Primary Care - Max Marchione
Why this is in the vault
Founder is building out the health-and-longevity targeting system. To know what sensors and actions to invest in, we need to understand why the existing primary-care system is producing the outcomes it does (Attia's "Medicine 2.0 fails most people" thesis from a structural-economics angle, not just a clinical angle).
Marchione's primary care essay is the most-rigorous critique of the digital-health investment thesis that's been published by anyone in his cohort. It also explains why DTC longevity platforms (Superpower, Function Health, Inside Tracker) exist at all: they're the structural response to the failures he diagnoses. Reading this clarifies which parts of the longevity-services market are real businesses vs marketing wrappers around lab requisitions.
Maps to the RDCO targeting-systems thesis (this is healthcare-as-targeting-system at the industry layer) and to the founder's "Med 3.0-aware doctor" sensor requirement in the project README.
The core content
Thesis
$100+ billion of digital-health investment since 2010 has failed to produce a generational primary-care company because every new model (virtual fee-for-service → membership → value-based → AI-powered) has targeted symptoms rather than the structural reimbursement and incentive problems. Public-market evidence: Teladoc down 90% since Feb 2021, One Medical down 85% pre-Amazon acquisition, Babylon down 95% before delisting.
Three load-bearing critiques
The unit economics of primary care don't work in fee-for-service. Median US health-system operating margins sit at 1.7%. Primary care is operated at a loss by 75% of health networks, subsidized by downstream specialist + procedure revenue. Independent primary care competing on access, quality, or price all compress margins because reimbursement is fixed by payers - there's no price discovery, only operational optimization.
AI startups are misreading the bottleneck. AI companies obsess over diagnosis (minority of physician work), and the diagnostic advantage commoditizes as foundation models distribute horizontally to incumbents. "Distribution will matter more than accuracy" - the EHR-integrated incumbents (Epic, Oracle) capture the AI productivity gains, not new entrants.
DTC membership-care CACs are crushing. Direct primary care companies have reported customer acquisition costs exceeding $2,000. At a $199-500/yr membership price point, this is multi-year payback assuming low churn - which doesn't hold up empirically.
Counter-evidence he cites
Parsley Health's employer-reimbursement model demonstrated 77% specialist reduction and 65% medication reduction - i.e. the value of high-touch primary care is real when the payer is aligned. The unsolved problem is the distribution channel, not the clinical model.
Where he leaves it
He defers solutions explicitly: "In the coming months, I'll break down individual companies and start suggesting solutions." His hinted direction is "outsiders working alongside insiders thinking from first principles" + driving the cost-to-build-and-distribute-healthcare-products toward zero. This is, not coincidentally, the company-building thesis behind Superpower.
Mapping against Ray Data Co
Targeting-systems thesis convergence
This essay is a healthcare-industry-scale version of RDCO's targeting-systems frame:
- Targeting layer (industry): Primary care has the wrong target (visit volume, RVU billing) instead of patient outcomes. The reimbursement structure forces optimization on the wrong objective function.
- Sensors: Primary care's instrumentation (15-minute visits, annual labs only) is too low-cadence to detect the conditions Med 3.0 cares about (cardiovascular trajectory over decades, metabolic drift).
- Actions: Physicians are constrained to a narrow action space (prescription, referral, basic counseling) because anything richer doesn't get reimbursed.
- Feedback loop: Patients see their primary care physician 1-2x/yr. Outcome-attribution to specific actions is impossible at that cadence.
This is the same structural failure RDCO is solving in agent-COO work: low-cadence feedback loop, mis-aligned targets, too-narrow action space. The market response (DTC longevity platforms, employer-reimbursed integrated care, wearables + bloodwork-as-a-service) all attack one or more of these layers directly.
Implication for the founder's sensor strategy
Marchione's diagnosis validates the project README's decision to not rely on the standard primary-care visit as the primary sensor layer. The annual physical is necessary (covers basic insurance-paid bloodwork, builds physician relationship) but not sufficient for the targeting system the project is building. We need:
- Higher-cadence sensors: Whoop (daily), Apple Health (continuous), DEXA (annual), Function Health or Inside Tracker (quarterly).
- A Med-3.0-literate physician as the action-prescription layer, not a standard PCP doing 15-minute visits. Worth paying out-of-pocket for, even if it means buying a longevity-medicine practice membership at a clinic like Parsley / Forward / a local concierge MD.
- Independent data sovereignty: bloodwork PDFs ingested locally, structured longitudinal view in the vault, not trapped in a portal we don't control. This is the data-sovereignty thesis in the project's 2026-05-10 architecture note made operational.
The bull case for DTC longevity (Superpower, Function Health, etc.)
These businesses are the first-principles answer to the structural failures Marchione diagnoses. Their unit economics work when:
- Customer is paying out of pocket (no payer-set reimbursement)
- Sensor cadence is high (quarterly bloodwork → monthly data → ongoing-relationship business model)
- Action space is broader (peptides, lifestyle coaching, supplement protocols)
- Distribution is direct (no health system middleman)
But: their CACs are real, their evidence base for non-FDA-approved interventions is thin, and they're regulatory-arbitrage-fragile. The founder's posture should be: use these as one of multiple sensors, never as the sole physician relationship, and never as the action-prescription source for novel interventions without an independent second opinion.
Personal-OS analog
Marchione's industry critique - mis-aligned targets + low-cadence feedback loops + too-narrow action spaces - is the same diagnosis RDCO has made of the founder's pre-Whoop health-tracking posture. The health-and-longevity project is, structurally, an attempt to build for the founder personally what Marchione is arguing the healthcare industry needs to build at scale.
Related
- [[~/rdco-vault/01-projects/health-and-longevity/README]] - the project this informs; specifically the sensor-inventory section
- [[~/rdco-vault/01-projects/health-and-longevity/2026-05-06-attia-longevity-framework]] - Attia's Medicine 2.0 vs 3.0 framing, clinical-side
- [[~/rdco-vault/01-projects/health-and-longevity/2026-05-10-data-sovereignty-outcome-procurement-bet-architecture]] - the data-sovereignty thesis Marchione's industry critique reinforces
- [[~/rdco-vault/01-projects/health-and-longevity/2026-05-11-marchione-peptides-101]] - Marchione's peptide bull case (downstream of the structural critique here)
- [[~/rdco-vault/06-reference/2026-04-30-rdco-thesis-targeting-systems-feedback-loops]] - RDCO's frame this maps to
- [[~/rdco-vault/03-contacts/max-marchione]] - author profile