Data-Sovereignty + Outcome-Procurement Bet — Architecture (V1)
Why this doc exists
Ben + Michael Holzum (BIL, finance + healthcare) decided 2026-05-10 22:01 ET to "treat this as serious" after the day's compounding signals: Michael's Whoop coaching idea (~12:00) → patient-data-sovereignty thesis (~16:00) → Michael's outcome-procurement / compute-escrow / data-trusts insight (~16:53) → Solve Everything chapter 9 grounding (~21:00) → patient-incentive structural-innovation spitball (~22:01). Capital state: "some between us, not loads." Ray's role: harness + execution layer. Michael's role: industry knowledge. Ben's role: tech + operating ground.
This doc is the V1 bet architecture. Treats Michael's incentive-alignment spitball as load-bearing (it is) and walks through the four business-model variants with capital, speed, and regulatory profiles.
The structural insight (Ben's 22:01 ET spitball, sharpened)
Currently, three parties have skin in healthcare outcomes:
- Insurance saves money when outcomes improve (lower claims)
- Provider earns more under value-based contracts when outcomes improve
- Patient does the actual work (lifestyle change, med adherence, sleep, exercise) and receives only "feels better" as upside
This is the unaligned-incentive problem. The patient bears the cost of behavior change but receives no financial share of the savings their behavior produces.
The proposed structural innovation: combine three EXISTING pieces in a new way:
- Value-based care contract (provider takes downside risk on patient outcomes)
- Patient incentive program (Vitality-style rewards, but tied to actual measured outcomes, not gym-check-ins)
- Continuous-data-sovereignty platform (patient's vault contains the measurement signal; both patient and provider can verify outcomes without re-collecting)
Bundled, the unit economics flip: the savings pool gets explicitly split three ways. Patient earns a meaningful share. Provider earns under the value-based contract. Platform takes a share for facilitating measurement + payments + governance.
Precedents (so we know what's been tried + what hasn't)
What's been done:
- Discovery Vitality (South Africa, global) — pays members for fitness activity, healthy grocery purchases, biometric improvements. ~$1B+ revenue. Insurance-adjacent (Discovery Health is the carrier; Vitality is the rewards layer).
- Medicare Advantage VBID Model — CMS pilot allowing MA plans to give beneficiaries reduced cost-sharing or supplemental benefits for managing chronic conditions. Active 2017-present. Limited uptake.
- Healthy Hawaii / state Medicaid wellness demos — state-level pilots paying Medicaid beneficiaries for behavior change. Sporadic, mostly token amounts.
- Employer wellness programs (GE, Toyota, Walmart) — pay employees for biometric screening completion + improvement. Most are gym-membership-reimbursement-tier, not outcome-tied.
- Vitality + Apple Watch partnership — health insurer pays members for activity rings, including subsidizing the Apple Watch itself. Closest analog to what we're proposing, but limited to fitness activity not measured medical outcomes.
What hasn't been done at scale:
- Tying patient cash payouts to actual measured medical outcome improvement (HbA1c reduction, blood pressure normalization, cardiovascular event avoidance, etc.) inside a value-based care contract structure, with the patient as the data-source-of-truth via a sovereign-data platform.
That gap is the wedge. Each piece exists separately. The combination is novel.
Regulatory landscape (the hard part)
Three statutes shape what's possible:
Anti-Kickback Statute (AKS) — prohibits offering remuneration to induce referrals for items/services payable by federal health care programs. Direct cash-to-patient for "improving outcomes" can implicate AKS if structured wrong. Exceptions: the Personal Services and Management Contracts safe harbor, the Beneficiary Inducement CMP exceptions (low-value gifts, promote-access-to-care, financial-need-based).
Beneficiary Inducement CMP — prohibits inducements to Medicare/Medicaid beneficiaries that influence choice of provider. The "Promotes Access to Care" exception (added 2016) explicitly allows certain remuneration that improves a beneficiary's ability to access care or comply with treatment. This is the cleanest pathway for the patient-incentive piece.
Stark Law — prohibits physician self-referral for designated health services payable by Medicare/Medicaid. Less relevant since we're not structuring physician self-referral; but worth flagging if the platform itself becomes a referrer of services.
Open regulatory questions for legal research (Notion follow-up #3):
- Can a state-chartered data trust, structured as a beneficiary fiduciary, accept value-based care payments AND make patient incentive payouts WITHOUT triggering AKS? The answer probably hinges on whether the trust is treated as the "provider" for AKS purposes or as a separate fiduciary.
- What state trust statutes (Delaware, Wyoming, South Dakota are common candidates) best support this construction?
- Does the Promote-Access-to-Care exception cover outcome-tied rewards or only access-tied rewards (transportation, scheduling, etc.)?
These are not internet-research questions. They are healthcare-IP-attorney questions. Likely $5-15K for an opinion letter once the structure is more concrete.
Four business-model variants
Each has different capital + speed + regulatory exposure profiles. Pick one, or sequence them.
Variant A: Pure platform (SaaS)
Shape. RDCO + Michael build the data platform + measurement infra + patient-incentive-payout rails. License to providers who run their own value-based care contracts and use the platform to measure + pay.
Capital required. Low. Engineering only. ~$50-150K to MVP, ~$500K-1M for first 5 provider customers.
Time to revenue. 6-12 months (sell to a provider that already has VBC contracts).
Regulatory exposure. Low-medium. Platform is HIPAA Business Associate of the providers; not a covered entity. Patient incentive payouts run through provider, not directly from platform.
Margin shape. SaaS gross margins (70-85%). Take rate likely 5-15% of contract value at scale.
Risk. Slow scale. Each provider sale is a multi-month enterprise sale. Network effects weak (each provider's data is siloed unless we add a federation layer).
Michael fit. Strong — his finance + healthcare network is the sales channel. He's the go-to-market lead. Ben/Ray are the product + tech.
Variant B: Vertical operator (RDCO becomes a chartered provider)
Shape. RDCO + Michael register a healthcare entity in a friendly state, contract directly with payers (CMS Innovation Center programs first, commercial payers later), recruit a clinical staff or contract clinicians on a panel, run patient cohorts directly. Pay patients directly from the savings pool.
Capital required. HIGH. State licensure + malpractice insurance + Medical Loss Ratio reserves + clinical staff + tech + 18-24 month runway before first contract pays out. ~$3-10M to first revenue, depending on state + payer.
Time to revenue. 18-30 months.
Regulatory exposure. High. Now a covered entity, fully HIPAA-regulated, AKS / Stark / CMP all directly apply. Need a healthcare general counsel from day 1.
Margin shape. Clinical operating margins (5-15% net). High capital intensity per panel (~$50-150 per-member-per-month operating cost, $100-300 PMPM revenue).
Risk. Capital-intensive, slow, regulatorily complex. But: highest economic share if the bet works (you keep 100% of the operator margin). And: the cleanest patient-incentive payout structure (you ARE the provider, no third-party AKS questions).
Michael fit. Critical — without his industry knowledge + healthcare-finance background, this variant is uninvestable. He likely needs to be CEO or COO, full-time.
Capital fit. Bad. "Some between us, not loads" doesn't cover this. Would need to raise $5-10M Series A from healthcare-experienced VCs (a16z Bio, Define Ventures, Optum Ventures, etc.). Multi-quarter fundraise.
Variant C: Hybrid (network operator + platform)
Shape. Platform AND orchestrator. RDCO + Michael are the data + measurement + payment platform, AND we orchestrate multi-provider/multi-payer relationships (e.g., contract a network of independent primary care physicians, plug them into ACO-like contracts via our infrastructure, take a share of the savings flow). We don't deliver care ourselves; we orchestrate the economic + data flows that make outcome-tied payments work.
Capital required. Medium. Higher than A (need credentialing + payer-relations + legal infrastructure), lower than B (no clinical staff, no MLR reserves). ~$1-3M to first revenue.
Time to revenue. 12-18 months.
Regulatory exposure. Medium-high. Likely a Management Services Organization (MSO) structure. AKS + CMP carefully navigated via fee structure (per-member-per-month flat, not per-outcome bonuses to providers).
Margin shape. 20-40% gross margin on contract throughput (split with providers + patients).
Risk. Two-sided market: need providers AND payers AND patients enrolled. Slower than A, faster than B. The "pioneering provider" framing from Michael's original spitball lives here.
Michael fit. Strong — he runs the payer + provider relationships, leverages industry network. Ben + Ray run the platform + ops.
Capital fit. Medium. "Some + targeted angel/seed raise of $1-2M" plausibly covers MVP + first contract. Healthcare-experienced angels (former CMS officials, payer execs, healthcare-IT founders) are the right capital.
Variant D: Cooperative / member-owned (most aligned with thesis, most novel)
Shape. Patients are the structural owners of the platform itself, organized as a cooperative or mutual benefit corporation. Co-op contracts with providers + payers; surplus from value-based contracts flows back to members (patients) as dividends. Platform fees cover ops; profit is structurally member-owned.
Capital required. Medium-low. Co-op formation is procedurally simple in most states. Member equity isn't traditional VC-backable; need a different capital stack (foundation grants, mission-aligned debt, member capital contributions, possibly USDA/HHS grants).
Time to revenue. 12-24 months. Slower because the legal structure is unfamiliar to payers.
Regulatory exposure. High but DIFFERENT shape — co-op governance is well-tested but co-op-as-healthcare-contracting-entity is less tested. Multi-state operations require state-by-state co-op statute compliance.
Margin shape. Low-margin by design (surplus goes to members, not equity holders). Sustainable but not VC-returns.
Risk. Hardest to capitalize via traditional VC. Best-aligned with the data-sovereignty thesis (members structurally OWN the platform that holds their data). Highest ceiling on "trust capital" with patients (because they own it).
Michael fit. Possibly stronger than Variants A-C — he's not driven by VC-equity outcomes if his finance background trends toward institutional/mission-driven capital.
Capital fit. Bad for traditional capital, possibly excellent for mission-aligned capital (Robert Wood Johnson Foundation, Commonwealth Fund, California Endowment, USDA Rural Health, etc.).
Recommended sequence (V1 hypothesis, founder + Michael to pressure-test)
Phase 1 (months 0-6, $50-150K): Variant A as MVP. Build the data-sovereignty platform + measurement infra + 5 outcome-category dashboards. Customer-discovery with 5-10 ACO-REACH operators. Goal: one paid pilot at $10-30K.
Phase 2 (months 6-18, $1-2M raised): Pivot decision based on Phase 1 traction. If providers love the platform but won't take patient-incentive risk: stay Variant A pure-SaaS. If providers want us to operate the bundle: pivot to Variant C, MSO structure. If a mission-aligned funder shows up early: consider Variant D from the start.
Phase 3 (months 18-36, depends): Variant B becomes plausible only if we have proven economics in C and the market wants vertical integration. Most likely we stay in C and license to multiple payer-provider networks.
Variant D has its own track: worth a dedicated grant-application sprint (1-2 months of one person's time) regardless of which other variant we pursue. Mission-aligned capital is non-dilutive and the data-sovereignty thesis maps to multiple funders' agendas.
Capital fit, frankly
"Some between us, not loads" is most consistent with Phase 1 Variant A (founder-funded MVP) followed by a healthcare-angel + mission-funder mixed raise for Phase 2. Variant B is off the table unless we land a strategic partner (a payer co-investing for product reasons).
Realistic capital plan:
- Months 0-6: Ben + Michael self-fund $100K combined, plus founder time. Build MVP.
- Months 3-9: Healthcare-experienced angel round, $1-2M target. Use the Phase 1 traction as the deck.
- Months 9-18: Series A only if metrics justify, $5-10M from healthcare-specialist VCs.
What Ray builds (regardless of variant)
The platform layer is the same across A/C/D. Ray's contributions, in priority order:
- Data ingestion MCPs (Whoop already in flight; Apple Health, MyChart, Quest, pharmacy follow). Each is a Layer-5 starter-kit module that doubles as a paying-product feature. The MCPs become the patient-facing "I own my data" install.
- Outcome measurement engine — automatically computes the contracted outcome metrics from the raw data feeds (HbA1c trend, BP normalization rate, CV event rate, sleep efficiency). The audit-script pattern from /process-newsletter generalizes here as deterministic outcome verification.
- Cohort dashboards — provider-facing UI showing patient panels, current outcome trajectories, projected savings vs benchmark. Uses HQ-style Astro routes pattern.
- Payment orchestration — the savings-split engine that calculates patient/provider/platform shares per outcome event. Webhook-based, integrates with payer claims feeds.
- Patient-facing app (later) — the consumer surface that shows the patient their data, their outcomes, their earnings. Probably React Native, ships post-Phase-1.
Open questions for Michael (pre-call)
If Ben sets up the intro call, these are the questions whose answers shape variant selection:
- What's your appetite for full-time on this vs side-project? Variants B/C/D require full-time founder commitment. Variant A can be a side-project for 6 months.
- What's your capital state? Joint angel investment = Phase 1 self-fund. Otherwise need to plan for raise immediately.
- What payer relationships do you have today? Direct relationships with ACO-REACH operators or commercial payers shorten Phase 1 by 3-6 months.
- Are you regulated industry (FINRA, etc.) or free to start a healthcare entity tomorrow? Affects timeline.
- Do you want to be CEO or co-founder-not-CEO? Variants B/C need a healthcare-side CEO; Variant A could have Ben as CEO with Michael as advisor/cofounder.
- Reaction to the patient-incentive frame: is this load-bearing for you, or a nice-to-have? The whole thesis pivots on whether we can structure patient payouts cleanly through the AKS exceptions.
Next concrete steps
- Founder + Michael have the call. Ray preps talking points (separate doc) when greenlit.
- Healthcare-IP attorney consult ($5-15K opinion letter) on the AKS / Beneficiary Inducement structure. Should happen BEFORE Phase 1 MVP commits to a specific variant.
- Customer-discovery interviews with 5-10 ACO-REACH operators. Validate pain + WTP.
- Foundation funder mapping for Variant D track (parallel work, low cost).
- Whoop MCP Phase 0 ships tonight (in flight, background subagent). First Phase-1 building block on the board.
Related
- [[06-reference/research/2026-05-10-healthcare-outcome-procurement-pioneering-provider]] — the deep-research brief sizing the bid windows ($25-50M cardiometabolic + $20M LEAD ACO at maturity)
- [[03-contacts/michael-holzum]] — cofounder-candidate context
- [[06-reference/concepts/2026-05-10-harness-moat-two-layers-portability]] — the meta-thesis that frames Ray's role
- [[06-reference/concepts/2026-05-10-ray-architecture-introspection]] — Ray's stack as the harness layer
- [[06-reference/book-solve-everything-ch9-build-the-rails-2026-04-13]] — Diamandis/Wissner-Gross frame: outcome procurement / compute escrow / data trusts as the rails healthcare needs
- [[06-reference/concepts/2026-04-24-targeting-system]] — the canonical RDCO framework being applied here
- [[06-reference/concepts/2026-04-23-unhobbling]] — what "unhobbled patient" actually means (data sovereignty + outcome verification + financial alignment)
- [[01-projects/health-and-longevity/]] — parent project folder