M31 Capital · Strategic Architecture · February 2026

The Architecture

A complete system design for how humanity prevents, diagnoses, and heals — reconciling every iteration of this thesis into one definitive blueprint.

Jen Berry · M31 Capital · Version 3.0
Part I
Why the System Fails

Five structural failures — not personality flaws, not underfunding — that guarantee bad outcomes

The American healthcare system does not fail because its individual participants are bad. It fails because its architecture makes good outcomes structurally impossible for the conditions that now represent 90% of health expenditure.

The Specialty Silo

Conditions like Ehlers-Danlos Syndrome, MCAS, POTS, ME/CFS, and autoimmune overlap syndromes require synthesis across cardiology, rheumatology, immunology, neurology, and gastroenterology. No physician in the current system is incentivized, trained, or given adequate time to perform this synthesis. The average diagnostic delay for EDS is measured in years to decades. The information to make the diagnosis typically exists in the patient's records — scattered across providers who never see each other's notes in context. This is not a knowledge problem. It is a data architecture problem.

I know this because I lived it. I had two concierge physicians, medical teams at Mt. Sinai, Weill Cornell, and Yale — fifteen specialists across three of the most prestigious hospital systems in the country — and not one of them connected the dots. I diagnosed myself with narcolepsy and Ehlers-Danlos Syndrome, then spent months convincing those same specialists to run the tests that proved it.

$4.5T
Annual US healthcare spending
34¢
Of every dollar → administration
~5%
Spent on prevention
70–80%
Chronic disease from lifestyle

Prevention Separated from Treatment

Prevention generates no revenue in fee-for-service systems. The person who keeps you healthy is almost never the person who treats you when you are sick, which means critical continuity is lost. Seventy to eighty percent of chronic disease burden is driven by lifestyle and environmental factors, yet less than five percent of healthcare spending addresses these causes. The economics of the system actively punish keeping people well.

Epistemological Rigidity

The current evidence hierarchy was designed for pharmaceutical regulation. It systematically excludes interventions that cannot be patented, cannot be blinded, or cannot be isolated from their delivery context — dietary interventions, movement practices, breathwork, plant medicines, psychedelic-assisted therapy. This is not scientific rigor. It is a category error that leaves the majority of the world's healing traditions outside the system, regardless of their efficacy.

Incentive Misalignment

Fee-for-service rewards volume. Specialists earn two to five times what generalists earn. Complex patients are financial liabilities — they consume more resources than their reimbursement covers. Insurance companies profit by denying care. Every incentive in the system points away from the outcomes patients actually need.

The Intermediation Web

Insurance companies, PBMs, prior authorization departments, billing coders, hospital administrators — each extracting value without creating health. The system is optimized for claims processing, not care delivery. The patient is the least powerful actor in an architecture designed to serve every other participant's financial interest.

The Coordination Gap in Alternative/Integrative Care
Supplement Quality StandardsNo trusted authority
Device Efficacy VerificationNo governing body
Protocol Synthesis (Individual)Does not exist
Practitioner CredentialingFragmented
Insurance Coverage for IntegrativeMinimal to none
Patient Data OwnershipEmerging

This chaos is exactly what incumbents point to when they argue against alternative healthcare. And they're not wrong — the chaos is real. But their solution — "come back to the pharmaceutical system" — is the equivalent of arguing that because the new world has pirates, we should return to the prison. The real answer is to build the coordination infrastructure without the extraction layer.

Part II
The Eight-Layer Architecture

Protocols over institutions. Transparency over gatekeeping. A system design for healthcare that actually heals.

What follows is the definitive system architecture — refined across multiple iterations, reconciling the best structural thinking from every version of this thesis. The principle: transparent protocols with aligned incentives outperform opaque institutions with misaligned incentives.

Layer 1 — Foundation
The Patient Graph
Every individual maintains a continuously updated health record organized not by encounter or specialty but as a systems map — a graph database capturing symptoms, biomarkers, genetics, environmental exposures, lifestyle data, microbiome profiles, hormonal patterns, and nervous system metrics with mapped relationships and correlations. Stored on decentralized infrastructure. Owned by the patient. AI pattern recognition layers on top to flag cross-system clusters and surface novel associations not yet formally described.
Layer 2 — Intelligence
AI-Assisted Systems Physician
The primary care relationship is inverted. Instead of a gatekeeper, the central figure is a deeply trained generalist — a "systems physician" — whose role is synthesis across organ systems. AI augments across three functions: continuous pattern surveillance, differential diagnosis generation with structured devil's advocacy, and treatment interaction mapping across the full therapeutic spectrum. The AI handles computation at superhuman scale. The human handles judgment, empathy, and wisdom.
Layer 3 — Prevention
Prevention as Default Operating Mode
Prevention is not a separate activity. It is the default state. Health coaches, movement specialists, nutritional therapists, and stress physiology practitioners are core members of the care team, not "wellness extras." Population-level environmental monitoring, personalized continuous biomarker surveillance, and behavioral/lifestyle medicine as first-line intervention — all integrated into the systems physician relationship and fully funded.
Layer 4 — Therapeutics
The Tiered Evidence Framework
A therapeutic spectrum that does not treat randomized controlled trials as the only valid evidence, while maintaining rigor. Three tiers of evidence — strong clinical, emerging with plausible mechanisms, and traditional/experiential — with every encounter in Tiers 2 and 3 generating structured outcome data, turning the entire system into a continuous pragmatic trial. See Part III for full detail.
Layer 5 — Community
Community Intelligence Network
Patient communities as a formal, load-bearing layer of the system — not a support group bolted on. Structured data collection from patient-reported outcomes, symptom triangulation, treatment response patterns, and comorbidity mapping across thousands of real cases. A parallel pattern-recognition engine alongside AI, specializing in the lived-experience data that clinical trials cannot capture at speed or scale.
Layer 6 — Verification
On-Chain Standards Body
Independent, decentralized standards organization operating on blockchain — supplement testing, device validation, practitioner credentialing, protocol development, vendor approval. All records immutable and transparent. Staffed by functional medicine doctors, integrative practitioners, researchers, pharmacologists, and patient advocates. Standards evolve based on evidence, not politics or industry capture.
Layer 7 — Sovereignty
Blockchain Data Sovereignty
Self-sovereign health identity. Health data encrypted with granular, programmable, time-limited consent. When anonymized data contributes to research or AI training, the individual is directly compensated. No single institution controls the data layer. Open, blockchain-verified global registry of therapeutic protocols with standardized evidence tiers, supply chain provenance, and decentralized peer review.
Layer 8 — Incentives
Outcomes-Based Funding Architecture
Fully capitated funding with outcomes-based bonuses. Complex patients carry higher capitation payments, not lower — making the hardest cases the most financially rewarding to serve well. Behavioral incentives structured as rewards for engagement rather than penalties for outcomes. Four funding layers detailed in Part V.
Part III
The Tiered Evidence Framework

Rigorous and epistemologically humble — the single most important intellectual contribution of this thesis

This is the answer to the question that will be asked first and loudest: "Where's the evidence for alternative medicine?" The answer is not to win the evidence argument retroactively. It is to build the system that generates the evidence prospectively.

Tier 1
Strong Mechanistic & Clinical Evidence
Interventions supported by RCTs, large observational studies, and well-understood mechanisms. Standard pharmaceuticals and surgical procedures — but also meditation for anxiety (robust trial data), acupuncture for specific pain conditions (strong systematic reviews), and herbal medicines with demonstrated pharmacological evidence. These are covered universally through the base capitated layer.
Examples: SSRIs for severe depression, knee replacement, MBSR for anxiety, acupuncture for chronic lower back pain, GLP-1 agonists for metabolic syndrome
Tier 2
Emerging Evidence with Plausible Mechanisms
Interventions where clinical trials are underway or where strong mechanistic reasoning plus historical use data suggests benefit. Available through the system with appropriate clinical supervision and mandatory data collection — every patient encounter feeds into ongoing effectiveness research. This is where the system becomes a continuous pragmatic trial.
Examples: Psilocybin for treatment-resistant depression, peptide therapeutics like BPC-157, functional nutrition protocols, breathwork for autonomic regulation, bioelectric stimulation for chronic pain, psychedelic-assisted therapy for PTSD
Tier 3
Traditional & Experiential Evidence
Interventions with long historical use and patient-reported benefit but limited formal research. Available with informed consent, practitioner certification, and mandatory outcome tracking. The key principle: absence of evidence is not evidence of absence, and the burden of proof should be proportional to the risk of the intervention. A low-risk herbal preparation with 500 years of traditional use should not face the same evidentiary bar as a novel synthetic molecule.
Examples: Ayurvedic protocols, traditional Chinese herbal formulations, energy healing modalities, frequency/vibration therapies, certain consciousness-based healing practices
"The critical design element: every therapeutic encounter in Tiers 2 and 3 generates structured outcome data that feeds back into the evidence base. This turns the entire healthcare system into a learning system — a massive, continuous pragmatic trial. Over time, interventions move between tiers based on accumulated real-world evidence, not just on whether someone funded a formal RCT." — Design Principle

Where Consciousness Enters the Framework

M31's research has identified consciousness-based healing as a high-conviction emerging paradigm — scored at investable levels across the five signals framework. The CIA's Gateway Process documents, the convergence of contemplative traditions, and emerging research in psychoneuroimmunology all point toward the same conclusion: consciousness is not merely epiphenomenal to health. It may be foundational.

Within this evidence framework, consciousness-based modalities enter at Tier 2 where mechanistic research exists (meditation, breathwork, psychedelic-assisted therapy) and at Tier 3 where the evidence is primarily experiential and traditional (energy healing, frequency medicine, certain biofeedback practices). The critical point is that the system does not pre-judge these modalities. It generates the data that allows them to rise or fall based on observed outcomes. If something works, the data will show it. If it doesn't, the data will show that too. This is genuine scientific epistemology — not the constrained version that only studies what can be patented.

Part IV
Community as Infrastructure

Not a support group. A load-bearing wall. A parallel intelligence system that outperforms institutions in specific domains.

After I was diagnosed, I found online patient communities — people with narcolepsy, EDS, and post-COVID conditions. What I discovered there changed the entire trajectory of this thesis: patients, in aggregate, know more about their conditions than the specialists treating them.

These communities were doing real-time pattern recognition. Triangulating symptoms across thousands of cases. Testing treatments through self-experimentation and sharing results. Building knowledge bases that no single physician could match. The narcolepsy community knew which medications worked best for which subtypes. The EDS community had mapped comorbidities that were barely mentioned in medical literature. The COVID long-haul community was identifying treatment protocols months before formal research caught up.

This is not anecdote replacing evidence. This is a different form of evidence — distributed, experiential, high-volume, and fast. It is the form of evidence that the institutional system dismisses and that the new system must integrate.

The Four Community Functions

Community Intelligence Architecture
Data LayerStructured patient-reported outcomes, symptom patterns, treatment responses across thousands of cases
Diagnostic LayerComorbidity mapping, symptom triangulation, pattern identification that accelerates diagnosis
Protocol LayerTreatment response data at volume and speed no clinical trial can match
Support LayerEmotional and practical support that improves adherence and outcomes

The Patient Graph (Layer 1) meets the Community Graph (Layer 5). AI systems can now ingest both — the individual's complete health record and the aggregate patterns from thousands of similar patients — to generate differential diagnoses and treatment recommendations that neither data source could produce alone. This is the synthesis engine that the siloed specialty system cannot build.

Part V
The Funding Model

Four layers — because "how do you pay for this" is the first question any serious person asks

This is where most utopian healthcare designs fall apart. We are being specific and realistic.

Layer 1

Universal Capitated Base

Every person receives the core package: their systems physician relationship, AI-assisted patient graph, preventive monitoring, behavioral and lifestyle medicine, and access to Tier 1 interventions. Funded through taxation or mandatory contribution. Non-negotiable and universal. The economic argument is straightforward: prevention and early detection are dramatically cheaper than late-stage treatment. Every dollar spent on metabolic health stability in the 30s saves tens of thousands in diabetes, cardiovascular, and kidney treatment in the 60s. The capitated model makes these savings accrue to the entity paying for care.

Layer 2

Expanded Therapeutic Access

Tier 2 and Tier 3 interventions covered through a combination of capitation top-ups and individual health savings accounts. Part required contribution (like Singapore's Medisave but expanded), part risk-pooled insurance. Patients have meaningful choice and skin in the game for interventions with less certain evidence, but the financial barrier is low enough that access is not determined by wealth.

Layer 3

Data Dividends & Participation Incentives

Patients who contribute anonymized data to research and AI training receive direct compensation. Patients who participate in structured outcome tracking for Tier 2 and 3 interventions receive reduced costs. This creates a virtuous cycle: participation improves the evidence base, which improves the AI, which improves everyone's care, and participants are directly rewarded. The blockchain data sovereignty layer makes this possible and auditable.

Layer 4

Outcome-Linked Pricing

For pharmaceutical companies and protocol developers, pricing is partially linked to real-world outcomes as measured by the system. If your drug works as well as claimed, you get full price. If outcomes fall short, the price adjusts automatically. This eliminates the current absurdity where drugs with marginal real-world benefit command premium prices based on carefully selected trial populations. The continuous data layer makes this measurable for the first time.

Incentive Design — Critical Distinction
Bonuses accrue forPreventive milestones, diagnostic speed, patient outcomes, complex-case management
Penalties apply forMissed early diagnoses, unnecessary fragmentation, over-testing without outcome improvement
Complex patientsHigher capitation — making the hardest cases the most rewarding to serve well
Behavioral incentivesRewards for engagement, not penalties for outcomes (health is not purely volitional)
"The penalty model assumes health behaviors are purely volitional. They are not. Obesity correlates with poverty, childhood trauma, gut microbiome composition, genetic predisposition, and endocrine disruptors. Penalizing outcomes that are only partially within individual control is ethically questionable and practically counterproductive — it drives the people who most need the system away from engagement." — Design Principle
Part VI
The Knowledge Architecture

Information flows up, not just down. Clinical observation feeds research. The system learns.

In the current system, knowledge flows one direction: from research down to practice. Clinical observations — the patterns that frontline practitioners see every day — have no formal mechanism to flow upward into research. This is an extraordinary waste of signal.

The new system has a formal mechanism for bidirectional knowledge flow. Every therapeutic encounter generates structured data. AI continuously analyzes this data for emerging patterns — unexpected efficacy signals, adverse combinations, population-specific responses, novel comorbidity clusters. When a pattern reaches statistical significance, it is automatically flagged for formal investigation.

This means the system can identify that a particular herbal protocol is showing remarkable results for a specific subset of autoimmune patients — before anyone designs a formal trial. It can detect that a combination of breathwork and a particular peptide is reducing inflammation markers in post-COVID patients — before a researcher writes a grant proposal. The system becomes a massive, continuous observational study that generates hypotheses at speed and scale impossible in the current research architecture.

The community intelligence layer (Layer 5) feeds directly into this. Patient-reported patterns — the symptom triangulations, the treatment responses, the comorbidity mappings that communities identify — become structured data inputs for the knowledge system. What patients already do informally in Reddit threads and Facebook groups becomes formal, quantified, and actionable.

Part VII
The Convergence

The core infrastructure already exists across three companies

The architecture described above is not theoretical. Its core components already exist.

Integration Map
Pearl Health → Layers 2, 8AI-assisted physician support + value-based incentive engine (3,500+ providers, 40+ states)
Garner Health → Layer 6Quality intelligence + provider routing (320M patient records, 500+ clinical metrics)
Oscar Health → Layers 7, 8Tech-native insurance platform + member experience layer ($9.18B revenue, 2M+ members)
Extension RequiredPatient Graph, Community Network, Tiered Evidence Framework, On-Chain Standards, Data Sovereignty

The combination of Pearl's value-based care operating system, Garner's provider quality intelligence, and Oscar's tech-native insurance platform — augmented by AI-assisted diagnostics, blockchain-enabled data sovereignty, an expanded therapeutic spectrum, and the community intelligence network — constitutes the architecture for a fundamentally new healthcare system. One that prevents disease instead of treating it, diagnoses complex multi-system conditions in months instead of decades, and is epistemologically open to the full range of human healing traditions.

The opportunity is to assemble these pieces before the market recognizes their combinatorial value.

The Synthesis

A System That Learns

This architecture — the patient graph, the systems physician, prevention as default, the tiered evidence framework, community intelligence, on-chain standards, data sovereignty, and outcomes-based funding — is not a utopian fantasy. Its individual components are either already built, currently being built, or technically achievable with existing infrastructure.

What does not yet exist is the integration layer — the entity that connects the components, aligns the incentives, and maintains the design coherence as the system scales. That is the M31 thesis. That is the investment opportunity.

The old system treats organs, not people. It manages disease, not health. It generates revenue from suffering, not savings from prevention. It siloes knowledge and hoards data. Every structural feature that makes it profitable makes it terrible at its stated purpose.

The new system inverts every one of those features. And for the first time, the technology, the economic models, the regulatory momentum, and the cultural demand are all present simultaneously.

This is the architecture. Now we build it.