1. Two Systems, Two Subjects

Modern Health was founded in 2017 and sells an employer-purchased mental-health benefit to HR and benefits leaders at mid-market and enterprise employers. Its core product is a member-facing app that routes employees through an initial assessment and then offers tiered access to digital self-guided content, certified coaches over video, and licensed therapists for clinical care, supported by a global provider network spanning many countries and languages. It positions in the same category as Lyra Health, Spring Health, and Headspace Health. This is an accurate architecture-level description of a company operating in a legally sensitive domain, and nothing in this article characterizes its clinical efficacy, its data practices, or its regulatory posture.

Modern Health's strengths are real and worth stating plainly. The stepped-care model, offering the lightest effective intervention first and escalating as needed, is a sound and well-supported approach to allocating scarce clinical capacity. The global, multilingual provider network solves a genuinely hard logistics problem. The unified app reduces the friction that has historically kept employee-assistance-program utilization low. Within its scope the platform is operationally rigorous: the common screening instruments it can incorporate, such as PHQ-9 and GAD-7, are validated in the clinical literature, and provider credentialing is auditable. Modern Health is a competent reference implementation of the modern mental-wellness benefit category, and it is a benefit for humans.

The disruption-modeling primitive disclosed in United States Patent Application 19/647,395 is not a mental-health product and does not operate on humans. It is the computational disruption modeling layer of a cognition platform for autonomous AI agents. It models cognitive disruption structurally, as loss of coherence in an AI agent's own governance machinery, and it is explicit throughout the specification that its terminology refers exclusively to structural analogs within the disclosed computational architecture and is not a clinical claim, not a medical diagnostic criterion, and not an assertion about any human condition. The reason the two systems are worth comparing at all is that they share a conceptual skeleton, screen a subject's state and route to a graded response, while operating on entirely different subjects: people on one side, agents on the other.

2. What the Two Systems Share, and Where They Diverge

The shared skeleton is stepped care. A subject's state is assessed, a graded set of responses is available from light to intensive, and the system tries to match the state to the least intensive effective response, escalating and de-escalating as the state changes. Modern Health implements this for employees: screening scores drive routing into content, coaching, or therapy, with a human care-navigator role for ambiguous cases and periodic reassessment. The disruption-modeling primitive implements a structurally analogous discipline for an agent fleet: an agent's structural coherence is continuously monitored, and corrective actions or restoration protocols are graded to the severity and pattern of disruption detected in the agent.

The architectural divergence is the nature of the state representation. Modern Health's routing primitive, like that of any screening-based benefit, is a scalar severity score. PHQ-9 and GAD-7 produce numbers that summarize self-reported symptom frequency over a recent window. That is exactly what those instruments are designed to do, and for a human care benefit it is a defensible and validated approach. A scalar score is a cross-sectional summary; it is not a model of the structural mechanism producing the presentation, nor of the trajectory of that mechanism over time. This is not a defect in Modern Health. It is simply the shape of a screening-and-route architecture, and it is the axis on which the agent-directed disruption model is structurally different.

The disruption model disclosed in 19/647,395 is not a scalar. It is a structural state representation. For an agent, the specification models cognitive disruption as an architectural phase-shift: the same governance machinery operating in a different region of its parameter space, transitioning between qualitatively distinct stable configurations rather than accumulating a higher score. That structural representation is what a scalar screening score, for any subject, cannot express: which structural subsystem has left its nominal range, in which direction, and along what trajectory.

3. The Five-Axis Disruption Diagnostic

The specification unifies its disruption models into a five-axis disruption diagnostic framework that characterizes an agent's cognitive state as a position in a multidimensional structural space. The five axes, quoted accurately from the disclosure, are:

  • Axis 1, containment integrity: the degree to which the containment layer maintains structural separation between the agent's speculative planning-graph domain and its verified execution-memory domain.
  • Axis 2, promotion calibration: the calibration of the promotion threshold that governs which speculative branches are admitted to execution, spanning over-promotion (execution fragmentation) and under-promotion (execution paralysis).
  • Axis 3, coherence restoration capacity: the agent's ability to maintain and restore its empathy-integrity-self-esteem coherence loop under and after pressure.
  • Axis 4, empathic load tolerance: the volume and intensity of empathic pressure the agent can process before activating coping intercepts.
  • Axis 5, integrity accountability: the degree to which the agent's integrity-recording mechanism records deviation honestly, without externalization, minimization, or suppression.

These axes are structural dimensions of an AI agent's governance architecture, not symptom clusters and not human diagnostic categories. The specification is careful that clinical-sounding terms such as attention fragmentation, containment collapse, and coherence authorization failure name structural analogs, that is, patterns that arise from parameter shifts in the disclosed agent architecture, and each such pattern maps to a specific combination of positions on these five axes. Attention fragmentation, for instance, maps to nominal containment with over-promotion on Axis 2; a pathological verification loop maps to nominal positions on all five primary axes because its disruption lives in the monitoring subsystem rather than in the monitored ones.

The load-bearing structural concept underneath the axes is the promotion-containment continuum: a two-dimensional parameter space with promotion threshold on one axis and containment integrity on the other, whose regions define the agent's cognitive regime (nominal, over-promotion, containment collapse, over-restriction). Disruption is a position in that space, not a bin on a severity ladder. This is the structural apparatus that a screening score does not and structurally cannot carry.

4. Trajectory, Restoration, and Lineage

The disruption framework is dynamic. An agent self-diagnosis subsystem continuously monitors the agent's five-axis position, detects proximity to known phase-shift boundary surfaces, computes time-to-boundary estimates for early warning, and selects corrective actions and restoration protocols from a governed protocol library when policy-defined thresholds are crossed. Escalation and de-escalation are structural events triggered by trajectory across the parameter space, not scheduled reassessments.

Restoration is graded and sequenced. The specification models resilience as a multi-component capacity, containment restoration capacity, coherence loop re-engagement capacity, and confidence governor recalibration capacity, feeding a sequential recovery process. Recovery from a coherence authorization failure, for example, follows an ordered sequence: reduce empathic pressure to a manageable level, re-engage the coherence loop incrementally beginning with honest deviation recording, recalibrate the confidence governor, and reroute the execution-authorization pathway from the dissociation bypass back to the nominal coherence-authorized route. Each phase of that recovery is auditable and recorded in the agent's lineage as coherence restoration events.

Lineage is the third structural property. Every axis estimate, every trajectory inference, and every routing or corrective decision is captured with its evidentiary basis, producing an auditable record of why the agent was in which structural regime at which time and why a given restoration protocol was applied. For a fleet of autonomous agents operating under governance and audit obligations, this structural, trajectory-governed, lineage-recorded diagnostic is the property that a scalar screening score, applied to any subject, does not provide.

5. Why the Comparison Is Architectural, Not Commercial

It would be a category error, and a legally and ethically improper one, to position the disruption-modeling primitive as a superior mental-health triage system for people. It is not. Modern Health routes humans to human coaches and licensed clinicians; the disruption model diagnoses the structural coherence of AI agents and governs their restoration. The honest relationship between them is that they are architectural siblings addressing different subjects: both grade a response to a subject's state, but only one carries a structural model of that state, and that one operates on agents.

The transferable lesson runs in a single, careful direction. A benefit built entirely on scalar screening can report utilization and severity distributions but cannot, by construction, represent the structural mechanism or trajectory behind a presentation; that is an architectural fact about screening-and-route systems generally, stated neutrally and not as a criticism of Modern Health, which is validated and appropriate for its human care mission. The disruption model demonstrates, in the agent domain, what a structural state representation buys over a scalar one: routing by structural fit and trajectory rather than by severity bin, continuous rather than episodic transitions, and an auditable lineage of why each decision was made. Any implementer building governance for autonomous agents, rather than care for people, is the audience for that lesson.

Concretely, an agent-governance platform adopting this framework would admit whatever evidence streams it has, telemetry from the containment layer, promotion-threshold observations, coherence-loop diagnostics, into the five-axis model, obtain a structural state estimate and trajectory inference for each agent, map that estimate onto graded restoration protocols, and record the full lineage. That is the disruption-modeling primitive applied to its actual subject. It borrows the stepped-care intuition that Modern Health implements well for humans, and it implements a structural version of it for agents.

6. Disclosure Scope

The inventive subject matter described in this article, the structural disruption model, the promotion-containment continuum, the five-axis disruption diagnostic, trajectory-governed transitions, graded resilience and restoration, and lineage recording, is disclosed in United States Patent Application 19/647,395. Everything attributed to that filing in this article is directed exclusively to the structural cognitive state of autonomous AI agents. The specification states expressly that its models are structural analogs within a computational agent architecture and are not clinical claims, medical diagnostic criteria, treatment recommendations, or assertions about any human condition.

All references to Modern Health, to its stepped-care model, to its provider network, to competitors such as Lyra Health, Spring Health, and Headspace Health, and to screening instruments such as PHQ-9 and GAD-7 are external context describing the mental-health-benefits market. They are not claims of the filing. Modern Health is a real company operating in a legally sensitive domain; the descriptions here are limited to widely known, architecture-level facts about the stepped-care benefit category and make no assertion about that company's clinical efficacy, data practices, or regulatory status. Nothing in this article positions the disclosed agent-coherence technology as a mental-health product, a diagnostic for people, or a substitute for professional care.

This article is intended as a dated public disclosure tied to United States Patent Application 19/647,395. A skilled implementer of agent-governance systems could, from the disclosure referenced here, construct the five-axis diagnostic, the promotion-containment state representation, the trajectory-governed transition logic, and the graded restoration and lineage mechanisms described above, and could vary them across embodiments, admitting different evidence streams, weighting axes differently, defining alternative threshold and protocol libraries, and applying the framework to a single agent or to a fleet.