintended Consequences to the Food Supply Chain

Since the start of Iran conflict in the Middle East, everyone is aware of the sharp rise in the price of a barrel of oil and the subsequent rise in gasoline and diesel prices at the pumps. Those were not the intended consequences of the conflict but they were consequences that could be foreseen. What is becoming more obvious is that unintended consequences of the conflict reach far beyond the energy supply chain. The staff at the Economist writes, “The longer the Strait of Hormuz stays closed, the larger another crisis looms. The waterway is vital for more than just fuel. Gulf states rely on it to import much of their food — roughly 80% of the region’s calories come from elsewhere. … Other consequences will be felt far more widely. Around a third of the world’s trade in the raw materials used to make fertilizers passes through the Strait. Iran is one of the world’s largest producers of urea, a crucial ingredient in synthetic nitrogen fertilizers. These are essential for modern agriculture: it is thought that half the global population would not be adequately fed without them. With planting season around the corner, a shortage could affect yields worldwide.”[1]

The staff at World Politics Review (WPR) discusses other areas that will be significantly affected by the conflict. They write, “In Brazil, which is heavily dependent on imported fertilizers for its soybean and corn production, the Agriculture Ministry assesses there is ‘a very high risk of supply shortages and rising domestic prices.’ And as Nima Shokri and Salome Shokri-Kuehni of the United Nations University recently wrote, farmers across much of sub-Saharan Africa are already having a hard time affording adequate fertilizer for their crops. The war-induced supply crunch will likely cause them to further reduce fertilizer usage further, exacerbating food insecurity.”[2] And the WPR staff observes that U.S. farmers will be affected as well. In a letter to President Trump, Zippy Duvall, President of the American Farm Bureau Federation, wrote, “Not only is this a threat to our food security — and by extension our national security — such a production shock could contribute to inflationary pressures across the U.S. economy.”

Fertilizer shortages are not the only unintended consequences of the conflict. Food journalist Flora Southey reports if the conflict continues there is likely to be a packaging crisis. She explains, “The food and beverage industry is inherently tied to plastic. As much as 40% of global plastic packaging is dedicated to the sector, with the vast majority of ready meals, bread, rice, cereals, meat, fish, and dairy products packaged in plastic. And that’s not even getting started on beverages: an estimated 600bn plastic bottles are produced globally for water alone, every year. So, it’s big news for F&B when plastic prices soar. And that’s what forecast to happen, imminently, due to rising tension in the Middle East and attacks on a vital choke point: the Strait of Hormuz.”[3] That means every consumer packaged goods (CPG) company in the world, and their customers, will feel the impacts of the conflict.

Beyond Food Supply Chains

Beyond the energy and food supply chains, other unintended consequences are being felt. Economics reporter Patricia Cohen explains, “In Kansas, home buyers saw 30-year mortgage rates edge above 6 percent. … In Western India, families mourning the death of a loved one discovered that gas-fired crematories had been temporarily closed. In Hanoi, Vietnam, gas station owners posted ‘sold out’ signs. In Kenya, tea growers and traders worried their exports to Iran would rot on the dock. … Cargo deliveries have been stranded, shipping charges have increased and insurance premiums have skyrocketed. Yes, the price of gas at the pump is affected. But so is the price of food, medicine, airplane tickets, electricity, cooking oil, semiconductors and more.”[4]

Air traffic to and from the Middle East has all but disappeared. As of mid-March, journalist Niraj Chokshi reported, “More than 52,000 flights to and from the Middle East — more than half of all flights planned in the region — have been canceled since the war began on Feb. 28, according to Cirium. An estimated six million passengers have been affected. The costs are adding up, too. Crews and planes for Middle Eastern carriers were displaced. And tourism to the region has effectively ground to a halt.”[5]

Consumers have also taken notice of worsening economic conditions. Journalist Matt Grossman reports, “Consumer sentiment declined to start March, according to the University of Michigan’s monthly survey, one of the first readings on public opinion about the economy since the start of the Iran war.”[6] Joanne Hsu, the survey’s director, told Grossman, “Interviews completed prior to the military action in Iran showed an improvement in sentiment from last month, but lower readings seen [following attacks on Iran] completely erased those initial gains.” When consumer sentiment declines, spending decreases, and the economy suffers.

Concluding Thoughts

Supply chain journalist Marina Mayer writes, “The threats of tariffs, trade wars and protectionism, as well as disruption to supply chains and shipping caused by regional conflicts in the Middle East and Russia/Ukraine are on the top of every board agenda.”[7] She reports that the greatest fear of business executives according to the Allianz Risk Barometer is: “Global supply chain paralysis due to a geopolitical conflict involving multiple major economies, halting the movement of goods and raw materials.” A company can plan a coherent strategy based on fear. It’s clear that companies need help navigating a future filled with uncertainty.

As I wrote in a LinkedIn article, “The systems we’ve built — global supply chains, AI systems, digital platforms, financial markets — have become far more complex than the frameworks we use to manage them. And it’s starting to show. We see it when: Supply chains collapse under pressure; AI systems generate answers they can’t explain; organizations collect more data than ever… but struggle to make better decisions; and complex systems fail in ways nobody predicted. The problem isn’t just volatility or uncertainty anymore. It’s that our systems have outgrown the frameworks used to govern them.” Ways out of this conundrum include leveraging explainable AI solutions like the Enterra System of Intelligence™. The system combines the power of a human-like reasoning and trusted generative AI with glass-box machine learning and real-world optimization to drive intelligent decision-making and fuel business growth.

When unintended consequences dominate the headlines, companies need a way to get ahead of those consequences. Only AI-powered solutions can help accomplish that goal.

Footnotes

[1] Staff, “The World in Brief,” The Economist email, 14 March 2026.

[2] Staff, “Daily Review,” World Politics Review email, 12 March 2026.

[3] Flora Southey, “The packaging crisis about to hit food and beverage,” Food Navigator, 16 March 2026.

[4] Patricia Cohen, “Oil Shock Sends Tremors Through World Economy: ‘This Really Is the Big One’,” The New York Times, 12 March 2026.

[5] Niraj Chokshi, “War Has Grounded High-Flying Gulf Airlines Like Emirates,” The New York Times, 15 March 2026.

[6] Matt Grossman, “Consumer Sentiment Declined This Month, Per Michigan Survey,” The Wall Street Journal, 13 March 2026.

[7] Marina Mayer, “Geopolitical Conflict to Create Black Swan Scenarios for U.S. Supply Chains,” Food Logistics, 7 March 2026.

Today is National Agriculture Day (or National Ag Day) and the theme for this year is “Together We Grow: Celebrating 250 Years of Progress in Agriculture.” The celebration is sponsored annually by the Agricultural Council of America (ACA). This year’s celebration focuses on “the dedication of America’s farmers and the bounty of American agriculture.”[1] Chuck Zimmerman, President of ZimmComm, writes, “This year’s theme emphasizes the collaborative efforts of farmers, ranchers, and all stakeholders in the agricultural sector who work tirelessly to provide safe, abundant, and affordable products and also recognizes 250 years of progress.”[2]

250 Years of Progress in Agriculture

Many people longing to return to the good old days probably don’t know what the good old days were like. Two-hundred-and-fifty years ago agriculture was primarily manual and subsistence-based, with roughly 95% of the U.S. population working in farming. Families used hand tools like hoes, scythes, and plows, relying on human and animal labor to produce crops for personal consumption and limited local trade. It was hard, tedious work. Farming has never been easy, but today automation plays a much larger role in food production and other agricultural activities.

Supply chain journalist Nick Bowman reports, “Traditionally, agriculture conjures images of workers tilling fields and picking crops. But as climate change has led to longer dry seasons, strained water supplies, and reduced yields, farmers have turned to a wide range of modern technologies that are anything but traditional.”[3] He observes that there are now machines that can test soil, harvest crops, leverage precision farming techniques, monitor the weather, and help with workforce assignments. He concludes, “By reducing the need for repetitive manual work and allowing farmers to focus their labor where it’s most valuable, these tools can not only help lower costs, but also improve consistency and quality across harvests. For farmers, the path forward isn’t about choosing between tradition and technology; it’s about using the best of both to stay resilient in the face of growing global climate risks. From water and soil to labor and yields, every smart tool that eases pressure today helps ensure there’s a harvest tomorrow.”

Thanks to technology, the percentage of the U.S. population working in agriculture today is surprisingly small. The staff at the Tennessee Farm Bureau observes, “Agriculture is no doubt our most important industry. However, too few people truly understand that, so we celebrate National Ag Day because it brings to light the less than 2% of the population who are working incredibly hard daily to provide all of us with food, fiber and fuel.”[4] Agriculture, however, is also a job creator. In addition to those working directly on farms, there are a host of supporting activities that move products from farm to table. Those activities include related industries like food processing, manufacturing, and distribution. When they are included, the agricultural and food sector supports approximately 10.4% of total U.S. employment.[5]

Looking to the Future

Although America’s semiquincentennial celebration is a great time to reflect on the past, it’s an even better time to look to the future. Joby Young, Executive Vice President at the American Farm Bureau Federation, explains, “Agriculture has long been defined by innovation, as farmers, especially in recent decades, have leveraged technology to improve yields, adapt to challenges and ensure their farms remain strong and sustainable for seasons to come. Innovation in areas such as data, crop protection tools, and biotech, have all led to better nutrition, improved environmental outcomes and greater food availability and affordability for Americans. But today, as the headwinds in agriculture continue to grow and technology advances, the conversations around innovation matter more than ever.”[6]

Like in most economic sectors, artificial intelligence (AI) is poised to play a significant role in future operations. Young observes, “Today, we’re seeing incredible advancements happening in the world of agricultural innovation. From new equipment that utilizes AI algorithms to platforms with the capability to analyze crop health in real-time, farmers are able to navigate numerous tasks and access insight that previously would have taken days or weeks to complete.” He adds, “We also need to make sure as more technology becomes available, we safeguard the interests of our farmers and work in partnership across the entire supply chain.” We also need to ensure that the technologies that help farmers produce more and better crops are affordable.

Concluding Thoughts

The staff at National Today observes, “Everything we eat, use, or wear every day is provided by agriculture. Most people don’t understand the contribution of agriculture to our lives. … So today we are celebrating agriculture and thanking all the people who work hard to feed the world, look after crops and livestock, and contribute to agricultural production.”[7] In addition to thanking people associated with agriculture, Young believes we should be actively engaged in ensuring that food value chain is resilient and affordable. He concludes, “Innovation touches every part of our food system. Whether you’re a farmer adopting new technology, an innovator with a groundbreaking idea, or a consumer enjoying a variety of healthy, safe, and affordable choices in the grocery store, innovation plays a role. We need to make sure there is partnership on every level so that farmers have access to new ideas and that our researchers and entrepreneurs know the needs of our farms and rural communities. By establishing a pipeline of innovation across the supply chain, we will have a food supply that remains strong and resilient for whatever tomorrow brings.”

Footnotes

[1] Staff, “Ag Day Social Media Toolkit 2026,” Agricultural Council of America.

[2] Chuck Zimmerman, “2026 National Ag Day,” AgWired, 26 January 2026.

[3] Nick Bowman, “AI and Robotics Take Root in Modern Agriculture,” SupplyChainBrain, 25 September 2025.

[4] Staff, “National Ag Day,” Tennessee Farm Bureau.

[5] Kathleen Kassel, “Agriculture and its related industries provide 10.4 percent of U.S. employment,” USDA Economic Research Service, 3 November 2023.

[6] Joby Young, “Ag Innovation: The Key to Tomorrow,” 17 April 2025.

[7] Staff, “National Ag Day,” National Today.

On the small screen, Rosie the Robot, from the animated series “The Jetsons,” convinced growing boomer children that domestic robots could be helpful around the house. Fast forward three score years and humanoid robots are once again making headlines — this time in the real world. In January, a domestic robot named Sprout was introduced. Journalist Matt O’Brien reports, “If its emotive expressions and blinking lights seem vaguely familiar, it might be from generations of Star Wars droids and other endearingly clunky robotic sidekicks dreamed up in animation studios and children’s literature.”[1] Journalist John Leonard reports, “Futurist Peter Diamandis has predicted that general purpose humanoid robots will start appearing in homes this year, able to do some household work and personal assistance tasks.”[2] Leonard points out, however, that the general opinion is that “commercialization of humanoid robots … will get their first jobs in warehouses or factories long before they are ready for homes.”

Humanoids in the Supply Chain

It has been clear for some time that supply chains are getting more automated and more autonomous. Mark H. Yahiro, vice president of business development at RealSense, observes, “Warehousing and logistics have undergone radical changes and several waves of ‘the next big thing’ over the past several decades, with varying impact. Radio frequency scanners made inventory visible. Voice-directed work made picking hands-free. Early automation and goods-to-person systems brought robots into the mix. These advances, while powerful, shared a common limitation: they did not really see the warehouse. They followed instructions, processed barcodes and confirmed voice prompts, but they had very little awareness of what was actually happening in three-dimensional space. We are rapidly entering an era where robots do not simply operate with human commands from static maps and basic point-to-point instructions; they do so autonomously — and with each other.”[3]

Even though most of robots described by Yahiro are unlikely to resemble humans, at some point in the future, humanoid robots are likely to find their way into supply chain operations. Abdil Tunca, a senior principal analyst in Gartner’s supply chain practice, explains, “The promise of humanoid robots is compelling, but the reality is that the technology remains immature and far from meeting expectations for versatility and cost-effectiveness.”[4] The staff at SupplyChainBrain reports, “Despite early enthusiasm for the technology, humanoid robots are expected to progress beyond the experimentation phase for less than 100 companies through 2028, with fewer than 20 companies pushing them live for supply chain and manufacturing purposes. According to an analysis from Gartner, the majority of production deployments of humanoid robots over the next couple years will be limited to ‘tightly controlled environments,’ with the hype surrounding the technology outpacing its actual readiness for widespread use.”[5] Gartner analysts point out, “Despite their potential, humanoid robots face significant barriers to supply chain, logistics and manufacturing adoption.”[6] Those barriers include:

• Technological limitations: “Current models lack the dexterity, intelligence, and adaptability required for complex, unstructured environments such as mixed SKU picking, trailer unloading or exception handling in high velocity warehouses.”

• Integration complexity: “Compatibility with existing systems and workflows remains a challenge.”

• High costs: “Substantial upfront investment and ongoing maintenance expenses must be weighed against uncertain returns. With the current technology and costs, humanoids cost multiple times more than task specific polyfunctional robots while delivering lower throughput and uptime.”

• Energy constraints: “Limited battery life restricts operational time for high-mobility tasks.”

Journalist Sarah Chea reports that advances in battery technology could overcome the energy constraint barrier. She writes, “A critical component for the expansion of the humanoid robot market is the development of solid-state batteries — with Korean companies at the forefront. Often dubbed the ‘dream battery,’ solid-state technology replaces liquid electrolytes with solid ones. … Solid-state batteries offer distinct advantages in energy density, safety and thermal management, potentially extending operational runtimes to five to eight hours or more. While high-nickel batteries remain the standard for humanoid robots in the near term, a gradual shift toward solid-state adoption is expected over the coming years.”[7]

Global analyst Jan Burian is skeptical that humanoid robots have a place in the supply chain. He writes, “Many assume industrial humanoids should look like humans, move like humans and perform tasks like industrial robots, just with legs. This assumption drives unrealistic expectations and skepticism. Compared to humanoids, industrial robots are deterministic, more precise, more reliable and safer. They can be mobile, collaborate with humans and perform high-volume tasks efficiently. Evaluating humanoids as replacements for industrial robots sets them up for inevitable failure. The core issue is that humanoids are not industrial robots yet.”[8] Admittedly, he has a point.

Concluding Thoughts

Science fiction films will likely to continue to feature humanoid robots; however, it may be a few years before (if ever) they are widely found working in the supply chain. Michael Tam, Chief Brand Officer at Chinese firm UBTech, admits his company’s robots “are at best only half as efficient as human workers.”[9] Nevertheless, companies around the world are working to improve humanoid machines. You might want to watch a recent episode of CBS’s “60 Minutes” to see how Hyundai and Boston Dynamics are working to integrate humanoid robots into the workforce. Predictions about how large the market for humanoid robots will be in the years ahead are staggering. Chea reports, “The humanoid robot market, which was valued at around $2.43 billion last year, will grow to $66 billion by 2032, according to a prediction by Fortune Business Insights, a global market research firm.” And journalist Muflih Hidayat reports, “China’s long-term projections indicate potential humanoid robot output reaching 59 million units by 2050.”[10]

Whether or not humanoid robots find a significant place in supply chain operations, robotics and automation will. That means that companies around the globe will fight for a market share of the robotics business. Boston Dynamics CEO Robert Playter believes the U.S. currently leads the race; however, he admits China is closing the gap. He told correspondent Bill Whitaker, “The Chinese government has a mission to win the robotics race. Technically I believe we remain in the lead. But there’s a real threat there that, simply through the scale of investment, we could fall behind.”[11] The race to develop ever more capable robots will also involve supply chains that provide actuators, sensors, motors, battery systems, and advanced artificial intelligence systems. Whether humanoid robots play a large role in supply chains or not, the future should be fun to watch.

Footnotes

[1] Matt O’Brien, “Not ready for robots at home? Friendly new humanoid maker thinks it may change minds,” Tech Xplore, 27 January 2026.

[2] John Leonard, “Are you ready for the humanoid robot bubble? - Asian Tech Roundup,” Computing, 30 January 2026.

[3] Mark H. Yahiro, “How Physical AI Will Reshape the Warehouse,” SupplyChainBrain, 29 January 2026.

[4] Jennifer Guhl, “Few Companies Will Scale Humanoid Robots by 2028, Predicts Gartner,” Consumer Goods Technology, 27 January 2026.

[5] Staff, “Hype for Humanoid Robots is Outpacing Supply Chain Readiness,” SupplyChainBrain, 27 January 2026.

[6] Staff, “Humanoid Robots for Supply Chain Will Stall at Pilot Scale,” Material Handling & Logistics, 26 January 2026.

[7] Sarah Chea, “Beaten by China in EV batteries, Korea finds an edge in humanoids,” Korea JoongAng Daily, 3 February 2026.

[8] Jan Burian, “Industrial Humanoid Robots: Are We Even Talking About the Same Thing?” IndustryWeek, 5 February 2026.

[9] Leonard, op. cit.

[10]Muflih Hidayat, “Tesla’s 2026 Humanoid Robot Production Timeline and Manufacturing Challenges,” Discovery Alert, 3 February 2026.

[11] Bill Whitaker, Aliza Chasan, Marc Lieberman, Cassidy McDonald, “Boston Dynamics is training an AI-powered humanoid robot to do factory work,” 60 Minutes Overtime, 4 January 2026.

Like Alice in Wonderland

In Lewis Carroll’s Alice’s Adventures in Wonderland, Alice came upon a Cheshire Cat in a tree and asked, “Would you tell me, please, which way I ought to go from here?” The Cat replied, “That depends a good deal on where you want to get to.” To which Alice replied, “I don’t much care where.” The wise Cat responded, “Then it doesn’t much matter which way you go.” Dismayed, Alice said, “So long as I get somewhere.” To which the Cat replied, “Oh, you’re sure to do that, if only you walk long enough.” Like Alice, many global business and political leaders are now asking, “Which way should I go from here?” With no clear answer to that question, the world will nevertheless end up somewhere. At the start of last year, BCG analysts outlined four broad “tectonic shifts in global trade corridors.”[2] Some of their observations remain valid, while recent developments have brought other predictions into question. Below are their forecasts.

• North America. A year ago, BCG analysts observed, “North America is solidifying into a resilient trade bloc that will continue to reduce its dependence on Asia, especially China. For now, this substitution of supply sources seems to be working.” Fast forward. A year to the day that BCG analysts published their findings, President Trump insisted the United States-Mexico-Canada trade agreement (USMCA) is no longer relevant for the U.S.[3] Although the USMCA may not be renewed, the fact remains that the U.S. does massive combined trade with Canada and Mexico (i.e., around 30% of its total goods trade). With the end of USMCA a real possibility, the Canadian Prime Minister struck a new trade deal with China.[4]

• China. According to BCG analysts, “China will emerge as the stronger trade partner for the rest of the world as its commerce with the West slows. Increasingly, indigenous technologies and deeper economic relationships with fast-growing emerging markets will drive growth.” One way China is integrating its economy with the rest of the world is by embracing the post-carbon economy. As America doubles down on the waning carbon economy, China leads in supplying the world with green technologies and is making itself irreplaceable in that arena. And that’s the arena in which the emerging global south is forging its future. Vikram Singh, a Senior Director at RMI, explains, “From Brazil to Morocco and Namibia, from Bangladesh to Egypt and Vietnam, [the Global South] has already overtaken the Global North in terms of the share of solar and wind in electricity generation, or the share of electricity in final energy.”[5]

• The Global South. Speaking of the Global South, BCG analysts noted, “The Global South will rise as a force in world trade, powered by dynamos such as India and Southeast Asia, as developing nations contribute more to global supply chains and develop new capabilities. South-South trade will also surge and is moving beyond exporting natural resource-based commodities to more sophisticated manufactured goods.” Demographics alone can help explain the importance of the Global South. Most of the world’s population (about 85% of it) lives in the Global South. While the Global North is growing old, the Global South remains young. Jason Hsu, founder and CIO at Rayliant Global Advisors, explains, “Famed French sociologist and philosopher Auguste Comte said that demography is destiny. Indeed, aging demographics is one of the most predictable macro challenges for humanity — the ultimate slow-motion train wreck. The world is getting older. … Over the next 20 years, aging developed markets will lean on emerging markets for their younger and more plentiful workforce, giving emerging markets newfound bargaining power.”[6]

• The European Union. A year ago, the BCG analysts noted, “The European Union’s trade growth with China will largely stagnate. The region is becoming more reliant both on long-standing trade partners such as the US and Japan and emerging markets such as India, Turkey, and Africa.” Much has changed over the past year, including new tariffs and the feud over Greenland. These developments have motivated the EU to look for growth opportunities beyond trade with the United States. Recently, for example, progress was made towards ratifying a huge trade deal with South America.[7] The Economist reports, “The pact creates a free-trade bloc of more than 700m people. By 2040, removing tariffs on around 90% of goods on both sides and simplifying trade in services will boost the EU’s exports by €49bn ($56bn) and Mercosur’s by €9bn, the Europeans estimate.”[8] The EU also completed “the mother of all trade deals” with India. If the deal is ratified and implemented by the EU and India, it will create a free trade zone that includes some 2 billion people and one-quarter of global GDP.[9]

As the world order continues to reorganize, companies must act. But, as the BCG analysts ask, “When and where?” Journalists from The Economist observe, “Growing geopolitical risks will force firms to incorporate more redundancy and flexibility into their operations. That will probably entail more diffuse production in multiple locations around the world, although many of these facilities will be final-assembly plants, not necessarily whole factories. Companies will have to think more carefully about which industries they enter where.”[10] Futurist Mark Van Rijmenam believes the next world order won’t be organized around ideological differences but resources. He explains, “Nations now jockey for energy and mineral access the same way they once fought over oil fields.”[11] What he doesn’t attempt is a description of what this new world order may look like.

Concluding Thoughts

Few observers doubt that the world order is experiencing fundamental changes. The question for business leaders is: How do I move forward in a chaotic world? BCG analysts insist, “The best way to plan in an unpredictable context is to think in terms of scenarios.” This advice is not new. Journalist Anjli Raval reports, “For decades, scenario planning has helped organizations map out a range of futures based on variables including economic shifts, technological leaps and regulatory changes. Pioneered at Shell — which anticipated the 1973 oil shock — scenario planning has been a corporate staple since.”[12] She adds, “Scenario planning was never about predicting the future — it’s about training for it.” The staff at John Galt highlights five ways that scenario planning can enhance business decisions.[13] They are: understanding tradeoffs and risks; evaluating impact probabilities; being adaptable and responsive; thinking strategically end-to-end; and unlocking hidden opportunities.

Scenario planning is an integral part of the Enterra System of Intelligence™. Built upon Enterra’s Autonomous Decision Science^® (ADS^®) & Generative AI technology platform, the System is a cross-enterprise analysis and control system that spans the data and process layers of SAP’s value-chain expansive transactional Systems of Record. Enterra’s System will autonomously perform end-to-end optimization, planning, and decision-making at scale and at the speed of the market with human-like intelligence and reasoning. A defining feature of the Enterra System of Intelligence is its architecture which is comprised of a set of interconnected business applications that leverage a common analysis, optimization, and decision-making/learning platform. These business applications include:

• Enterra Consumer Insights Intelligence System™

• Enterra Revenue Growth Intelligence System™

• Enterra Demand and Supply Intelligence System™

• Enterra Global Insights and Decision Superiority System™ (which is part of Enterra Business WarGaming™)

Enterra’s System of Intelligence can help business leaders rapidly explore a multitude of options and scenarios. Regardless of where the world order is headed, companies benefit from exploring numerous possible outcomes, especially the most likely outcomes. Doing so they can best mitigate dangers and take advantage of emerging opportunities.

Footnotes

[1] Aparna Bharadwaj, Dominic DeSapio, Marc Gilbert, Nikolaus Lang, Kasey Maggard, Michael McAdoo, Morten Seja, and Peter Ullrich, “Trade in Transition: How to Prepare for a Patchwork World Order,” Boston Consulting Group, 8 January 2026.

[2] Priscille Arbour, Aparna Bharadwaj, Tim Figures, Marc Gilbert, Nikolaus Lang, Georgia Mavropoulos, Michael McAdoo, and Cristián Rodríguez-Chiffelle, “Great Powers, Geopolitics, and the Future of Trade,” Boston Consulting Group, 13 January 2025.

[3] Bo Erickson and David Shepardson, “Trump says USMCA is irrelevant for US,” Reuters, 13 January 2026.

[4] Maria Cheng, “Canada, China slash EV, canola tariffs in reset of ties,” Reuters, 16 January 2026.

[5] Vikram Singh, “Powering Up the Global South,” Rocky Mountain Institute.

[6] Jason Hsu, “Demographic Shifts: How Aging Economies Impact Emerging Market Assets,” Rayliant Insights.

[7] Patricia Cohen, “E.U. and South America to Form Free-Trade Zone With 700 Million People,” The New York Times, 9 January 2026.

[8] Staff, “Europe and South America seal a trade pact for the Trump era,” The Economist, 11 January 2026.

[9] Staff, “‘The Mother of All Trade Deals’,” World Politics Review, 27 January 2026.

[10] Staff, “Geopolitics is warping multinationals’ commercial decisions,” The Economist, 15 January 2026.

[11] Mark Van Rijmenam, “What Will Decide the Next World Order After Oil?” The Digital Speaker, 13 January 2026.

[12] Anjli Raval, “Scenario planning is getting a stress test,” Financial Times, 12 May 2025.

[13] Staff, “5 Ways Scenario Planning Transforms Your Supply Chain Strategy,” John Galt Blog, 25 June 2025.

Economist Mark Thornton, a Senior Fellow at the Mises Institute in Austria, explains, “For a visual description, imagine a graph where the vertical axis forms the vertical height line of the K. From the vertical axis two lines emerge to form the K — one upward sloping and one downward sloping. The economic interpretation starts at a point in time when the economy was moving in a kind of equilibrium with all sectors and industries moving together in unison, including aggregate income and labor markets. Then, at that point in time, instead of the overall economy moving in unison upward with economic growth, or downward into recession, there is a noticeable divergence. The upward trend line of the K represents the fortunes of high-income salary earners, the wealthy, and the high-tech and luxury goods industries. The downward trend line of the K represents the fortunes of the working class and the poor. Essentially, the rich are getting richer, the poor are getting poorer, and the middle class is shrinking.”[2] He adds, “This divergence is too stark to deny.”

The K-shaped Economy — Not New, but Important and Changing

Due to the recent spate of articles about the K-shaped economy, one might conclude it is a recent phenomenon. However, economists Beth Ann Bovino and Matt Schoeppner explain, “The K-shaped economy reflects a long-standing divergence where wealthier households benefit from asset appreciation, better schools and technology, while lower-income families face structural disadvantages and financial strain.”[3] They note, however, “Today, the K-shape has become the defining lens for understanding resilience and vulnerability between the haves and the have-nots. It continues to steer consumption and borrowing patterns, widen generational divides and influence policy debates. Moreover, new forces — higher interest rates, persistent inflation and the rapid adoption of transformative technologies such as artificial intelligence (AI) — appear to be amplifying these disparities even further. After decades of debate on the subject, it’s no surprise that the effects of rising income inequality are complex. To some extent, it’s necessary for a market economy to function, as it incentivizes investment and expansion. However, excessive income concentration can also undermine growth by fueling political polarization, while reducing overall demand in the economy.”

Bovino and Schoeppner assert that “rapid technological change” is reshaping economic dynamics and calls this shift “inevitable.” If you are wondering whether the term “K-shaped economy” is simply a buzzword used by the popular press, Mark Zandi, chief economist at Moody’s Analytics, insists it is not. He explains, “This is not a cyclical or temporary phenomena. This is a structural, fundamental issue.”[4] Bovino and Schoeppner conclude, “How the U.S. adapts workforce skills will ultimately determine whether AI-driven innovation — and the new industries it spawns — flattens the K or keeps the economy on its familiar path. Therefore, building digital literacy will be a critical step in preparing current and future workers to integrate these tools into their workflows.”

How the K-shaped Economy Affects Consumer Spending

Business reporter Lora Kelley observes, “‘K-shaped’ entered the popular parlance in 2020 and is now ubiquitous, used for divergent consumer outlooks, unsteady spending at stores and restaurants, and even corporate stumbles, as some companies, especially tech-focused ones, thrive while the prospects of others decline. Moody’s Analytics recently estimated that the top 10 percent of households were responsible for nearly half of all spending.”[5] While a model — like the K-shaped model — can be useful, Katie Thomas, an analyst with the Kearney Consumer Institute, cautions that any simplified model fails to capture the complexity of the economy. She explains, “Economic models like the K economy can be great tools for simplifying and illustrating complex economic arguments. They can also limit our understanding of what is happening to consumers at all levels of the economies they are trying to describe. Getting at consumer motivations and a deeper understanding of consumer behaviors requires a much more nuanced approach than just pointing out asymmetry.”[6]

Thomas laments, “K economy analytics tend to blend divergent consumer behaviors together and then spread them like the data equivalent of peanut butter across a series of demographic cohorts, creating an ‘average’ consumer who does not, in fact, actually exist.” Thomas describes consumers located on either the “arm” or the “leg” of the economy’s K. At the tip of the arm, Thomas places “individuals whose names and faces include those who regularly appear in the media with super-high incomes.” She continues, “Slightly below them, but still well above everyone else, we find the ‘stable/responsible,’ consumers who are stable and getting by just fine. … At the bottom of the arm, dangling perilously close to the middle or leg of the K, we have the ‘on thin ice’ — high earners whose lack of budgeting and profligate spending has them overleveraged and exposed.”

Thomas also divides consumers on the leg into three segments. She explains, “Starting at the top of the ‘leg’ [we find] consumers who are ‘comfortable.’ Their incomes technically put them in the bottom of the K, but their overall financial position is more secure on a day-to-day, year-to-year basis thanks to a variety of factors. Next are those who are ‘stable/responsible,’ relatively lower-income individuals who are constrained in their ability to spend but who have managed by choice and/or necessity to find ways to live within their means. Toward the bottom of the leg, we find low- to no-income people who are truly ‘on thin ice.’ These are folks struggling to pay their basic bills at the same time they are being crushed under a mountain of structural debt-promoting forces.”

Mark Mathews, Chief Economist and Executive Director of Research at the National Retail Federation, reports that K-shaped economy is becoming evident in consumer spending. He explains, “What is clear is that across lower- to middle-income households, growth in spending has begun to slow. However, top-line spending remains robust, and some sectors have even managed to retain or grow their share across income groups. We expect to see continued growth at a top-line level for retail into 2026, but it is clear that not all segments of the consumer will be driving this growth.”[7] Mathews conclusions are backed up by data from the Federal Reserve Bank of New York, which reports, “Spending among higher-income U.S. consumers has increased by 2.3% over the last three years, compared to just 1.6% for middle-class households making between $40,000 and $125,000, and less than 1% for those earning under $40,000.”[8]

Concluding Thoughts

When you look at variations in spending by consumers of different ages, patterns emerge that only relate peripherally to the K-shaped economy. Younger consumers are markedly more pessimistic about the economy than older consumers. Journalist Helen Atkinson writes, “Retailers could get whiplash trying to keep up with youthful consumer spending habits, but they better get good at it as the upcoming generations emerge as mature consumers.”[9] At Enterra Solutions^®, we avoid generalizations about consumers and look more deeply into the consumer landscape. Our Enterra Consumer Insights Intelligence System™ allows clients to quantitatively uncover and logically understand the inter-relationships that lead to heightened consumer understanding, hyper-personalized product recommendations, and new product innovation. This is critical in today’s complex consumer landscape. The bottom line, however, is that, for consumers on the leg of the K, this economy is nothing special.

Footnotes

[1] Christopher Rugaber, “Here’s why everyone’s talking about a ‘K-shaped’ economy,” Associated Press, 1 December 2026.

[2] Mark Thornton, “The K-Shaped Economy,” Mises Wire, 15 December 2025.

[3] Beth Ann Bovino and Matt Schoeppner, “The K-economy in 2026: Same story, new amplifiers,” US Bank Economic Commentary, 7 January 2026.

[4] Thornton, op. cit.

[5] Lora Kelley, “When Did Everything Become ‘K-Shaped’?” The New York Times, 19 December 2025.

[6] Katie Thomas, “Hidden dimensions of the K-shaped economy: detailing how income, lifestyle, and circumstance shape consumer stress and spending,” Kearney, Fourth Quarter 2025.

[7] Mark Mathews, “Is retail spending really K-shaped?” National Retail Federation, 5 February 2026.

[8] Staff, “‘K-Shaped’ Spending Gap Widens for U.S. Consumers,” SupplyChainBrain, 4 February 2026.

[9] Helen Atkinson, “Retailers Warned of the Counter-Intuitive Spending Habits of Gloomy Young People,” SupplyChainBrain, 14 January 2026.

If you’ve spent any time in a boardroom or a strategy session over the past three decades, you’ve heard the diagnosis: we live in a VUCA world — volatile, uncertain, complex, ambiguous. The term was born at the U.S. Army War College in the late 1980s, military shorthand for the post-Cold War mess. Business consultants loved the martial ring of it. They adopted it eagerly, and for years it served as the organizing grammar of corporate strategy: the world is VUCA, therefore be agile, be adaptive, be ready.

I used that language myself. In the aftermath of September 11th, I founded my first company to help the government fix a problem that had suddenly become existential: national security agencies sitting on mountains of data they couldn’t talk to each other about, let alone integrate. We were building resilience models at Carnegie Mellon’s Software Engineering Institute, trying to figure out how critical infrastructure — public and private — could withstand threats that no one had gamed out. Volatility was the right word for what we faced. Uncertainty was the water we swam in. And complexity? That was the landscape we were trying to map with tools that hadn’t been built yet.

But something shifted. VUCA didn’t fail because it was wrong — it failed because it became wallpaper. It described the weather without offering an umbrella. Worse, it normalized perpetual crisis management. Leaders got comfortable saying “the world is VUCA” the way they might say “the sky is blue” — true, but not useful. As Arthur D. Little’s Shift Institute argued just this month, VUCA “presents a depiction of the world with four separate dimensions rather than interconnected and intertwined elements.” Diagnosis without a recipe.

Meanwhile, VUCA masked the deeper fragility building beneath the surface of globalization. Just-in-time supply chains juiced efficiency but left systems brittle as glass. Deregulated financial instruments spread risk so wide that no one could see where it had pooled. We chased short-term wins and offshored not just jobs but institutional memory. That’s strategic nearsightedness at its worst — and it echoes Keynes’s warnings about Versailles, where the architects of peace mistook order for stability and kicked the can down the road toward the next collapse.

Enter BANI

In 2018, the futurist Jamais Cascio named what many of us were feeling but hadn’t yet articulated. He called it BANI: brittleness, anxiety, non-linearity, incomprehensibility. Where VUCA described conditions, BANI describes what those conditions do to us — the emotional and structural damage of living inside systems that fail not gracefully but catastrophically.

I want to be precise about what Cascio got right, because it matters.

Start with brittleness. This isn’t volatility wearing a different hat. A volatile market swings — it moves, and you can ride the movement if you’re paying attention. A brittle supply chain doesn’t swing. It snaps. No warning, no gradual degradation. One day it works; the next day you’re on the phone with your board explaining why you can’t ship product.

Anxiety is the one that gets under my skin, because I watch it warp decision-making in real time. Uncertainty is an intellectual state — you don’t know what’s coming, and you can work with that. Anxiety is different. It’s physiological. It’s the room full of smart people who can’t stop bracing for impact long enough to think clearly.

Non-linearity is the one that keeps the mathematicians up at night. (I say this as someone who has spent far too many nights in that category.) Complexity tells you there are many moving parts. Non-linearity tells you that a tiny input over here can produce a wildly disproportionate output over there — and good luck predicting which one.

And then there’s incomprehensibility, which is genuinely new. Ambiguity means the picture is fuzzy. Incomprehensibility means there may not be a picture at all.

Cascio’s book, Navigating the Age of Chaos, published last October with Bob Johansen and Angela Williams, makes a compelling case that BANI is the honest description of our present condition. I agree. His proposed responses — resilience, empathy, improvisation, intuition — are admirable human capacities.

But here’s where I part company — respectfully, and based on three decades of building operational systems for Fortune 500 companies and governmental agencies.

The Missing Layer

BANI’s response set is entirely human and behavioral. Resilience as a personal quality. Empathy as a leadership practice. Improvisation as a cultural capacity. Those are necessary. They are not sufficient.

Here’s why. The challenge of our moment is not merely that people feel anxious and leaders need more empathy. The challenge is that the systems on which seven billion people depend — energy grids, financial networks, health care logistics, food supply chains, democratic governance itself — are operating beyond the cognitive capacity of any human or team of humans to monitor, diagnose, and correct in real time.

I sit with Fortune 500 executives regularly. When someone’s supply chain spans 40 countries and 10,000 SKUs, and the conversation turns to “how do we build resilience,” the answer cannot be “practice empathy.” I don’t say that to be glib. I say it because I’ve watched empathetic, well-led organizations get blindsided by cascading failures that no amount of emotional intelligence could have prevented. The answer has to be engineered. It has to be mathematical. It has to be embedded in the architecture of the system itself — not layered on as a leadership seminar after the fact.

The behavioral responses and the engineered responses are complementary. But only one of them scales.

What Engineering Resilience Actually Looks Like

I/We learned this lesson the hard way. After September 11th, the national security community had more data than any human could process — signals intelligence, financial flows, communications intercepts, satellite imagery — and no way to integrate it fast enough to act. The volume was the enemy. That’s what drove me to applied mathematics and the intersection of complexity science and artificial intelligence: not because I was fascinated by algorithms (though I am — ask anyone who’s made the mistake of sitting next to me at dinner), but because the data volume from counterterrorism operations demanded it.

The same structural problem now confronts every enterprise. The CEO of a consumer goods company managing demand signals across six continents faces a version of what I faced in 2001: too much data, too many interdependencies, too little time, and frameworks built for a simpler world that no longer hold.

VUCA told leaders to be adaptive. Useful, but vague. BANI tells leaders to be empathetic and improvisational. Also useful. Also vague. What neither framework offers is a systems-level engineering specification for how to build organizations, technologies, and governance structures that can actually function under the conditions both frameworks describe.

The Profitable Equilibrium of Chaos

Here’s the part no one wants to say out loud: BANI is not just an unfortunate side effect of technological progress. It’s a profitable equilibrium.

Platforms monetize anxiety. Consultancies thrive on permanent crisis. And opacity? Opacity feeds authoritarian appeal. If the world is incomprehensible, the strongman who claims to have simple answers gains traction precisely because the complexity exhausts everyone else.

Erich Fromm diagnosed this in Escape from Freedom back in 1941. Individuals flee from the isolating anxiety of modern freedom into authoritarian submission. That psychological pattern hasn’t changed. What’s changed is the scale — and the technology that amplifies it.

Brittleness and incomprehensibility are not bugs in the system. For some actors, they are features. A world that can’t understand itself is a world that can be governed by those who claim to understand it — even when they don’t. We’ve seen this movie before. Weimar Germany. The interwar period. The same structural conditions producing the same political consequences.

Breaking out of that equilibrium requires more than a new acronym. It requires building alternative systems that don’t reduce complexity but render it governable — through transparency, through mathematical rigor, through technology that shows its work.

The Explainer Gap

My friend and longtime collaborator Tom Barnett — the strategist behind The Pentagon’s New Map — and I have spent more than two decades watching a pattern that genuinely alarms us. In the mid-twentieth century, figures like Einstein, Keynes, von Neumann, Oppenheimer, Churchill, and FDR served as what we call “Explainers-in-Chief.” They took the most complex, terrifying developments of their era — quantum physics, economic collapse, nuclear weapons, totalitarianism — and translated them into shared understanding that allowed democratic societies to act coherently.

Where are those figures today?

Complexity has accelerated. The density of high-visibility explainers has declined. Today’s equivalents publish academic papers and give TED talks, but they lack the cultural penetration to create the kind of shared understanding that a democracy needs to govern itself through radical transformation.

That gap — the explainer gap — is not incidental to our current crisis. It’s a root cause. When the people who understand the systems can’t explain them, and the people who can explain things don’t understand the systems, the field is wide open for demagogues offering false clarity.

What Comes Next

I don’t have a bumper sticker for the answer yet, and frankly, I’m suspicious of anyone who does. But I’ve spent twenty-five years building pieces of what I think it looks like, and I want to think through it out loud.

Take brittleness. The response can’t just be personal toughness or “grit” — it has to be engineered resilience. Systems designed to absorb shocks, learn from them, and reconfigure without waiting for a human to intervene at every step. The principles behind that draw on decades of complexity science and applied mathematics — and they’re already working in practice, across industries.

Anxiety is the human signal that our systems have become too complex to trust without seeing how they think. Trusting intuition alone ignores the prompt. What people crave instead is radical transparency: AI and decision systems that show their reasoning, trace their logic from input to output, and submit to audit. The black-box era in AI is ending, and not because of some philosophical preference for openness. It’s ending because regulators are mandating it. The EU AI Act’s explainability provisions take full effect in August 2026. Every enterprise deploying high-risk AI systems will need to demonstrate traceability or face penalties up to 35 million euros. The market is about to discover that opacity is not just an ethical problem — it’s a commercial liability.

Non-linearity demands more than improvisation—fine for jazz, but brittle for systems under stress. It requires anticipatory capacity: modeling multiple futures and preparing for disruptions before they strike, instead of reacting to the wreckage. Most miss this key distinction: prediction extrapolates backward from what happened; anticipation projects forward, mapping what could happen. In a non-linear world, that difference is everything.

Incomprehensibility — arguably the scariest of Cascio’s four horsemen \— demands lucidity: a shared, culturally embedded commitment to clarity that slices through fog, resists manipulation, and prioritizes understanding over sloganeering. Lucidity is democracy’s lifeline in a world of trillion-parameter models and converging crises. Without it, we fulfill Fromm’s warning: an escape from freedom into the arms of anyone promising to make the confusion stop.

The Stakes

I’ve been working at the intersection of AI, complexity science, and national security since before most people knew what a neural network was. I started because the data volumes from post-9/11 intelligence operations demanded computational tools that didn’t yet exist. I continued because I discovered something I didn’t expect: the same mathematical frameworks that helped the national security community integrate chaotic data could help a consumer goods company optimize a global supply chain — or help a government anticipate a pandemic’s second-order effects before they cascaded.

That connection isn’t accidental. It reflects a deeper truth: the problems of the 21st century are not sector-specific. They are structural. And the tools to address them must be structural too.

Next time, I want to dig into the first of those capacities — and it’s one that most AI systems are currently designed to prevent: the ability to break and recover. Not as a metaphor, but as a measurable, diagnosable, engineerable property of a system. The field of resilience engineering is older than most people think, and its lessons are more urgent than most leaders realize.

VUCA is dead. BANI named the body. The question now is what we build on the grave.

Stephen F. DeAngelis is CEO of Enterra Solutions and co-founder of Massive Dynamics. He has served as a Visiting Professional Executive at Princeton University, a Visiting Scientist at Carnegie Mellon’s Software Engineering Institute and Oak Ridge National Laboratory, and a collaborator with MIT CSAIL.

The Wide Variety of Cyber Vulnerabilities

Supply chain risk managers would be delighted if they could focus on a single area of cybersecurity. Unfortunately, that is not the case. The threats are coming from all directions. Below are a few of the supply chain areas witnessing cyberattacks.

Operational Technologies. Graham’s particular focus is on operational technologies. He notes, “Many organizations believe they are isolated, only to discover accessible endpoints, unpatched vulnerabilities, and devices with little or no authentication. The supply chain only amplifies this risk.” Trying to button-up all those loose ends is more difficult than most people imagine. Jess Smith, leader of the Prevention and Protection team at Pacific Northwest National Laboratory, observes, “It used to be that physical systems, such as the devices that open valves or turn on or off transformers, were distinct from traditional computer systems. That is no longer the case. There is no longer a line between these functions, and anything that is digital could be vulnerable to being hacked. We need to be vigilant about every single device in these incredibly complex networks.”[2]

Logistics. The staff at Supply Chain 24/7 reports, “A new report from Everstream Analytics finds that cyberattacks targeting logistics companies are expected to double in 2026, following several years of sharp growth. The research tracks incidents affecting carriers, ports, 3PLs, and other logistics providers and shows attacks are up nearly 1,000% since 2021. … Everstream found that these attacks are becoming more coordinated and harder to contain. In many cases, the compromised systems weren’t owned by the affected company at all, but by a third-party provider, leaving shippers and carriers with little control over the situation.”[3]

Fraud and Theft. Journalist Dev Kundaliya reports, “Cyber-enabled fraud has become one of the most pervasive digital threats facing governments, businesses and individuals worldwide, according to Global Cybersecurity Outlook 2026 report by the World Economic Forum (WEF). The report warns that cybercrime is expanding at an unprecedented pace, fueled by advances in AI, fragmented global politics, and growing weaknesses in supply chains.”[4] He adds, “Chief executives now rank cyber-enabled fraud as their top digital risk, overtaking ransomware. By contrast, chief information security officers continue to focus on ransomware attacks and the fragility of supply chains.”

What Can Be Done?

Defending against cyberattacks is difficult. The WEF report notes, “The digital supply chain is highly interconnected, with dependencies within and across industries that are often not clearly mapped. A breach or disruption of one supplier can cascade through the entire ecosystem, affecting production, operations and even other suppliers or customers. This complexity makes it difficult to assess and manage cyber risk effectively. Attacks on widely used software or service providers can have global and systemic impacts.” The report adds, “Cyber risk is no longer a technical issue alone — it is a strategic, economic and societal concern that demands coordinated action across sectors and borders.” Enterprises around the world are asking, what can be done? Below are a few suggestions.

Collaborate. As the WEF report notes, today’s circumstances “demand coordinated action.” When that occurs, the report notes, “There are reasons for optimism.” It explains, “Organizations that embed resilience into leadership agendas, proactively manage supply chain and AI risks, and engage their broader ecosystems are better positioned to withstand shocks and adapt to uncertainty. … Building a secure digital future requires more than technical solutions. It calls for decisive leadership, shared accountability and a commitment to lifting the collective baseline — ensuring that resilience is accessible to all, not just the most well-resourced.”

Be Proactive. Hizmy Hassen, Chief Digital Officer at Apollo Tyres Ltd, asserts, “Complacency in supply chain cybersecurity could be your biggest risk.”[5] He explains, “No organization is too big or too well-resourced to be targeted. … Yet the latest State of Supply Chain Security report found that 94% of organizations are confident they could respond to a supply chain attack, and around a fifth believe they would not be affected if a key supplier was unable to operate for five days. Set against those case studies, too much confidence in cybersecurity looks less like resilience and more like wishful thinking.” Complacency can only be overcome through proactive programs. Hassen believes companies should start by “[mapping] out which partners are critical to which plants and customers, which systems are connected, and what data flows in each direction. If you model the impact of a key supplier being offline for a week, the single points of failure reveal themselves very quickly. This informs how you plan and rehearse your response with the most critical suppliers. Manufacturers should increase investment in independent external assessments of those partners, including targeted penetration tests and ‘white-hat’ exercises. At a time when cyber insurance premiums are rising, being able to evidence that kind of proactive testing can lead to more favorable conversations with insurers.”

Let AI Help. As noted at the beginning of this article, the benefits of artificial intelligence in supply chain operations have been clearly proven. At the same time, we have seen the first cyberattacks carried out by AI agents. When it comes to cybersecurity, companies must fight fire with fire. Paolo Dal Cin global lead at Accenture Cybersecurity, notes, “The weaponization of AI, persistent geopolitical friction, and systemic supply chain risks are upending traditional cyber defenses. For C-suite leaders, the imperative is clear; they must pivot from traditional cyber protection to cyber defense powered by advanced and agentic AI to be resilient against AI-driven threat actors.”[6]

Concluding Thoughts

The WEF report on cybersecurity makes an interesting point. It states, “In 2026, geopolitics remains the top factor influencing overall cyber risk mitigation strategies. Some 64% of organizations are accounting for geopolitically motivated cyberattacks — such as disruption of critical infrastructure or espionage.” Many of those attacks originate in China, Russia, and Iran. Graham adds, “Threat actors are exploiting weaknesses in global supply chains. Whether motivated by financial gain, political objectives, or disruption, they are targeting the very systems we rely on to keep lights on, shelves stocked, and public services running.” The Supply Chain 24/7 staff bluntly states, “The message is clear. Cyber risk in logistics [or elsewhere in the supply chain] is no longer just an IT concern or a rare disruption. It’s becoming a regular supply chain challenge, and one that can move faster than weather events, labor issues, or equipment breakdowns.” Supply chains without a good cybersecurity program in place will remain vulnerable to disruption and collapse.

Footnotes

[1] Mark Graham, “Supply chains are the overlooked risk in industrial cybersecurity,” Computer Weekly, 17 November 2025.

[2] Pacific Northwest National Laboratory, “Scientists address risks to supply chain in a connected world,” EurekAlert!, 28 October 2024.

[3] Staff, “Cyberattacks on Logistics Are Set to Double in 2026, Report Finds,” Supply Chain 24/7, 12 January 2026.

[4] Dev Kundaliya, “Cyber fraud and geopolitics reshaping global threat landscape, warns WEF,” Computing, 13 January 2026.

[5] Hizmy Hassen, “Complacency in supply chain cybersecurity could be your biggest risk,” Computing, 11 December 2025.

[6] Emma Woollacott, “Supply chain and AI security in the spotlight for cyber leaders in 2026,” IT Pro, 13 January 2026.

India’s Global AI Summit declared the end of the black-box era and called for AI transparency to be mandated as national infrastructure. Anthropic’s CEO told the Pentagon he won’t let his AI models be deployed without ethical constraints — and the Pentagon told him to comply or face consequences. And Forbes published a piece arguing that VUCA, the framework that has governed strategic thinking for 30 years, is no longer sufficient.

These aren’t three separate stories. They’re one story: the frameworks we built to manage complexity are breaking down at the same time the systems we built on top of them are becoming too opaque to govern.

I’ve been thinking about this convergence for a long time. Since September 12, 2001, to be precise — when I started building data integration and resilience systems for the governmental agency community and discovered that the same structural problem (too much complexity, too little explainability) was showing up everywhere: in defense, in supply chains, in financial systems, in democratic governance itself.

Over the next several weeks, I’m going to lay out an argument in five parts. It starts with why VUCA failed and what BANI — the framework that replaced it — gets right and where it stops short. It draws on what I’ve learned over twenty-five years of engineering resilience, explainability, and anticipatory systems for Fortune 500 companies and government agencies. And it ends with a question I think we’re all going to have to answer soon: are we going to build systems that render complexity governable, or are we going to keep managing the symptoms while the structure deteriorates underneath?

This isn’t abstract for me. I build these systems. I’ve watched them work — and I’ve watched what happens when organizations try to navigate a brittle, anxious, non-linear, incomprehensible world with tools designed for a merely volatile, uncertain, complex, and ambiguous one.

The first full essay drops next week. If the gap between how fast the world is breaking and how slowly our frameworks are adapting keeps you up at night — it keeps me up too.

Smart Cities and Smart Supply Charts

Journalist Susan Fourtané reports, “The world’s urban population is expected to rise to 70% by 2050. This means an increase from four billion to eight billion people living in urban areas using 75% of the planet’s natural resources. … The rapid increase in freight deliveries that result from the continuing growth of e-commerce and online shopping will lead to unsustainable traffic congestion, CO2 emissions, noise, and unhealthy air pollution levels within urban areas.”[3] She adds, “One of the biggest challenges that this increasing urbanization brings is how to provide for last-mile logistics. Smart city logistics proposes that logistics providers must leverage various innovations and technologies currently used in the digital transformation of the supply chain to find solutions for the challenges.”

Although technology is the link that binds smart city planning and urban logistics, there are many more variables that must be taken into consideration. Smart city initiatives often clash with the efficient movement of goods in urban environments (e.g., the closing of streets). It’s at this intersection of competing objectives that smart technologies and new thinking must play a role. Lily Xiang, Head of Brand at JUSDA International Supply Chain Co. Ltd, explains, “Smart cities transform urban landscapes with advanced technology and connectivity. Global logistics, essential for moving goods efficiently, intersects with these innovations. Understanding this intersection enhances supply chain efficiency.”[4] There is a lot of work to do. Rignesh Soni, Executive Vice President at Premium Parking, explains, “Urban logistics is at a crossroads, with the last-mile delivery segment facing increasing pressure from growing e-commerce demands. As cities become denser and more complex, traditional logistics models need help to keep pace, as they face inefficiencies, congestion, and environmental concerns.”[5] Inevitably, smart-city planners and urban supply chain professionals must solve problems together.

According to the staff at Seko Logistics, there are many benefits associated with smart city logistics.[6] When smart-city technologies are intertwined with every stage of the logistics process, they expect the following advantages to emerge:

• Faster delivery speed. “Data-driven urban logistics use real-time analytics to plan optimal delivery routes, anticipate roadblocks, and proactively prevent delays.”

• Lower-cost delivery solutions. “Smart transportation technologies lower operational expenses through route optimization and energy-efficient vehicles.”

• Minimized environmental impact. “Cities can cut emissions and promote cleaner air with sustainable city logistics systems.”

• Higher quality of life in urban areas. “Reduced congestion and efficient freight systems contribute to a better quality of life for city residents.”

Clearly, it would be a stretch to say that smart urban logistics is a reality in most urban areas. Nevertheless, without a vision there is no direction.

The Way Ahead

There are no silver bullet solutions for urban supply chains. Below are a few expert recommendations to improve smart-city logistics.

• Access to Technology. Xiang reports, “Many cities lack the necessary technology to support smart logistics. Inadequate internet connectivity and outdated systems pose significant challenges. Urban planners must invest in modernizing infrastructure to enable efficient logistics.” Once infrastructure is in place, it must be made widely available. The staff at Vintly insists, “Smart cities must ensure that the benefits of technology are accessible to all residents, businesses, and communities. This includes addressing issues such as the digital divide and ensuring that small businesses have access to the tools and resources needed to participate in smart city supply chains.”

• Smart Curbside Management. Soni reports, “Smart curbside management is a crucial innovation in this space. By deploying IoT sensors, cameras and AI-driven algorithms, cities can monitor and manage curbside usage in real time. This allows for better pricing and the allocation of dedicated loading zones, which reduces congestion and improves delivery times.”

• Smart Transportation Systems. The Seko Logistics staff writes, “Sustainable urban logistics requires new and innovative technologies to reach its full potential. Some examples include: Tracking enabled by IoT devices: This provides visibility into goods’ locations so providers can make quick adjustments as issues arise. Deliveries are more likely to show up on time, which contributes to a superior customer experience. AI-powered optimization: AI technology can analyze huge swaths of data to build efficient delivery routes, schedule freight at optimal times, and even predict demand patterns and prepare accordingly. Smart traffic management systems: These evaluate traffic flow and adjust signals accordingly. This helps to minimize congestion and, in high-demand periods, gives freight the right of way.”

• Smart Logistics Hubs. Xiang predicts that smart logistics hubs will emerge in most smart cities. She explains, “Smart logistics hubs emerge as a pivotal trend in smart cities. These hubs integrate advanced technologies to streamline operations. Urban areas benefit from enhanced efficiency and reduced congestion. Logistics hubs utilize IoT for real-time data exchange. This connectivity ensures seamless coordination among stakeholders. Companies achieve faster processing times and improved service delivery.” Brandon reports that the firm KPF Urban envisions these kinds of smart hubs “where automated cranes and robots would collect the cargo and distribute it to logistic centers scattered around the city. From there, goods would be delivered using a variety of micromobility options like electric bikes, unmanned aerial vehicles (UAVs), or drones.”

Concluding Thoughts

Soni insists, “As urbanization continues and e-commerce grows, the demand for efficient and sustainable logistics solutions will only intensify.” Efficiency, however, requires connectivity and connectivity often raises privacy issues. Xiang notes, “Public acceptance poses a social challenge for smart city logistics. Residents may resist changes brought by smart city initiatives. Concerns about privacy, surveillance, and data usage affect public perception. Authorities must engage communities in the planning process. Transparent communication and education can build trust and acceptance. Public support is crucial for the successful implementation of smart city projects.” Soni adds, “This means balancing the demands of logistics companies with the needs of local businesses and residents, creating a harmonious urban environment.” As most city planners and supply chain professionals know, smart urban logistics, as envisioned, remains aspirational. The Seko Logistics staff concludes, “The sustainable, connected cities of the future are only possible with smart urban logistics strategies. This demands an investment in green energy, AI, and IoT, as well as collaborative public-private partnerships.”

Footnotes

[1] Staff, “Smart Cities and Their Impact on Urban Supply Chains: Revolutionizing Urban Logistics,” Vintly Blog, 10 February 2025.

[2] Elissaveta M. Brandon, “NYC has a major delivery problem. These architects have a big vision to fix it,” Fast Company, 16 January 2026.

[3] Susan Fourtané, “Smart Cities Logistics & Implications for Supply Chains, EE Times, 8 May 2017.

[4] Lily Xiang, “The Impact of Smart Cities on Global Logistics Trends,” JUSDA Supply Chain Management International Co., 26 September 2024.

[5] Rignesh Soni, “How Smart Cities Are Transforming Last-Mile Delivery And Logistics,” Forbes, 30 September 2024.

[6] Staff, “Optimizing Urban Logistics with Smart City Infrastructure,” Seko Knowledge Hub, 2 December 2024.

Most experts agree that supply chains will encounter plenty of turbulent seas in the years ahead. Journalist David H. Coburn reports, “Experts say we can expect more frequent and less predictable challenges to stability from geopolitical upheaval, wars, extreme weather, supply‑and‑demand fluctuations — and even the occasional black swan event. As disruption becomes the new normal, companies are evolving new mindsets, strategies, and tactics to help them mitigate or even avoid the next shock.”[1]

Risk Managers Should Broaden Their Perspective

Political analyst Bruce Mehlman spends most of his time trying to make sense of current events. He writes, “I’ve spent the past [two years] trying to explain the volatile period we find ourselves in — the seismic technological, geopolitical and cultural macro-trends that brought us here, define our unsettled reality and portend a dynamic future.”[2] To help others “maintain [their] perspective, objectivity & sanity in [this] age of disruption,” he suggests people focus on six areas. They are:

1. Media. According to Mehlman, humans are prone to the same “garbage in, garbage out” flaws that plaque artificial intelligence. To counter that tendency, he suggests balancing your media diet. He explains, “If an unbalanced nutritional diet undermines your athletic performance, won’t an unbalanced information diet impair your effectiveness as a leader, investor, or citizen? We’re drowning in information sources and media options, making it very easy to fall into the confirmation bias trap — only consuming things that make us feel good. But that’s not balanced. Overcoming the algorithms & media business models takes thought and intentionality … aim for diversity in viewpoint, medium, geography, time frame, subject & format.”

2. History. Another bias people are prone to, according to Mehlman, is “recency bias.” To overcome that bias, he suggests people read a little history. He explains, “Studying history reminds us that we’ve always faced challenges that felt insurmountable at the time … and consistently made progress.”

3. Curiosity. I’ve written numerous articles about the importance of curiosity. The late Albert Einstein once wrote, “The important thing is to not stop questioning. Curiosity has its own reason for existence. One cannot help but be in awe when he contemplates the mysteries of eternity, of life, of the marvelous structure of reality. It is enough if one tries merely to comprehend a little of this mystery each day.”[3] Mehlman believes we should have the same curiosity about the people around us as we do about the world around us. He cites author Morgan Housel, who asks, “What have you experienced that I haven’t that makes you believe what you do? And would I think about the world like you do if I experienced what you have?” Mehlman writes, “I love those questions. … Too many of us presume the worst in others, convinced we understand the other side based on our own preconceived notions. We’re judgmental, not curious. And we’re frequently wrong … about who they are and what they believe.”

4. Success. Mehlman believes individuals need to reimagine what it means to be successful in life. He cites the late Clayton Christensen who discussed meeting people who were financially successful but nevertheless unhappy. Christensen wrote, “They didn’t keep the purpose of their lives front and center as they decided how to spend their time, talents, and energy.” Supply chain professionals should also take time to reimagine what success looks like for their company during turbulent times. What matters most? What hurdles stand in the way of success?

5. Long-term thinking. Many prominent business consultants have stressed the importance of playing the long game. Mehlman writes, “Success compounds over time. Shortcuts rarely do. Great businesses, investors and families play the long game … seeking slow & steady progress while avoiding big mistakes.”

6. Be there. Mehlman argues that the most successful businesses and people are the ones that show up when people need them. He cites numerous surveys that show that people have a great distrust for government, big business, the healthcare system, etc. Nevertheless, they trust their representative, their employer, their doctor, and so forth. He writes, “We trust those things we experience in real life, first hand, offline. If you want to be trusted, you need to show up … to tell your own story.”

Mehlman concludes, “I remain bullish on the future, optimistic we will make it ‘through the night with a light from above.’ America always has. Admittedly my optimism is based on hope… but it’s a hope informed by history & human nature. As Morgan Housel put it, ‘The past wasn’t as good as you remember. The present isn’t as bad as you think. The future will be better than you anticipate.’”

Navigating the Turbulence

I would love to provide a simple chart of things companies need to do to steer successfully through a turbulent future. As Ralf W. Seifert, a Professor of Operations Management at IMD, and Richard Markoff, a supply chain researcher, explain, “Each industry will adapt in different ways and at different paces.”[4] If Mehlman’s observations teach us anything, they teach that the more you know the better off you will be. Seifert and Markoff agree. They suggest supply chain professionals expand their knowledge in five different ways: 1) Take advantage of internal company networks and experts; 2) Cultivate relationships with regional experts and local partners; 3) Subscribe to expert intelligence and analysis; 4) Build geopolitical visibility from end-to-end in your supply chain; and, 5) Use this intelligence as the basis for regular dialogue across the business. The fact of the matter is, today’s supply chains are so complicated that professionals will come to realize they need artificial intelligence to help them make sense of it all.

To help clients gather and analyze information in today’s complex business environment, Enterra Solutions^® has developed the Enterra System of Intelligence^®. This System merges cutting-edge analytical techniques with a business’ data and knowledge to Sense, Think, Act, and Learn^® on enterprise data to meet the changing needs of the market. Enterra’s System acts as central “brain” within an organization, ingesting diverse datasets, business logic and practices, and strategy, to uncover unique insights and generate autonomous recommendations across the enterprise at market speed. Insights and recommendations generated by the Enterra System of Intelligence are acted upon through deep integrations with an organization’s established systems of record and engagement, akin to how the brain (decision-making) and central nervous system (actions) interact within the human body. Enterra’s system uniquely learns the environmental reasons that recommendations are successful or not and persists that learning in its Ontologies and Generative AI knowledge bases to improve future insights and recommendations. The business application modules included in the Enterra System of Intelligence are:

• Enterra Consumer Insights Intelligence System™. This System allows clients to quantitatively uncover and logically understand the inter-relationships that lead to heightened consumer understanding, hyper-personalized product recommendations, and new product innovation.

• Enterra Revenue Growth Intelligence System™ (ERGIS™). ERGIS systemically performs holistic revenue growth optimization (including optimizing strategic and tactical pricing, trade promotion, trade architecture, price pack architecture, media mix, customer segmentation, and assortment).

• Enterra Demand and Supply Chain Intelligence System™. This System concurrently performs non-linear demand and supply planning optimization.

• Enterra Business WarGaming™. Business WarGaming enables organizations to leverage their data to make strategic decisions by anticipating the moves of their competitors and taking direct action to beat the competition, mitigate risk, navigate uncertainty, and maximize market opportunity. Part of Enterra Business WarGaming is the Enterra Global Insights and Decision Superiority System™ (EGIDS™) — powered by the Enterra Autonomous Decision Science™ platform — which can help business leaders rapidly explore a multitude of options and scenarios.

Concluding Thoughts

Chris McDivitt, the Global Lead for Autonomous Supply Chain at Accenture, insists, “In today’s world of constant supply chain shocks, autonomy is no longer just innovation, it’s a resilience strategy.”[5] Coburn adds, “The dream vision of AI — fully autonomous supply chains run by teams of digital agents — may be years away from becoming a widespread reality. However, some technology observers are convinced that autonomy will be the next frontier in building disruption‑proof supply chains.” I am not sure than any complex system can be entirely “disruption-proof”; however, I do believe that Autonomous Decision Science (ADS^®) will make supply chains more resilient to future disruptive events.

Footnotes

[1] David H. Coburn, “Building Disruption-Proof Supply Chains,” MHI Solutions, 18 September 2025.

[2] Bruce Mehlman, “Six-Chart Sunday (#100) – How to Navigate an Age of Disruption,” Bruce Mehlman’s Age of Disruption, 23 November 2025.

[3] Albert Einstein, “Old Man’s Advice to Youth: ‘Never Lose a Holy Curiosity.’,” Life, 2 May 1955.

[4] Ralf W. Seifert and Richard Markoff, “The geopoliticization of supply chains: How it started and where it’s going,” I by IMD, 3 November 2025.

[5] Coburn, op. cit.

What is Retrieval-Augmented Generation?

The staff at Amazon Web Services explains, “Retrieval-Augmented Generation is the process of optimizing the output of a large language model, so it references an authoritative knowledge base outside of its training data sources before generating a response. Large Language Models are trained on vast volumes of data and use billions of parameters to generate original output for tasks like answering questions, translating languages, and completing sentences. RAG extends the already powerful capabilities of LLMs to specific domains or an organization’s internal knowledge base, all without the need to retrain the model. It is a cost-effective approach to improving LLM output so it remains relevant, accurate, and useful in various contexts.”[2]

The staff at Google Cloud explains RAG in a similar manner. They write, “RAG is an AI framework that combines the strengths of traditional information retrieval systems (such as search and databases) with the capabilities of generative large language models. By combining your data and world knowledge with LLM language skills, grounded generation is more accurate, up-to-date, and relevant to your specific needs.”[3]

Merritt compares RAG to the legal staff supporting a judge. He writes, “Judges hear and decide cases based on their general understanding of the law. Sometimes a case — like a malpractice suit or a labor dispute — requires special expertise, so judges send court clerks to a law library looking for precedents and specific cases they can cite. Like a good judge, large language models can respond to a wide variety of human queries. But to deliver authoritative answers — grounded in specific court proceedings or similar ones — the model needs to be provided that information. The court clerk of AI is a process called retrieval-augmented generation, or RAG for short.” In other words, LLMs are most useful when they have context.

The Benefits of RAG

Journalist Steven Rosenbush observes that the public is often “bewildered” at the enthusiasm companies are showing for generative AI.[4] The public, he notes, only sees consumer-facing chatbots. What they don’t see, he writes, is “behind the scenes, a lot of companies are deploying more specialized, internal AI tools that tap their data — and that is where they are looking for the real payoff from AI.” How big is the movement towards RAG-aided generative AI. According to Sylvain Duranton, global leader of BCG X, “It’s massive.” He told Rosenbush, “Most of what we do [for large corporations] is RAG-based.”

The most obvious benefit of RAG is succinctly stated by the IBM staff, “RAG helps large language models deliver more relevant responses at a higher quality.”[5] They go on to note, “RAG empowers organizations to avoid high retraining costs when adapting generative AI models to domain-specific use cases. Enterprises can use RAG to complete gaps in a machine learning model’s knowledge base so it can provide better answers. The primary benefits of RAG include: Cost-efficient AI implementation and AI scaling; access to current domain-specific data; lower risk of AI hallucinations; increased user trust; expanded use cases; enhanced developer control and model maintenance; [and] greater data security.” The Google Cloud staff adds, “Providing ‘facts’ to the LLM as part of the input prompt can mitigate ‘gen AI hallucinations.’ The crux of this approach is ensuring that the most relevant facts are provided to the LLM, and that the LLM output is entirely grounded on those facts while also answering the user’s question and adhering to system instructions and safety constraints.”

RAG also enhances prompt engineering efforts (i.e., the process of asking questions and giving instructions to a large language model). Journalist Belle Lin explains, “Prompt engineering has emerged … as an essential new skill for employees, so that they can generate better text summaries, data analyses and email drafts from AI chatbots and other applications. It is also used as a way to provide general large language models with specific company information, so that it provides more tailored responses.”[6] The IBM staff notes, “RAG systems essentially enable users to query databases with conversational language.” Because RAG systems are user-friendly, they can be leveraged in a number of ways. The IBM staff explains, “The data-powered question-answering abilities of RAG systems have been applied across a range of use cases, including: Specialized chatbots and virtual assistants; research; content generation; market analysis and product development; knowledge engines; [and] recommendation services.”

Concluding Thoughts

Author and keynote speaker, Dean DeBiase, explains, “For many companies, LLMs are still the best choice for specific projects. For others, though, they can be expensive for businesses to run, as measured in dollars, energy, and computing resources. … I suspect there are emerging alternatives that will work better in certain instances — and my discussions with dozens of CEOs support that [prediction].” RAG has certainly made LLMs more useful. Although as DeBiase notes, other alternatives, like small language models, may achieve similar results at less cost. As I noted in a LinkedIn post, “There’s often a misconception that small language models (SLMs) are less effective than large language models (LLMs) — but the reality is, each serves different functions/needs based on the source data and information they use and the trustworthiness they bring.” RAG was designed to improve trust for users of LLMs.

Despite the many benefits of RAG, Varun Raj, a cloud and AI engineering executive, offers a word of caution. He writes, “Many organizations are discovering that retrieval is no longer a feature bolted onto model inference — it has become a foundational system dependency. Once AI systems are deployed to support decision-making, automate workflows or operate semi-autonomously, failures in retrieval propagate directly into business risk. Stale context, ungoverned access paths and poorly evaluated retrieval pipelines do not merely degrade answer quality; they undermine trust, compliance and operational reliability.”[8] He argues that freshness, governance, and evaluation of data must be built into systems employing RAG. He explains, “Freshness, governance and evaluation are not optional optimizations; they are prerequisites for deploying AI systems that operate reliably in real-world environments. As organizations push beyond experimental RAG deployments toward autonomous and decision-support systems, the architectural treatment of retrieval will increasingly determine success or failure.”

Footnotes

[1] Rick Merritt, “What Is Retrieval-Augmented Generation, aka RAG?” Nvidia Blog, 31 January 2025.

[2] Staff, “What is RAG (Retrieval-Augmented Generation)?” Amazon Web Services.

[3] Staff, “What is Retrieval-Augmented Generation (RAG)?” Google Cloud.

[4] Steven Rosenbush, “Companies Look Past Chatbots for AI Payoff,” The Wall Street Journal, 23 October 2024.

[5] Staff, “What is retrieval augmented generation (RAG)?” IBM.

[6] Belle Lin, “From RAGs to Vectors: How Businesses Are Customizing AI Models,” The Wall Street Journal, 21 May 2024.

[7] Dean DeBiase, “Why Small Language Models Are The Next Big Thing In AI,” Forbes, 25 November 2024.

[8] Varun Raj, “Enterprises are measuring the wrong part of RAG,” Venture Beat, 1 February 2026.

According to the staff at Random Acts of Kindness Foundation, the full story about how the movement got started has a few twists. They write, “We finally sat down and pulled together what we believe is the most accurate version about the history of Random Acts of Kindness Day, which occurs annually on February 17th. We have been a bit elusive in our answer because, frankly, we didn’t know the whole story. There have been various anecdotes written, but nothing comprehensive.”[2] Below is the summary version of how the movement began:

“The Random Acts of Kindness movement started more than 40 years ago in the San Francisco Bay Area. In 1982 Berkeley writer and activist Anne Herbert published the first known account of ‘Practice Random Acts of Kindness and Acts of Senseless Beauty’ in a CoEvolution Quarterly. After her article appeared, the kindness movement began to spread in surrounding communities. Fast forward to 1991 when a local woman noticed the phrase scrawled across a warehouse wall in her neighborhood. She shared the phrase with her husband, a then 7th grade teacher, who decided to share it with his students. One of the kids happened to be the daughter of a San Francisco Chronicle columnist, who then wrote about Anne Herbert and the phrase. The article was picked up nationally by Reader’s Digest and later reprinted by the editors of Conari Press, a small press in Berkeley, California. Inspired by the phrase and the people involved in the movement, the editors at Conari Press published a book highlighting stories of kindness. The book, aptly titled Random Acts of Kindness, was published in February 1993 and was immediately embraced by hundreds of thousands who helped continue the movement. Readers of the book and admirers of the phrase began creating local ‘Random Acts of Kindness Days’ in mid-1993. In February of 1995, the first national Random Acts of Kindness Day took place with participants coast to coast. As a gift to many grassroots organizations, Conari Press funded and facilitated the kickoff year. Random Acts of Kindness Foundation (RAK) was created in 1995 in the Bay Area to facilitate future celebrations, always taking place in February during Valentine’s Day week. RAK was purchased soon after by a private foundation and moved to Denver, Colorado where it is located today.”

The best thing about helping celebrate Random Acts of Kindness Day is that everyone — regardless of where they live, what they do, or how well off they are — can participate and feel better for having done so. The Save the Children staff suggests a few ways to perform random acts of kindness you might want to consider. They include complimenting a stranger; donating blood; visiting a senior home and delivering a surprise; donating anonymously to a charity (they would certainly welcome the gift); sending an encouraging email to a colleague; taking someone to your favorite place; praising a local business or service provider online; organizing a community cleanup; writing positive messages on sticky notes and leaving them for others to find; and volunteering at a local charity, homeless shelter, animal shelter, or other non-profit organization.

The Random Acts of Kindness Foundation hopes that the emotional lift you get from performing a simple act of kindness for another person will inspire you to make acting kindly a normal part of your day throughout the year. That’s why the foundation’s motto is “make kindness the norm.” Writer Cassidy Rush reminds us, “We all know the feeling of a bad day turning around because of a small gesture. Maybe a stranger held the door for you, or a coworker brought you a coffee when you were stressed. These moments remind us that we are connected. Random Acts of Kindness Day is a special time dedicated to celebrating these moments. It encourages us to step out of our daily routine and do something good for others without expecting anything in return.”[3]

There are few things in this world that cost you nothing but a little effort yet bring you immense returns on your investment. Acting kindly is one of those things. Rush concludes, “Kindness is like a muscle. The more we use it, the stronger it gets. By making small, intentional choices to be generous and patient today, you help create the kind of world we all want to live in. So, look around you. Who needs a smile, a hand, or a kind word? You might be surprised at how much power you have to brighten someone’s day.” We all know the idiom “actions speak louder than words.” If you want your children to grow up to become kind and compassionate individuals, show them how it is done.

Footnotes

[1] Staff, “Celebrating Random Acts of Kindness Day 2026,” Save the Children.

[2] Staff, “The History of Random Acts of Kindness Day - February 17th,” Random Acts of Kindness Foundation.

[3] Cassidy Rush, “Random Acts of Kindness Day 2026: Simple Ways to Make a Difference,” Remitly, 12 January 2026.

• “The Genesis Mission charges the Secretary of Energy with leveraging National Laboratories to unite America’s brightest minds, most powerful computers, and vast scientific data into one cooperative system for research.”

• “The Order directs the Department of Energy to create a closed-loop AI experimentation platform that integrates the Nation’s world-class supercomputers and unique data assets to generate scientific foundation models and power robotic laboratories.”

• “The Order instructs the Assistant to the President for Science and Technology (APST) to coordinate the national initiative and the integration of data and infrastructure from across the Federal government.”

The Executive Order also directs the Secretary of Energy, APST, and the Special Advisor for AI & Crypto to “collaborate with academia and private-sector innovators to support and enhance the Genesis Mission.” The Order also prioritizes Genesis Mission efforts to focus on “the greatest scientific challenges of our time that can dramatically improve our Nation’s national, economic, and health security, including biotechnology, critical materials, nuclear fission and fusion energy, space exploration, quantum information science, and semiconductors and microelectronics.”

A Beginning

The Oxford Languages Dictionary defines the word “genesis” as “the origin or mode of formation of something.” The administration is hoping that the Genesis Mission is the beginning of a new scientific method of discovery — a method that keeps the United States in the forefront of new advances and discovery. AI and other advanced technologies reside at the heart of this new approach. Technologist Chuck Brooks insists, “Today’s world is characterized by exponential changes in technology.”[2] He views the Genesis Mission as “not merely another government program.” He writes, “It represents a bold strategic move that aligns with my belief that science, data, and computing should be regarded as essential components of our national strength rather than optional extras. … The real world and the digital world have merged. Deep-learning AI, quantum computing, and networked sensors were once only concepts in science fiction. They are real and strong.”

According to Michael Kratsios, head of the White House Office of Science and Technology Policy, scientific breakthroughs using AI require access to federal datasets that can “massively accelerate the rate of scientific breakthrough.”[3] Using that data, Kratsios says the Genesis Mission will be able to “automate experiment design, accelerate simulation and generate predictive models for everything from protein folding to fusion plasma dynamics. This will shorten discovery timelines from years to days or even hours.” Government support for basic scientific research has been declining for years. Hopefully, the Genesis Mission will spark a new wave of research. Energy Secretary Chris Wright insists government efforts will enhance private-sector investment in AI and focus more “on scientific discovery and engineering advancements.”[4] He adds, “To do that, you need the data sets that are contained across our national labs.”

Traditionally, government research support has been directed more at academia than the private-sector. This appears to be changing. As part of the Genesis Mission, “Administration officials said AI companies will get access to scientific data sets held by the government, which operates 17 National Laboratories through the Department of Energy. The Labs conduct cutting-edge research and have some of the world’s most advanced supercomputers.”[5] Journalist Zac Anderson notes, “[This approach] comes with concerns about accessing copyrighted or sensitive material that could have national security implications.”[6] To ease these concerns, Administration officials indicate, “Controls will be put in place to protect such data.” Journalist Ali Azhar reports, “The idea is to connect massive, and often underused, scientific datasets across agencies like the DOE, NIH, and NOAA to national lab supercomputers and wrap it all into an AI experimentation platform that supports scientific discovery. The policy outlines key players: public research agencies, academic institutions, and hand-picked private partners.”[7] In December, the Department of Energy announced it had signed collaboration agreements with two dozen organizations. They are: Accenture; AMD; Anthropic; Armada; Amazon Web Services; Cerebras; CoreWeave; Dell; DrivenData; Google; Groq; Hewlett Packard Enterprise; IBM; Intel; Microsoft; NVIDIA; OpenAI; Oracle; Periodic Labs; Palantir; Project Prometheus; Radical AI; xAI; and XPRIZE.

Azhar is both encouraged by this approach and concerned that the gap between research groups favored by the government and those not in favor could grow. He explains, “In theory, this could create a long-term framework where machine learning, high-end simulations, and domain-specific data all operate alongside each other. It could speed up discovery in energy, healthcare, climate, and more. However, it also raises some concerns. As AI models and compute power become central to modern research, what happens to institutions that do not have access to either? That gap could get wider (and very fast).” Where the Genesis Mission is headed will only become clear once it transforms from an idea into a reality.

Concluding Thoughts

Technology journalist Brian Buntz notes advanced research finds itself at a critical moment in America’s history. He reports, “While the U.S. remains the world’s research superpower ... China is quickly catching up and could be the world’s top R&D spender by 2030.”[8] Remaining a research superpower is an imperative. The Executive Order notes, “From the founding of our Republic, scientific discovery and technological innovation have driven American progress and prosperity. Today, America is in a race for global technology dominance in the development of artificial intelligence, an important frontier of scientific discovery and economic growth.” As long as it is executed well, Brooks harbors great hopes for the Genesis Mission. He writes, “I have consistently stated that new technology without risk management is a bad idea. Technologies don’t just bring hope; they also make things less safe. The focus of the Genesis Mission on cybersecurity, data protection, supply chain resilience, and governance shows that its creators recognize the significance of these elements. This technology is not an afterthought; it is a crucial part of the plan. If executed well, the Genesis Mission can produce a cascading set of positive outcomes.”

Footnotes

[1] Staff, “Fact Sheet: President Donald J. Trump Unveils the Genesis Mission to Accelerate AI for Scientific Discovery,” The White House, 24 November 2025.

[2] Chuck Brooks, “The Genesis Mission To Harness AI For Innovation And Security,” Forbes, 25 November 2025.

[3] David Shepardson, “Trump aims to boost AI innovation, build platform to harness government data,” Reuters, 24 November 2025.

[4] Ibid.

[5] Zac Anderson, “Trump signs order harnessing federal resources for AI boom,” USA Today, 24 November 2025.

[6] Ibid.

[7] Ali Azhar, “What Trump’s ‘Genesis Mission’ Really Means for Data and AI-Powered Science,” HPCwire, 25 November 2025.

[8] Brian Buntz, “NSF layoffs in 2025: Deep budget cuts headed for U.S. research sector,” R&D World, 15 February 2025.

The Future of Supply Chain Orchestration

The need for supply chain orchestration is real. Dominik Metzger, President and Chief Product Officer at SAP Supply Chain Management, explains, “In an era defined by geopolitical instability, tariffs, inflation, climate pressures, and rapid technological change, traditional supply chains — built on rigid, linear models — are ill-equipped to meet the flexibility and responsiveness that today’s fast-paced markets demand.”[3] He adds, “To thrive in this environment, enterprises must evolve from reactive management to proactive orchestration, powered by artificial intelligence.”

Metzger agrees with the DHL Supply Chain staff that digitalization is the foundation upon which orchestration builds. “While digitalization increases visibility and resilience,” he writes, “it focuses heavily on fixing foundational data challenges, such as data completeness and correctness and using this data for critical business decision-making.” And, like the DHL staff, he argues that businesses must take the next step forward. He explains, “Leading organizations are going further, by leveraging AI to automate decisions entirely, which enables the orchestration of supply chains by integrating applications, data and new automation technologies into truly agile operations.”

Metzger goes on to explain, “Applications trained on industry-specific business data, such as AI agents, can detect threats, analyze their potential impact, recommend mitigation strategies, and even execute responses before disruptions occur. This proactive capability pushes supply chain operations beyond basic digitalization, empowering teams to make faster, smarter decisions.” Gilmore writes, “OK, that sounds pretty good.” He then asks the really big question, “How do we get there?”[4]

Getting There

For the past several years, Enterra Solutions^® has focused on advancing Enterra Autonomous Decision Science^® (ADS^®) to help clients reach the next level of supply chain optimization. The culmination of this effort is a set of solutions we call the Enterra System of Intelligence^®. This System ushers in a new era of AI-enabled management science by merging cutting-edge analytical techniques with a business’ data and knowledge to Sense, Think, Act, and Learn^® on enterprise data to meet the changing needs of the market. Enterra’s System acts as central “brain” within an organization, ingesting diverse datasets, business logic and practices, and strategy, to uncover unique insights and generate autonomous recommendations across the enterprise at market speed. Insights and recommendations generated by the Enterra System of Intelligence are acted upon through deep integrations with an organization’s established systems of record and engagement, akin to how the brain (decision-making) and central nervous system (actions) interact within the human body. Enterra’s system uniquely learns the environmental reasons that recommendations are successful or not and persists that learning in its Ontologies and Generative AI knowledge bases to improve future insights and recommendations. The business application modules included in the Enterra System of Intelligence are:

• Enterra Consumer Insights Intelligence System™. This System allows clients to quantitatively uncover and logically understand the inter-relationships that lead to heightened consumer understanding, hyper-personalized product recommendations, and new product innovation.

• Enterra Revenue Growth Intelligence System™ (ERGIS™). ERGIS systemically performs holistic revenue growth optimization (including optimizing strategic and tactical pricing, trade promotion, trade architecture, price pack architecture, media mix, customer segmentation, and assortment).

• Enterra Demand and Supply Chain Intelligence System™. This System concurrently performs non-linear demand and supply planning optimization.

• Enterra Business WarGaming™. Business WarGaming enables organizations to leverage their data to make strategic decisions by anticipating the moves of their competitors and taking direct action to beat the competition, mitigate risk, navigate uncertainty, and maximize market opportunity. Part of Enterra Business WarGaming is the Enterra Global Insights and Decision Superiority System™ (EGIDS™) — powered by the Enterra Autonomous Decision Science™ platform — which can help business leaders rapidly explore a multitude of options and scenarios.

The following video provides a glimpse of what is possible.

Video

This powerful capability uniquely enables end-to-end Value Chain Optimization and decision-making at scale. It allows clients to uncover and understand the inter-relationships that lead to innovative new product development and innovation, heightened consumer understanding and targeted marketing, revenue growth tactics, and intelligent demand and supply-chain planning.

Concluding Thoughts

The DHL staff concludes, “The true potential of orchestration will be realized when it is applied to the planning, coordination, and execution of all supply chain activities, including transportation, warehousing, packaging and inventory.” With so much focus being given to supply chain orchestration, it is little wonder that Metzger reports, “Advanced analytics and AI are the top technology investment priorities for supply chain leaders over the next three years.” Gilmore is convinced. He concludes, “The autonomous orchestrating supply chain is coming, with our supply chain offices increasingly crowded AI agents. It’s going to be a brave new world.”

Footnotes

[1] DHL Supply Chain Staff, “Orchestration: The future of supply chain,” Fast Company, 15 July 2024.

[2] Dan Gilmore, “Maestros Wanted to Orchestrate Supply Chains, Part 1,” Supply Chain Digest, 14 November 2025.

[3] Dominik Metzger, “Why autonomous orchestration is the next frontier in supply chain management,” World Economic Forum, 3 November 2025.

[4] Dan Gilmore, “Maestros Wanted to Orchestrate Supply Chains, Part 2,” Supply Chain Digest, 21 November 2025.

What is Agentic Commerce?

Laura Furlong, a Senior Marketing Manager at VGS, defines agentic commerce as “AI agents acting on behalf of an individual (usually a consumer) to make a purchase online.”[4] She continues, “In agentic commerce, agents generally use conversational AI models to help consumers make purchasing decisions and ultimately complete a purchase on their behalf based on a set of consumer-defined parameters. Essentially, agentic commerce is using AI commerce tools and platforms to enhance and streamline e-commerce operations, enhancing customer service experiences through machine learning, data analytics, and intelligent automation.”

Consumers have mixed views about their interactions with chatbots used by online websites over the past several years. They have often found chatbots helpful and almost as often found them frustrating. The Salesforce staff notes, “In contrast to copilots and chatbots that rely on human requests and struggle with complex or multi-step tasks, ecommerce agents offer a new level of sophistication. … Consumer-facing agents take chatbots and other AI shopping interactions to the next level. Unlike traditional chatbots, agents can reason, learn, and adapt.” They suggest that e-commerce agents possess five specific attributes that enable them to follow through on a task. Those attributes are:

1. A proper role (What job should they do?).

2. Access to the right data (What knowledge can they access?).

3. The capacity to carry out correct actions (What capabilities do they have?)

4. Constrained by appropriate guardrails (What shouldn’t they do?)

5. Be available on the profitable channels (Where do they work?)

The Salesforce staff concludes, “You can think of shopper agents as a digital concierge, providing personalized assistance by guiding product searches and offering tailored assistance on ecommerce sites.”

Is Agentic Commerce the Future of E-commerce?

The simple answer to that question is a qualified yes. Schumacher and her McKinsey colleagues explain, “The stakes are high. By 2030, the US B2C retail market alone could see up to $1 trillion in orchestrated revenue from agentic commerce, with global projections reaching as high as $3 trillion to $5 trillion, according to McKinsey research. This trend will have the breadth of impact of prior web and mobile-commerce revolutions, but it can move even faster since agents can traverse the same digital paths to purchase as humans, allowing them to ‘ride on the rails’ laid down by these prior revolutions.”

With all this excitement about agentic commerce, why did I qualify my “yes” answer about agentic commerce’s future? Aaron Cheris, a Partner at Bain & Company, reports that many consumers harbor serious doubts about letting AI agents do their shopping.[5] The PYMNTS staff notes, “95% of consumers have at least one concern about agentic commerce. Despite many consumers’ willingness to let AI complete purchases on their behalf, most still harbor concerns about agentic AI controlling commerce.”[6] Journalist Paresh Dave reports that most consumers prefer AI to assist them with their shopping rather than do it for them. He writes, “In one of the frankest comments on agentic shopping made by a top tech boss, Amazon CEO Andy Jassy recently criticized how agentic shopping currently works on other platforms. ‘I would say the customer experience is not good,’ Jassy said on an earnings call last month. ‘There’s no personalization. There’s no shopping history. The delivery estimates are frequently wrong. The prices are often wrong. We have got to find a way to make the customer experience better and have the right exchange of value.’”[7]

If agentic commerce is going to meet expectations, it must overcome this lack of consumer confidence. Most analysts believe this will happen; especially as AI natives (i.e., people born since the beginning of the AI era) become primary consumers. Cheris writes, “This is the moment to strengthen customer relationships, build trust, and make retailer-value obvious — to humans and to algorithms alike.”

Concluding Thoughts

There is enough smoke in the air about agentic commerce that retailers would be foolish think there isn’t a fire to which they must respond. The question is: How should retailers respond to agentic commerce? Boston Consulting Group (BCG) analysts, Mike Evans, Robert Derow, Mitch Krogman, and Lia Gregor, insist, “To win in the new normal, retailers must invest across three pillars: third-party agents; retailer-owned agents, and agentic foundations (i.e., AI and data platform; measurement; operating model & governance; and talent & skills).”[8]

The staff at Google Cloud suggests retailers need to take advantage of two distinct opportunities. They are: Owning the consumer experience end-to-end; and, owning the transaction, regardless of origin.[9] In order to own the consumer experience end-to-end, the staff recommends creating “a branded agentic experience that curates the entire journey, from discovery to loyalty.” To do that, they suggest: 1) guiding intelligent product discovery by helping agents proactively connect consumers with the right products and cross-sell or upsell opportunities; 2) enabling cross-retailer shopping through a branded environment, aiming to facilitate negotiation and personalization; and 3) building loyalty through personalized experiences based on preferences and purchase history. To own the transaction, regardless of origin, the Google Cloud staff recommends capturing the sale no matter where it originates — the company’s site, a consumer’s personal agent, or another platform. To accomplish this, the staff recommends: 1) participating in the broader agentic commerce network through agent-to-agent transactions; 2) meeting industry-leading standards for payments, checkout, and agent interoperability; and 3) making products available wherever purchasing decisions happen.

Evans and his BCG colleagues believe the window of opportunity for retailers will close fast. They explain, “AI agents aren’t just another wave of digital innovation — they are fundamentally reshaping the retail landscape. While almost all global retailers are exploring agentic commerce, they are not moving fast enough. Those who move now by investing in agentic capabilities and building AI-native experiences will define the next era of customer engagement, brand relevance, and operational efficiency. Those who wait risk becoming invisible, disintermediated from the customers they once knew. The window to lead is rapidly closing, but for those bold enough to act, the upside is transformative.”

Footnotes

[1] Ben Fox Rubin, “What is agentic commerce? Your guide to AI-assisted retail,” Mastercard Stories, 4 September 2025.

[2] Staff, “What Is Agentic Commerce?” Salesforce.

[3] Katharina Schumacher and Roger Roberts with Katharina Giebel, “The agentic commerce opportunity: How AI agents are ushering in a new era for consumers and merchants,” Quantum Black, 17 October 2025.

[4] Laura Furlong, “Agentic Commerce: What you need to know,” VGS Blog, 13 May 2025.

[5] Aaron Cheris, “Agentic AI poised to disrupt retail, even with 50% of consumers cautious of fully autonomous purchases—Bain & Company,” Bain & Company Press Release, 13 November 2025.

[6] Staff, “Agents of Change: How Agentic AI Is Redefining Commerce,” PYMNTS, November 2025.

[7] Paresh Dave, “You Won’t Be Able to Offload Your Holiday Shopping to AI Agents Anytime Soon,” Wired, 14 November 2025.

[8] Mike Evans, Robert Derow, Mitch Krogman, and Lia Gregor, “Agentic Commerce is Redefining Retail—Here’s How to Respond,” Boston Consulting Group, 6 October 2025.

[9] Staff, “Agentic commerce is here: How retailers can prepare for the new shopping era,” Google Cloud, 7 October 2025.