Why the next mental-health transformation will require redesigning the systems around people, not simply expanding the systems that treat them

Trang Phan | AMOS — Absolute Operating System

The modern mental-health crisis is usually described as a healthcare problem, a workplace problem, a youth problem, a social-media problem or an individual resilience problem. Each description captures part of the picture, but none adequately explains the scale. Before COVID-19, approximately 970 million people—around one in eight globally—were living with a mental disorder, with anxiety and depressive disorders the most common. The first year of the pandemic was associated with an increase of more than 25% in anxiety and depressive disorders. In the workplace, WHO estimates that depression and anxiety contribute to roughly 12 billion lost working days every year and approximately US$1 trillion in lost productivity. In the United States, the CDC reported that 39.7% of high-school students experienced persistent feelings of sadness or hopelessness in 2023. These are different populations, different measurements and different forms of distress, and they should not be collapsed into a single causal explanation. But collectively they establish something strategically important: mental health is no longer a peripheral clinical issue. It has become a material constraint on human, organizational and economic performance. (World Health Organization)

The conventional response has been to expand the machinery downstream of distress. More therapy, more medication, more employee-assistance programs, more wellbeing applications, more resilience training, more diagnostic capacity, more mental-health benefits and increasingly more digital and AI-enabled interventions. Much of this is necessary, and for many people clinical treatment is essential. The mistake is not providing care. The mistake is assuming that expanding care is equivalent to correcting the conditions that contribute to distress. A society can become substantially better at treating people while simultaneously becoming worse at designing environments in which people can remain well. That distinction is becoming increasingly important because WHO itself identifies excessive workloads, low job control, job insecurity, discrimination and inequality among risks to mental health at work. Mental health therefore cannot be understood exclusively as something located inside the individual. It also reflects the interaction between the individual and the system in which that individual is expected to function. (World Health Organization)

This report approaches that problem through AMOS — the Absolute Operating System, created by Trang Phan. AMOS is a broader systems architecture for reasoning about complex human, organizational, technological and societal systems through deterministic constraints, dependencies, feedback, boundaries and consequences. One of its central ideas is deceptively simple: before attempting to optimize an outcome, identify the conditions the underlying system requires to remain viable. Applied to mental health, this changes the governing question. Instead of beginning with How do we make people more resilient to the environment?, it begins with What conditions are we repeatedly asking human beings to absorb, and are those conditions compatible with sustained biological regulation? This does not replace psychiatry, psychology, neuroscience, medicine or public health. It changes the unit of analysis from the isolated person to the coupled human–environment system.

That distinction matters because modern institutions have become extraordinarily sophisticated at optimizing everything surrounding the human being while retaining comparatively primitive assumptions about the human being inside the system. Capital is optimized. Supply chains are optimized. Inventory is optimized. Customer journeys are optimized. Software utilization is optimized. Labor is scheduled algorithmically. Communication is instantaneous. Performance is measured continuously. Artificial intelligence is beginning to compress work that previously required days into minutes. Yet the human nervous system remains constrained by recovery, sleep, predictability, social connection, perceived control, physical limits and finite attentional capacity. Technology has changed the velocity of the operating environment much faster than human biology has changed the capacity of the operator.

The strategic question is therefore larger than mental-health provision. It is whether modern institutions have unintentionally constructed an economy that consumes human regulatory capacity as an unpriced input. If that is occurring, rising expenditure on mental-health support can coexist with deteriorating population-level outcomes because treatment is operating downstream while load continues to accumulate upstream. In business terms, this would be analogous to continually increasing maintenance expenditure on machinery while refusing to examine whether the production system is operating the machinery outside its sustainable design envelope. The maintenance may be excellent. The machine may still keep failing.

1. Mental Health Has Become Economic Infrastructure

Mental health was historically treated by many organizations as a private health matter that became relevant to employers primarily when illness resulted in absence. That distinction is increasingly untenable. Mental health now influences labor availability, productivity, presenteeism, retention, leadership capacity, workplace safety, healthcare expenditure and institutional resilience. WHO estimates that 15% of working-age adults were living with a mental disorder in 2019, while depression and anxiety alone account for the equivalent of approximately 12 billion lost working days each year. Importantly, WHO explicitly recognizes that work can either protect or undermine mental health: good work can provide income, identity, purpose, relationships and inclusion, while poor working environments—including excessive workload, low control and insecurity—can create material mental-health risks. The workplace is therefore not simply where mental-health consequences become visible. It can be one of the environments shaping those consequences. (World Health Organization)

This changes the economics of the issue. Traditional accounting records labor as an input purchased for a defined period, but it does not adequately account for the biological state in which that labor is produced. Two organizations may employ the same number of people for the same number of contracted hours while extracting radically different levels of physiological and cognitive load. One may operate with predictable schedules, clear decision rights, sufficient staffing, protected recovery, high trust and meaningful employee agency. Another may produce identical nominal working hours through constant interruption, ambiguous accountability, continuous digital availability, volatile scheduling, understaffing and fear-based performance management. Conventional labor accounting can make these organizations look comparable. Biologically, they are not.

AMOS frames this as a systems-integrity problem. The output of a human system cannot be assessed independently from the rate at which the system consumes the capacity required to generate that output. A company that increases short-term productivity by progressively reducing recovery may record an efficiency improvement while accumulating what is effectively biological operating debt. The debt does not necessarily appear immediately. It emerges through error rates, disengagement, interpersonal conflict, absenteeism, health deterioration, unwanted attrition, reduced creativity and eventually loss of organizational capacity. The delay between extraction and consequence makes the model particularly dangerous because managers can observe rising output before they observe the degradation that produced it.

This is why the WHO/ILO estimate of approximately US$1 trillion in annual productivity losses should be read as more than a healthcare statistic. It is an indication that mental health has become part of the operating economics of the global labor system. The relevant question for business leaders is no longer whether organizations should care about employee wellbeing as an expression of corporate responsibility. It is whether an organization can sustainably operate when the biological infrastructure on which its strategy depends is degrading. (World Health Organization)

2. The Central Error Is Treating Human Capacity as Static

Most modern operating models implicitly assume that human capacity is approximately static: an employee has a certain level of capability, a role has a certain amount of work, technology improves productivity, and management determines how efficiently the two can be combined. What this model underweights is that human capacity is state-dependent. Attention changes under sleep deprivation. Judgment changes under sustained uncertainty. Cognitive flexibility changes under chronic threat. Social behavior changes under insecurity. Recovery changes what a person can sustainably produce tomorrow. The same individual therefore does not represent a constant unit of productive capacity across different operating environments.

AMOS treats this distinction as fundamental. Human performance should not be modeled simply as capability × time. It is better understood as capability expressed through a changing regulatory state. This means organizations can inadvertently destroy the very capacity they are attempting to optimize. Increasing workload can initially increase output because people compensate. Increasing urgency can initially accelerate decisions because attention narrows. Increasing surveillance can initially improve compliance because employees respond to evaluation. Reducing headcount can initially improve apparent productivity per employee because remaining staff absorb additional work. But none of those short-term responses proves that the resulting operating state is sustainable.

This is one reason burnout is so frequently misunderstood. Burnout is often addressed only after performance deteriorates, at which point the individual is offered leave, counseling, coaching, resilience training or another form of support. The organization observes the endpoint and intervenes there. A systems view asks what sequence preceded the endpoint: workload, control, uncertainty, recovery, social support, conflicting expectations, resource constraints and duration. It then asks whether the pattern was isolated to the person or repeated across the operating environment. When similar outcomes cluster around particular teams, roles, managers, work designs or periods of transformation, the system itself becomes part of the diagnostic field.

This does not mean every case of anxiety, depression or burnout is caused by work or by modern economic structures. Such a claim would exceed the evidence and erase biological, psychological, social and clinical differences between people. The business implication is narrower and stronger: environmental conditions are material variables in human functioning, and an operating model that excludes them is incomplete. WHO's own workplace guidance supports this multidimensional framing by identifying organizational conditions alongside individual support as legitimate intervention targets. (World Health Organization)

The management consequence is significant. Human-capital strategy cannot stop at recruitment, skills, incentives and performance. It must increasingly include regulatory sustainability: whether the organization is producing conditions under which people can repeatedly mobilize effort and subsequently recover it. The distinction between peak capacity and renewable capacity may become as important to human-intensive businesses as the distinction between gross revenue and recurring revenue is to investors.

3. The Modern Economy Has Industrialized Ambient Threat

Historically, many threats were acute and identifiable. Modern professional and economic environments increasingly generate another category: ambient threat. Nothing catastrophic needs to happen on a particular day. Instead, uncertainty remains unresolved in the background. A worker may simultaneously face restructuring risk, rising living costs, performance measurement, technological displacement, constant digital communication, changing organizational priorities and the expectation of continuous availability. Each variable may be manageable individually. The combined environment can produce persistent anticipatory load.

The importance of this pattern is that uncertainty itself consumes capacity. A system does not need to be continuously dangerous to remain difficult to regulate within; it only needs to remain continuously unresolved. Employees who do not know whether a restructuring affects them, gig workers who do not know what next month's income will be, junior professionals who do not know which unwritten expectations determine progression, and managers accountable for outcomes without sufficient resources all operate inside different versions of the same structural condition: high consequence combined with incomplete control.

Digital technology has intensified this problem because it removed many of the natural boundaries that once ended exposure. The office closed. The journey home created separation. Messages waited. Information arrived at a finite speed. Modern communication architecture allows work to remain psychologically active long after formal working hours end. Smartphones transformed availability from an exceptional condition into an ambient possibility. Collaboration tools increased organizational coordination while also increasing the number of channels through which attention can be claimed. Remote work removed commuting for millions of people but, in some contexts, also weakened the physical boundary separating professional and private life. Artificial intelligence may now accelerate this transition again by raising expectations about how much cognitive work can be produced in a given period.

The danger is not technology itself. The danger is velocity without corresponding regulatory redesign. Every productivity technology creates the possibility of capturing its benefit in at least two ways: the organization can produce the same output with less human load, or it can convert the efficiency gain into a higher output expectation. If the second response dominates repeatedly, productivity technology does not necessarily create recovery. It can simply increase the speed of the treadmill.

For sectors such as professional services, finance, technology, healthcare and logistics, this becomes a strategic design question. AI could reduce the cognitive burden associated with documentation, research, analysis and administrative work. Alternatively, organizations could use AI to increase utilization expectations, compress deadlines and expand the number of simultaneous responsibilities assigned to each person. The same technology therefore has two fundamentally different biological consequences. Automation can return capacity to the human system, or it can become another mechanism for extracting more capacity from it.

4. Recovery Has Been Reclassified From Requirement to Reward

Perhaps the most consequential change in modern work is the cultural treatment of recovery. Recovery is often presented as something earned after sufficient productivity rather than as one of the inputs required to produce sustainable productivity. Rest therefore sits outside the operating model. Organizations model staffing, utilization, project demand, revenue and deadlines in detail while assuming that recovery will occur somewhere in the remaining space.

This is structurally similar to running an industrial asset without explicitly scheduling maintenance. The system may continue functioning for a considerable period, particularly when individual components compensate for degradation. Human systems are remarkably adaptive. People work later. Managers cover vacancies. Parents sacrifice sleep. Teams absorb additional projects. High performers carry weaker parts of the organization. None of this necessarily produces immediate system failure. In fact, the ability of people to compensate can conceal poor system design for years.

The difficulty is that compensation is frequently mistaken for capacity. Because the work was completed, the workload is considered achievable. Because the deadline was met, the timeline is considered reasonable. Because the employee continued performing, the operating model is considered functional. This is a profound measurement error. Observed survival under load does not establish sustainable capacity under load.

The distinction matters particularly in industries built around utilization. Consulting, investment banking, law, medicine, technology and other high-intensity professions often create enormous economic value from highly trained human cognition. Yet the utilization logic applied to those people can resemble asset utilization: unused capacity appears inefficient. From a biological perspective, however, some apparent unused capacity is precisely what creates resilience. Slack allows recovery, learning, adaptation, mentoring, error correction and response to unexpected demand. A system operating continuously near maximum capacity may therefore appear efficient while becoming progressively more fragile.

This principle extends far beyond elite professional work. In logistics, insufficient slack appears when demand spikes. In healthcare, it appears when patient surges collide with already stretched staffing. In retail and hospitality, it appears through volatile scheduling and understaffing. In manufacturing, it appears where labor intensity increases without adequate ergonomic or recovery design. Different industries generate different forms of load, but the systems principle is the same: capacity that is never allowed to regenerate eventually ceases to be capacity.

5. Mental Health Became a Market Because Downstream Problems Are Easier to Productize

Once distress is individualized, it becomes economically legible. A person can receive a diagnosis, appointment, prescription, coaching program, wellness subscription, digital intervention, insurance claim or workplace accommodation. Each intervention can be purchased, delivered, measured and administered. The system therefore develops a market around the distressed individual because the individual is an addressable unit.

Environmental causation is harder. No single application can eliminate job insecurity. No therapy platform can redesign an understaffed hospital. No mindfulness program can create affordable housing. No employee-assistance program can resolve contradictory performance incentives. No antidepressant can change an abusive manager. No resilience course can manufacture genuine control over a volatile schedule. Clinical interventions may help people enormously in coping with, recovering from or treating mental-health conditions; the point is not that they are ineffective. It is that the intervention and the cause do not necessarily occupy the same level of the system.

This creates a powerful economic asymmetry. Downstream care is modular. Structural redesign is political. Treatment can often be purchased from a vendor. Redesign requires decisions about workload, staffing, compensation, management quality, scheduling, autonomy, technology, performance systems and ultimately the distribution of power. Organizations therefore have a natural incentive to purchase interventions that leave the operating architecture intact.

The mental-health market can consequently expand while underlying sources of load remain largely unchanged. This does not require conspiracy or bad intent. It is a predictable consequence of institutional incentives. Organizations generally prefer interventions that are measurable, bounded, delegable and minimally disruptive to the production model. A wellbeing platform fits those criteria. Reconsidering whether the business systematically depends on chronic overextension does not.

The market therefore performs an important but incomplete function. It creates more ways to help individuals after distress becomes visible. What it does not automatically create is a mechanism through which aggregated distress forces redesign of the system producing it.

That missing feedback loop may be the central structural weakness of the current model.

6. Care Becomes Dangerous Only When It Prevents the System From Learning

Care is indispensable. People experiencing mental-health conditions deserve access to appropriate professional support, and structural explanations must never become an excuse for withholding individual treatment. But from a systems perspective, treatment should perform two functions simultaneously: restore the individual where possible and generate information about the environment in which impairment emerged.

Modern institutions are considerably better at the first than the second.

Imagine a business in which one employee from a division takes stress leave. The case may be individual. If twenty employees from the same operating unit develop similar problems, the information content changes. Yet confidentiality requirements and organizational silos often mean each case continues to be processed separately. HR sees absence. Healthcare providers see patients. Managers see vacancies. Finance sees productivity effects. Insurers see claims. Executives see aggregate engagement data. The underlying pattern may exist across all of these datasets without any single owner being responsible for reconstructing it.

AMOS treats this as a feedback failure. A viable system must be capable not merely of correcting local failure but of learning when repeated local failures indicate a higher-level design defect. If the same type of breakdown repeatedly occurs under the same conditions and every breakdown is treated independently, the organization can become extraordinarily efficient at repair while remaining incapable of prevention.

The analogy with industrial safety is instructive. Mature safety systems do not simply treat injured workers and return them to the same machine. They investigate incidents, near misses, common mechanisms, equipment design, procedures, incentives and environmental conditions. Aviation does not conceptualize every accident as a pilot wellbeing problem. Nuclear engineering does not respond to repeated anomalies exclusively by training operators to become more resilient. High-reliability industries learned that human failure can be information about system design.

Mental health has not yet fully made that transition.

A mature organizational model would therefore aggregate anonymized signals across absence, turnover, overtime, schedule volatility, engagement, manager quality, workload, safety events and employee support utilization to identify whether specific operating conditions repeatedly precede deterioration. The objective would not be to diagnose individuals from organizational data. It would be to diagnose the organization.

7. High Prevalence Changes the Strategic Question

The prevalence numbers are important not because they establish one universal cause, but because they challenge the assumption that mental health can remain a specialist issue at the margin of management.

In the CDC's 2023 Youth Risk Behavior Survey, 39.7% of U.S. high-school students reported persistent feelings of sadness or hopelessness, while 20.4% seriously considered attempting suicide and 9.5% reported an attempt. These outcomes vary significantly by demographic group and require careful interpretation; they cannot simply be attributed to economic design. Nevertheless, when distress indicators occur at this scale, youth mental health becomes an education, family, community, technology and public-policy issue as well as a clinical one. (CDC)

The same systems logic applies to work. If one employee cannot tolerate a role, individual fit may be the primary explanation. If an entire occupational category repeatedly exhibits exhaustion, turnover and psychological strain, organizational and sector design become legitimate explanatory variables. The inference does not need to be ideological. It is simply how systems diagnosis works: as prevalence rises, shared exposures deserve increasing analytical weight.

This matters for corporate benchmarking. Organizations often benchmark compensation, productivity, headcount, spans of control, digital adoption and employee engagement against peers. Far fewer benchmark what could be called human-system sustainability: schedule predictability, recovery opportunity, after-hours intrusion, perceived control, managerial volatility, concentration of chronic overtime, staffing resilience, psychological safety and the distribution of load across teams.

The omission means organizations may benchmark themselves into collective dysfunction. If every competitor operates with unsustainable hours, being average does not make the practice sustainable. If an industry normalizes permanent availability, sector benchmarking simply reproduces the same biological externality across companies. Relative performance is not equivalent to absolute viability.

AMOS introduces a different benchmark: does the system preserve the capacity on which its future performance depends?

That is a much harder standard.

It is also a strategically superior one.

8. The Five Variables Organizations Should Begin Measuring

A biologically compatible operating model requires a different management dashboard. It does not replace clinical measures, engagement surveys or traditional workforce analytics. It adds upstream variables capable of identifying deterioration before it becomes illness, absence or attrition.

The first is predictability. Human systems can tolerate substantial demand when they can anticipate it, prepare for it and understand when it will end. Uncertainty changes the nature of the same workload. Organizations should therefore measure not only hours worked but schedule volatility, priority changes, deadline instability, role ambiguity and the frequency with which employees are forced to replan their lives around unexpected organizational demands. The relevant variable is not simply load; it is load variance.

The second is recovery capacity. Organizations typically measure vacation entitlement rather than whether recovery actually occurs. A more meaningful model would examine consecutive high-load periods, after-hours communication, interruption during leave, recovery time following peak demand and whether teams possess enough spare capacity for workload to fall after a surge. A nominal benefit that employees cannot safely use is not a functional recovery mechanism.

The third is agency. Employee autonomy is often discussed culturally, but its biological importance is deeper. A person who can influence sequence, pace, method and boundaries experiences the same demand differently from someone who experiences identical demand without control. The critical measurement is therefore not whether organizations claim employees have flexibility, but whether people can exercise meaningful refusal or adjustment without hidden career, financial or social penalties.

The fourth is social buffering. Workplaces are social systems as much as production systems. Relationships with colleagues and managers can absorb uncertainty, distribute load, provide information and reduce isolation. Conversely, low trust, internal competition, bullying and weak management can amplify threat. WHO explicitly recognizes positive relationships and inclusion among the ways good work can support mental health. (World Health Organization)

The fifth is sustainable load. Most organizations know how much work they need performed but have a much weaker model of how much load a given organizational configuration can absorb repeatedly. Sustainable load must incorporate intensity, duration, complexity, uncertainty, emotional demand, interruption, recovery and staffing resilience. This is not a universal physiological formula. It is a management discipline: stop treating every completed workload as evidence that the workload was sustainable.

Taken together, these variables shift the conversation from “Are employees well?” to “Does our operating system repeatedly create conditions in which people can remain functional?”

That is a far more actionable question.

9. The Business Case Is Larger Than Absenteeism

The conventional financial case for mental health usually begins with absenteeism and lost productivity. Those costs are significant, but they capture only the most visible end of the system. The larger value at risk lies in the degradation of decision quality, innovation, customer experience, leadership judgment, safety, institutional memory and adaptive capacity long before an employee formally leaves work.

A software engineer under sustained overload may continue shipping code while producing more technical debt. A clinician may continue treating patients while losing attentional margin. A relationship manager may continue serving clients while becoming less capable of emotional regulation. A senior executive may remain highly productive while increasingly relying on narrow, short-horizon decisions. A warehouse employee may remain present while reaction time and physical safety deteriorate. In each case, conventional attendance data records a functioning worker. The underlying system may already be losing quality.

This is why presenteeism is strategically important. WHO's broader economic framing explicitly recognizes that mental-health costs extend beyond simple absence and include impaired productivity and turnover. (Iris) The implication for management is that the cost curve begins before the medical event.

For knowledge businesses in particular, this is critical because value creation depends disproportionately on judgment rather than time. An exhausted professional can remain at a desk for twelve hours while the quality of the marginal hour approaches zero or becomes negative through errors requiring later correction. Yet utilization systems may still classify all twelve hours as productive effort.

The same principle applies at executive level. Organizations routinely conduct financial stress tests, cyber resilience exercises, liquidity scenarios and supply-chain simulations. Few conduct equivalent tests of human operating resilience. What happens if a transformation requires eighteen months of sustained additional effort? What happens if AI adoption coincides with restructuring anxiety? What happens if the highest-performing 10% of employees are simultaneously the group carrying the greatest hidden load? What happens if an organization achieves its transformation target but loses the people capable of operating the transformed business?

Those are not HR questions.

They are enterprise-risk questions.

10. AI Could Become Either the Largest Recovery Technology or the Largest Load Accelerator in Modern Work

Artificial intelligence makes the argument urgent because it is rapidly altering the relationship between human time and productive output. Generative AI can compress research, drafting, coding, analysis, customer support, administrative processing and knowledge retrieval. The economic opportunity is substantial. But whether those productivity gains improve human sustainability depends almost entirely on how organizations capture them.

Consider two companies implementing the same AI capability. Company A automates routine administrative work, reduces unnecessary meetings, protects the recovered time and redirects part of the productivity gain toward deeper work, development and reduced overload. Company B observes that employees can now produce 30% more and immediately raises targets by 30%, removes headcount and compresses deadlines. Both companies may record productivity improvements. Only one has increased the regenerative capacity of the human system.

The second company has effectively converted technological productivity into a new baseline expectation. When the next productivity technology arrives, the process repeats. Over time, the organization becomes more technologically capable while individual cognitive slack approaches zero.

AMOS describes this as a failure to distinguish efficiency from integrity. An optimization is not genuinely successful if it improves a local performance measure by degrading a load-bearing system required for future performance. This principle applies beyond mental health—to cybersecurity, financial leverage, supply-chain concentration and environmental degradation—but human capacity makes the trade-off particularly visible.

AI therefore creates an unusual strategic opportunity. For the first time in decades, organizations may be able to remove substantial amounts of low-value cognitive work at scale. The question is whether leaders will bank some of that dividend as human recovery capital, or whether every minute saved will immediately be resold to the organization as additional output.

The decision will shape the next era of work.

11. Healthcare Illustrates What Happens When Purpose and Operating Reality Diverge

Healthcare provides one of the clearest examples of the systems problem because the mission is explicitly human while the operating environment can place extraordinary load on the humans delivering it. Clinicians work in systems where errors carry significant consequences, demand can be unpredictable, emotional exposure is high, staffing can be constrained and administrative requirements consume increasing amounts of cognitive time.

The conventional response to clinician distress often includes wellbeing programs, counseling and resilience support. These interventions can be valuable. But if the underlying drivers include excessive administrative load, insufficient staffing, unstable scheduling, low control and chronic capacity pressure, individual support cannot substitute for operating-model correction.

The same applies to education. A teacher can be trained in resilience, but resilience cannot reduce class size. A nurse can receive mindfulness training, but mindfulness cannot create an additional nurse on an understaffed shift. A call-center employee can access an assistance program, but that program cannot independently change algorithmically determined performance pressure. A junior banker can receive wellbeing resources while continuing to operate under unpredictable overnight demands.

The recurring pattern is important: individual interventions are strongest when the impairment is individual; they become progressively less sufficient as the cause moves upward into system design.

This does not require abandoning personal responsibility. People differ substantially in vulnerability, coping strategies, health, circumstances and preferences. Organizations cannot design away every source of distress, nor should normal difficulty be medicalized. The objective is instead to assign responsibility to the correct level. Individual problems require individual interventions. Management problems require management interventions. Structural problems require structural interventions.

A mature system knows the difference.

12. Mental Health Should Become a Feedback Signal for Organizational Design

The most important change is therefore conceptual. Mental-health data should not merely tell an organization how many people require support. Properly aggregated and protected, it should help reveal whether the organization itself requires redesign.

This creates a closed learning loop. A company changes workload allocation, scheduling or management practices. It then observes not merely productivity but turnover, absence, perceived control, recovery, safety and employee experience. If performance improves without deterioration elsewhere, the design may be superior. If output rises while biological and social indicators deteriorate, the apparent productivity improvement should be treated as provisional rather than celebrated immediately.

This is analogous to how sophisticated businesses already manage other complex systems. A bank does not judge a loan portfolio solely by current revenue; it monitors future credit risk. A manufacturer does not judge a plant solely by output; it monitors defects and maintenance. A cloud operator does not judge infrastructure solely by utilization; it maintains redundancy because 100% utilization is dangerous. An airline does not optimize solely for aircraft hours; it incorporates mandatory maintenance because an aircraft that never stops eventually stops involuntarily.

Human systems require the same maturity.

The irony is that many businesses understand this principle perfectly everywhere except labor. They maintain spare server capacity but remove spare human capacity. They create redundancy in data centers while celebrating organizations dependent on single critical employees. They schedule machine maintenance while allowing executive teams to operate continuously through transformation after transformation. They stress-test balance sheets while assuming that people will simply continue absorbing volatility.

That inconsistency is becoming increasingly expensive.

13. The AMOS Alternative: Design From Biological Constraints Outward

AMOS begins from a different premise: systems should be optimized only after their load-bearing constraints have been identified. In human systems, those constraints include physical, cognitive, emotional, social and temporal limits. They do not dictate a single organizational model. They establish boundaries within which different models can be tested.

At the individual level, the concern is whether a person retains sufficient capacity to regulate, recover, decide and act. At the organizational level, the concern is whether work design, management, technology, incentives and social conditions preserve that capacity across the workforce. At the societal level, the concern becomes whether housing, education, healthcare, employment, information systems, community structures and economic volatility collectively create conditions compatible with sustainable human functioning.

These levels interact. Financial insecurity at the societal level enters the workplace as employee stress. Poor management at the organizational level enters the household as depleted attention. Social isolation alters the capacity with which individuals arrive at work. Digital platforms cross all three levels. Clinical services often receive the downstream consequences generated elsewhere.

This is why mental health cannot be solved by a single industry. Healthcare can treat. Employers can redesign work. Governments can influence security and public infrastructure. Schools can strengthen connection and development. Technology companies can change digital environments. Communities can create social buffering. Individuals can build skills and seek appropriate care. The failure occurs when one level is expected to compensate indefinitely for failures at all the others.

The AMOS framing therefore rejects the idea that resilience means making the individual infinitely adaptable. A resilient system is one that does not require infinite adaptation from its components.

That distinction may be the most important principle in the entire debate.

14. The Benchmark for the Next Generation of Companies Will Be Regenerative Productivity

For much of the industrial era, the dominant productivity question was how much output could be generated from a unit of labor. The next generation of human-capital strategy will need a second question: how much of the capacity used to generate that output can be regenerated?

This suggests a different concept of productivity—regenerative productivity. The objective is not lower performance. It is high performance that preserves the capability to perform again. A high-performing organization should therefore be capable of intense mobilization when necessary while also being capable of returning to a lower-load state. Permanent emergency is not high performance. It is a failure to complete the recovery cycle.

The distinction is commercially important. Companies capable of sustaining talent may experience lower replacement costs, greater institutional memory, stronger customer relationships and more reliable execution. Organizations that systematically exhaust people may still outperform temporarily, particularly during growth periods, but their apparent advantage incorporates a liability that conventional accounting rarely captures.

Leaders should consequently ask a different set of questions during strategy and transformation. Not only: How many roles can AI remove? But: How much cognitive load can AI remove? Not only: How can utilization increase? But: What level of utilization preserves resilience? Not only: How do we reduce mental-health absence? But: What operating conditions repeatedly precede that absence? Not only: Do employees have wellbeing benefits? But: Can they use those benefits without paying a professional penalty? Not only: How resilient are our people? But: How much unnecessary resilience does our operating model demand from them?

These questions move mental health from the periphery of corporate responsibility into the center of operating-model design.

15. What Leaders Should Do Differently

The immediate implication is not a wholesale rejection of modern work, markets, technology or mental-health treatment. It is the creation of a more disciplined hierarchy of intervention.

Clinical conditions should receive appropriate clinical care. Individual capability gaps should receive individual support. Management failures should be corrected through management. Work-design failures should be corrected through work design. Economic and policy problems should not be disguised as wellness problems simply because the organization lacks authority to solve them.

Companies should begin by identifying where human regulatory capacity is being consumed unnecessarily. Repeated priority changes, excessive meetings, low-quality management, ambiguous accountability, avoidable after-hours communication, inefficient administrative work and poor workflow design frequently generate load without generating proportional value. Removing these burdens can improve both productivity and human sustainability; there is no inherent trade-off.

The second priority is to distinguish peak-load systems from chronic-load systems. Many industries require periods of extraordinary effort. That is not inherently incompatible with human functioning. The greater danger is when surge conditions become the permanent baseline. Organizations need explicit mechanisms through which load returns to normal after a peak.

Third, leaders should protect genuine agency. Flexible-work policies mean little if employees who exercise flexibility are penalized informally. Leave entitlements mean little if work accumulates while people are away. Psychological safety means little if challenging unrealistic targets damages careers. Systems should be evaluated by actual consequences, not policy language.

Fourth, AI productivity programs should include an explicit capacity-dividend decision. Before efficiency gains are entirely converted into higher output or lower headcount, leaders should determine whether some portion should be retained as reduced administrative load, greater decision quality, stronger customer interaction, learning or recovery. This transforms AI from a pure extraction technology into a system-improvement technology.

Finally, organizations should connect care data to prevention without compromising individual privacy. The objective is not employee surveillance. It is pattern detection at the system level. If particular operating conditions repeatedly precede absence, attrition or distress, leadership should treat those conditions as management information.

16. What the Mental-Health Market Has Revealed

The growth of mental-health services should not be interpreted simply as evidence of societal failure. Greater awareness, reduced stigma and expanded access can represent real progress. Millions of people who previously suffered without support can now seek help. Any serious systems argument must preserve that achievement.

But the expansion of the market reveals another truth: society has become much better at monetizing recovery than at protecting the conditions that make recovery possible.

We have created products for sleep while designing lives that erode it. We sell mindfulness into environments built around interruption. We provide resilience programs inside organizations that sometimes reward permanent availability. We treat loneliness while weakening physical community. We build digital detox products on top of an attention economy. We prescribe recovery after burnout while continuing to celebrate operating models that make burnout predictable.

The contradiction does not invalidate the interventions. It exposes the incompleteness of the architecture around them.

A healthcare system should not be required to repair every consequence of economic design. Employers should not be expected to solve every source of human distress. Individuals should not be expected to regulate every instability created by institutions. Governments cannot legislate away every difficulty. Responsibility is distributed—but distributed responsibility cannot mean responsibility disappears.

The next phase of mental-health strategy therefore needs to move from treatment expansion alone toward causal allocation: identifying which level of the system is generating which part of the problem and assigning intervention accordingly.

17. The Larger Economic Question Is Whether Growth Is Consuming the Capacity Required to Sustain Growth

This is ultimately why mental health belongs in strategic and economic debate.

Every economic system depends on human beings capable of learning, deciding, trusting, cooperating, creating, caring, leading and adapting. Those capabilities are routinely treated as renewable simply because new people enter the labor market. But population-level deterioration in mental health, social connection and workforce sustainability raises a harder question: whether parts of the modern economy are drawing down the quality of their own human substrate.

WHO's estimate that 970 million people were living with mental disorders in 2019, together with the approximately US$1 trillion annual productivity loss associated with depression and anxiety, does not by itself prove that modern economic systems are causing those conditions. It does establish that the human and economic burden is large enough that environmental and organizational design cannot reasonably remain outside the analysis. (World Health Organization)

The appropriate response is therefore neither anti-market nor anti-growth. It is to improve the definition of growth.

Growth that increases current output while reducing future human capacity contains an unrecognized liability. Productivity that requires continuously escalating compensatory healthcare expenditure is not fully measured productivity. Technology that accelerates output while degrading judgment, social connection or recovery is not costless efficiency. An organization that achieves exceptional results while systematically losing its most capable people has not necessarily discovered a superior operating model.

The broader economic challenge is to distinguish value creation from capacity extraction.

The two can look identical in the short term.

Only time reveals the difference.

18. The Human System Is Not the Residual Variable

For more than a century, management science has progressively optimized the organization around the worker. Scientific management optimized tasks. Industrial engineering optimized production. Enterprise software optimized processes. Digital platforms optimized information. Analytics optimized decisions. Automation optimized repetitive work. Artificial intelligence is beginning to optimize cognition itself.

The remaining danger is that the human becomes the residual variable expected to adapt to whatever configuration produces the highest measured efficiency.

AMOS reverses that logic.

The human is not an infinitely elastic component inside the operating system. Human biological, cognitive and social constraints are part of the operating specification. A system that repeatedly violates those constraints should not automatically classify the resulting distress as failure of the component.

This principle does not make organizations softer.

It makes systems engineering more complete.

The strongest systems are not those that place the least demand on their components. They are those capable of generating extraordinary performance without destroying the components on which extraordinary performance depends.

That is the distinction between extraction and regeneration.

Between utilization and sustainability.

Between surviving a system and being able to continue operating within it.

Conclusion: Mental Health Is Telling Us Something About the Systems We Built

The mental-health crisis is real, but its meaning is larger than the growth of mental illness alone. It is also exposing a structural weakness in the way modern institutions think about people.

We have built sophisticated systems for treating distress after it becomes visible. We have not built equally sophisticated systems for identifying when environments are repeatedly producing unnecessary regulatory load. We measure absence more reliably than exhaustion, utilization more reliably than recovery, productivity more reliably than sustainability and treatment more reliably than prevention.

That imbalance matters because the human organism has limits regardless of whether economic systems recognize them.

The central insight of the AMOS approach is therefore not that every mental-health condition is caused by society, work or technology. That would be unsupported and clinically irresponsible. It is that human beings operate inside systems, and those systems participate materially in determining the load humans must regulate. Once that relationship is acknowledged, mental health becomes more than a healthcare outcome. It becomes a measure of system quality.

The future of mental-health strategy should therefore not be framed as a choice between treatment and structural change. We need both. People experiencing illness require high-quality care. People under acute pressure require support. But the information generated by widespread distress must also travel upstream. It must influence management, technology, organizational design, education, urban systems, labor policy and ultimately the assumptions by which economic performance is measured.

AMOS — the Absolute Operating System created by Trang Phan — provides one way of framing that transition: begin with the constraints of the system, trace where load is created, distinguish symptoms from causes, preserve the boundaries required for stability, and do not call an optimization successful when it damages the capacity required to sustain the optimization.

The business implication is substantial.

Mental health is not simply something organizations support. It is something operating systems produce.

A company produces it partly through workload, predictability, management, autonomy and recovery. A school produces it partly through belonging, pressure, safety and connection. A city influences it through housing, transport, community and security. Technology influences it through attention, pace, comparison and availability. Economic institutions influence it through stability, opportunity and uncertainty. Healthcare then receives a portion of the consequences.

The strategic objective should therefore evolve from building ever larger systems for helping people withstand instability toward building systems that generate less unnecessary instability in the first place.

The organizations that understand this earliest may gain more than healthier workforces. They may develop a fundamentally superior model of human productivity—one in which technology increases capacity rather than simply raising expectations, in which recovery is treated as productive infrastructure, in which mental-health signals feed organizational learning, and in which human sustainability becomes part of enterprise resilience.

For decades, the prevailing question has been:

How much can we get from people?

The more important question for the next economy is:

How much can people produce while preserving their capacity to produce, think, care, create and adapt again tomorrow?

That is not a wellness question.

It is not an HR question.

And it is not principally a question of compassion.

It is a question of operating-system design.

The market learned how to sell recovery.

The next transformation is learning how to stop consuming it faster than humans can regenerate it.

Selected evidence base

World Health Organization, Mental Health and World Mental Health Report: approximately 970 million people globally were living with mental disorders in 2019; anxiety and depressive disorders were the most common. (World Health Organization)

World Health Organization and International Labour Organization, Mental Health at Work: approximately 12 billion working days are lost annually to depression and anxiety, representing approximately US$1 trillion in lost productivity; WHO also identifies excessive workload, low job control and job insecurity among workplace mental-health risks. (World Health Organization)

World Health Organization, World Mental Health Day 2022: anxiety and depressive disorders were estimated to have risen by more than 25% during the first year of the COVID-19 pandemic. (World Health Organization)

U.S. Centers for Disease Control and Prevention, Youth Risk Behavior Survey 2023: 39.7% of U.S. high-school students reported persistent sadness or hopelessness, 20.4% seriously considered attempting suicide and 9.5% reported an attempt. (CDC)

Institution for Trust, Intelligence & Autonomy

Design. Intelligence. Automation.

Home - Work - Capabilities - Insights. - About - Contact

INQUIRIES

solutions@itia.ai

North America — Vietnam — Australia

© 2026 ITIA-Production-grade AI engineering for the global enterprise.