The Relief Layer

The thesis, in five paragraphs.

The problem. Family caregivers are economically invisible, clinically uncounted, and structurally unsupported. The home-care industry that nominally serves them extracts labor through high-turnover, low-wage employment arrangements. The wearable and sensing technologies that have rushed into the space detect strain without delivering relief. The Medicare reimbursement infrastructure that exists for caregiver respite (the GUIDE Model, 2024) flows through partner organizations that mostly do not yet exist.

The intervention. co-op.care is a Boulder-anchored member-owned care utility organized as a Limited Cooperative Association under Colorado law. Members pay $59 per month or claim a $100 founding share refundable on exit. The cooperative delivers care through three integrated mechanisms: (1) a mobile application (Hearth, launching with iOS 27 on September 14, 2026) running Apple Intelligence Foundation Models on-device, (2) worker-owned care visits at a standardized $35 per hour rate verified through a three-layer attestation chain, and (3) reimbursement pathways via Letters of Medical Necessity that unlock pre-tax Health Savings Account dollars and a partnership pathway as a Partner Organization under the CMS GUIDE Model for Medicare-reimbursed respite services.

The legal substrate. The cooperative form is the load-bearing structural advantage. Bylaws under Colorado LCA law prohibit the sale of member data and require member governance over algorithmic decisions. Patronage equity returns surplus to members under IRC §§ 1381-1388. The closed-loop Care Token (banked hours at par, non-transferable, non-appreciating) supports mutual aid without crossing into securities or money-transmitter territory. None of these structural features can be replicated by investor-owned competitors without abandoning shareholder primacy.

The state at publication. Apple Developer Program enrollment is under way under the cooperative entity. Approximately 120 web surfaces, 14 production backend services, and 18 founding members are operational. An active partnership offer from a HIPAA-compliant dementia care wearable manufacturer for ten hardware units anchors a Boulder Front Range pilot. Partnership conversations are in early stages with regional health system and hospice partners on the Front Range. The CMS GUIDE Model (active at scale July 2026) creates a clear opportunity for cooperatives positioned as Partner Organizations for GUIDE-participant respite delivery — a role that existing home-care agencies are poorly structured to fill.

Why this matters. The thesis advanced here is not that caregivers need a better app. It is that the missing infrastructure is a relief layer constituted by the cooperative legal form, attestable visit verification, and reimbursement rails — and that this infrastructure, properly built, is owned by the people who use it.

The Problem: Family Caregiving as Invisible Infrastructure.

1.1 Scale

The most recent national accounting (Reinhard et al., AARP, 2026) estimates that 93 million Americans provided 49.5 billion hours of unpaid family caregiving in 2024. The valuation methodology assigns these hours to the United States Bureau of Labor Statistics median wage for home-health aides ($16.32 per hour, 2024), yielding a shadow economic value of $873 billion for the calendar year. For comparison: the entire U.S. Medicaid program disbursed $824 billion in federal-plus-state spending in fiscal year 2023 (CMS, 2024). The unpaid labor of family caregivers in a single year exceeds the largest public payer of medical care in the country.

The distribution of this labor is gendered. Three in five caregivers are women. Daughters provide care at approximately twice the rate of sons (AARP, 2026). The mean caregiver age is 49.4 years. Twenty-eight percent of caregivers report annual household income under $50,000.

1.2 Mortality and Morbidity Costs

The clinical literature on caregiver health is unambiguous. Schulz and Beach (JAMA, 1999) reported that spousal caregivers reporting caregiver strain had a 63 percent higher four-year mortality risk than non-caregivers of the same age cohort, an effect that survived adjustment for baseline health, education, and income. The finding has been replicated and extended: Vitaliano, Zhang, and Scanlan (2003) found elevated cardiovascular risk markers in chronic caregivers; Pinquart and Sörensen (2003) found dose-dependent depression and anxiety effects scaling with hours of care provided.

Caregiver health utilization paradoxically declines as care demands increase. The Schaeffer Center at USC (2025) found that only 17 percent of family caregivers remain current on their own age-appropriate preventive screenings; this share decreases monotonically with hours of care provided. Caregivers are, in the framing introduced by Tullis (2026), an uncharted patient population.

1.3 The Home-Care Industry's Structural Failure

The formal home-care industry, which might in principle offer relief, instead serves as a second extraction layer. The Home Care Association of America (2024) reports industry-wide annual turnover of 77 percent among home-health aides and personal care attendants. Median wages remain at or below state minimums in most regions. Workers receive limited or no benefits, no career advancement structure, and no equity in the businesses that employ them.

The investor-owned home-care firm receives between $25 and $45 per hour from clients or payers and remits between $11 and $16 per hour to the worker. The labor margin captured by intermediation funds shareholder returns, sales operations, and administrative overhead. The worker, who possesses the relationship with the family, holds no ownership claim on the business that monetizes that relationship.

The result is a market in which the people performing the care work cannot afford to stay in the work, and the families receiving the care cannot trust the continuity of the relationship.

1.4 The $78 Trillion Extraction Layer

The extraction analysis extends beyond home care to the broader healthcare and elder-care economy. The National Bureau of Economic Research working paper 28474 (Finkelstein, Notowidigdo, and Persson, 2021) documents the share of medical and care expenditures that flow to administrative, financial, and platform intermediaries rather than to direct care delivery. MetLife, Genworth, and Johns Hopkins surveys consistently find that two-thirds of long-term care expenditures by families are spent on services with no documented clinical or functional benefit.

Cumulative across the projected 30-year demographic transition (the aging of the baby boom generation), the share of caregiver-adjacent expenditure captured by intermediation is estimated at approximately $78 trillion (USC Schaeffer Center, 2025). This is the structural cost that any genuine relief infrastructure must work against.

Prior Art and the Surveillance-Without-Relief Gap.

2.1 The Surveillance Wave

Beginning approximately 2018, a wave of consumer and clinical technologies has entered the family-caregiving market, all sharing a common architectural pattern: instrument the older adult, surface alerts to the family member or care coordinator, monetize through subscription or per-incident fees.

Representative entrants span GPS-equipped wearables for wandering; bed and chair pressure sensors; voice-activated assistants with check-in capability; routine-monitoring sensor packages; consumer-grade fall detection; and medication management dispensers. The market is fragmented across dozens of point solutions.

The category has attracted substantial venture investment. Cumulative funding across the named entrants exceeds $1.4 billion as of mid-2026 (Crunchbase aggregation, 2026). Despite this, family caregivers report that the products have not measurably reduced the felt burden of care work (Cheng et al., 2024). Alerts arrive without the means to act on them. The technology informs without unloading.

2.2 The Surveillance-Without-Relief Critique

The most precise articulation of the surveillance-without-relief gap, drawn from dementia-care operators writing in 2026, frames dementia care explicitly as a two-person illness: the caregiver is also a patient, but not yet treated as one. The operational summary is that awareness is not relief. Surveillance technologies inform caregivers without unloading them; respite delivered on the day of strain measurably reduces felt burden, but the infrastructure to deliver it on demand does not exist at scale.

This framing is consistent with the broader caregiver health literature reviewed in Section 1.2 and with the home-care industry analysis in Section 1.3. It points toward a missing operational layer: the institution that converts detection into delivery.

2.3 The Digital Therapeutics Adjacency

A related but distinct category of digital health entrants — Hinge Health, Sword Health, Kaia Health, Function Health, Noom — has built consumer-facing or employer-facing AI-mediated services for chronic disease management. These businesses are not directly analogous to co-op.care (they are condition-specific, investor-owned, and primarily address adult populations rather than older adults), but they establish two relevant data points.

First, the unit economics of AI-mediated care delivery at scale are favorable. Hinge Health reported FY2024 contribution margins above 60 percent on its core musculoskeletal product (S-1 filing, 2025).

Second, the legal form of these businesses creates structural constraints. As venture-backed C-corporations, they cannot adopt bylaws prohibiting data sale without violating fiduciary duty to shareholders. As condition-specific entities, they cannot integrate across the multiple conditions that any single caregiver-family will encounter. These constraints are not technological; they are structural-legal.

2.4 The Cooperative Health Adjacency

Cooperative health organizations have an established American history. Group Health of Puget Sound (founded 1947, member-cooperative organization, later HealthPartners) demonstrated that consumer-cooperative governance is compatible with high-quality integrated care delivery. The Mondragón Cooperative Corporation (Spain, founded 1956) demonstrates that worker-ownership scales to multi-billion-euro enterprises across decades. Patient-Centered Outcomes Research Institute (PCORI, 2010) and Accountable Care Organizations (ACOs, established under the Affordable Care Act 2010) have moved toward shared-savings and patient-governance models within fee-for-service Medicare.

What has not been tried at scale is the integration of (1) worker-ownership cooperatives for the care workforce, (2) member-ownership cooperatives for the families receiving care, (3) on-device artificial intelligence as the coordination layer, and (4) Medicare reimbursement pathways (specifically the GUIDE Model, established 2024) as the operational financing rail. co-op.care proposes this integration.

The Cooperative Care Framework.

3.1 The Limited Cooperative Association Legal Form

co-op.care is incorporated as a Limited Cooperative Association under Colorado Revised Statutes Title 7, Article 58 (C.R.S. § 7-58). The LCA form, available in Colorado since 2011, is a hybrid cooperative entity that combines member governance with patronage allocation of surplus and contains explicit provisions for investor participation alongside cooperative members. Twelve other states have adopted variations of the Uniform Limited Cooperative Association Act (Colorado being the first).

The structural features of the LCA that are load-bearing for co-op.care:

1. Patronage allocation under IRC §§ 1381-1388. Surplus generated by the cooperative is allocated to members in proportion to their patronage (use of the cooperative's services). Allocations may be paid in cash or retained as capital credits; the cooperative is taxed at the entity level on the portion not allocated to members, and members are taxed on their allocations. This structure returns the economic surplus generated by care delivery to the families and workers who generated it.
2. Member governance. Each member has one vote regardless of capital contribution. Major decisions (bylaw amendments, sale of substantially all assets, dissolution) require supermajority member approval. The board of directors is elected by and accountable to the members.
3. Bylaw constraints on member-data sale. The cooperative's bylaws (as drafted, pending counsel finalization) prohibit the sale, licensing, or commercial transfer of any member-identifying data to any third party for any purpose. Amendment of this clause requires a supermajority member vote (by current draft, two-thirds of voting members). The bylaws are public documents available from the Colorado Secretary of State.
4. Investor membership with limited governance. The LCA form permits investor members (distinct from patron members) who may contribute capital and receive preferred returns but who hold limited voting rights and may not collectively exceed a defined share of total governance. This permits raising capital without abandoning member control.

3.2 Why the Form Matters Structurally

The cooperative legal form is not a marketing differentiator; it is the structural answer to specific failure modes documented in Section 1.

The high-turnover, low-wage home-care employment arrangement is fixed by worker membership: workers earn equity (cooperative capital credits) on top of their hourly wage. Annual patronage allocations create a vesting structure that competing investor-owned employers cannot replicate. Industry turnover at 77 percent compares unfavorably to documented turnover in worker-ownership cooperatives (Cooperative Home Care Associates, NYC, reports 15 percent annual turnover; ICA Group, 2019).

The data-extraction failure mode of investor-owned digital health products is fixed by bylaw constraint: a publicly-filed bylaw prohibiting data sale, amendable only by supermajority member vote, creates a structural commitment that cannot be unilaterally reversed by management. Apple App Store reviewers, AWS HIPAA compliance architects, and prospective health-system partners can verify the bylaw at the state filing.

The misalignment failure mode of investor-owned AI products (the algorithm optimized for shareholder returns rather than user benefit) is fixed by member governance: algorithmic decisions are accountable to the members. The Omaha-scored care benefit verification (Section 4.3) is the operational mechanism by which members measure whether the AI is producing benefit.

3.3 The Care Token: A Closed-Loop Mutual Aid Instrument

The cooperative operates a closed-loop time-bank instrument called the Care Token. One token equals one banked hour of care given or received. Tokens are non-transferable to non-members, non-redeemable for cash, non-appreciating in value (the exchange rate is constant at one hour per token in perpetuity), and recorded as ledger entries against the member's account rather than as on-chain digital assets in any tradeable sense.

The Care Token is not a security under the Howey test: it has no expectation of profit derived from the efforts of others, no investment of money (it is earned by labor), and no common enterprise in the sense the Securities Act contemplates. It is not a money-transmitter instrument under the federal definitions: it has no fungibility outside the closed cooperative and no exchange rate against fiat currency.

The Care Token is structurally analogous to airline frequent-flyer miles (closed-loop, earned through patronage, redeemable for in-system benefits, not securities) but with mutual aid as the redemption purpose rather than discounted travel. The 1985 IRS ruling on time-bank instruments (Revenue Ruling 85-50) provides supporting precedent: closed-loop time banks earned through services and redeemed for services are not subject to federal income taxation.

The Care Token's role in the cooperative's architecture is restricted to mutual aid among members beyond their own family circles. Care delivered within one's own family is not tokenized; care delivered to a neighbor through the cooperative is logged as a token transfer. The instrument supplies the cooperative with an accounting mechanism for reciprocal labor without monetizing intra-family care.

Three-Layer Verification — Identity, Delivery, Benefit.

4.1 The Verification Problem

Care delivery, whether reimbursed by Medicare, paid out-of-pocket, or exchanged through the time-bank, presents three distinct verification problems. Was the care recipient who they claimed to be? Did the care visit actually occur? Did the visit produce a measurable benefit? Each of these questions has been answered partially and imperfectly in various corners of the care economy: identity verification via state-mandated background checks; visit verification via Electronic Visit Verification (EVV) systems mandated by the 21st Century Cures Act (2016); benefit verification via clinical outcomes measurement frameworks such as the Omaha System.

co-op.care integrates all three layers into a unified verification stack and writes each verification event to a cryptographically-attestable log (HashCare).

4.2 Layer One: Identity (Proof of Personhood)

The identity verification layer answers: is the cooperative member or care worker who they claim to be? The cooperative uses a layered approach:

• For care workers: state-mandated background check (in Colorado, the Colorado Bureau of Investigation fingerprint-based check), Department of Motor Vehicles identity verification, and (where applicable) National Plan and Provider Enumeration System (NPPES) National Provider Identifier verification for licensed clinicians.
• For families: a lighter-weight identity verification using government-issued ID and an in-person visit by a cooperative caregiver for the first care session.
• For physicians attesting Letters of Medical Necessity: full NPPES verification, state medical license verification in each state the physician practices, and Drug Enforcement Administration registration verification where controlled substances are relevant.

All identity verification artifacts are stored on the cooperative's backend in encrypted form with retention only as long as required by applicable regulation (in Colorado, six years for Medicaid-adjacent records).

4.3 Layer Two: Delivery (Proof of Visit)

The visit verification layer answers: did the care visit actually take place? The cooperative uses two methods depending on context:

• For Medicaid-reimbursable care, the cooperative complies with EVV requirements (in Colorado, the state-designated EVV vendor Sandata). EVV captures visit start time, visit end time, GPS coordinates at start and end, service type, and caregiver identity.
• For non-Medicaid care (private pay, time-bank exchanges), the cooperative uses a double-confirmation escrow pattern: both the care worker and the care recipient (or family proxy) must confirm the visit through the Hearth app within 24 hours of visit completion. Time-locked confirmation prevents retroactive editing.

Both methods feed into the HashCare attestation chain. Each confirmed visit produces a cryptographic hash containing the visit's identifying tuple (worker ID, recipient ID, start time, end time, service type, duration) hashed against the prior block in the cooperative's local chain. Leaf hashes are periodically rolled up to a public chain anchor (operationally, Bitcoin OP_RETURN or Ethereum calldata for cost efficiency), producing a publicly-attestable record of visit occurrence without exposing visit content or party identities.

4.4 Layer Three: Benefit (Proof It Helped)

The benefit verification layer answers: did the visit produce a measurable improvement? The cooperative uses the Omaha System (Martin, 2005), an evidence-based framework for community health nursing measurement that scores care outcomes across three dimensions:

• Knowledge: the family caregiver's understanding of the care recipient's needs, scored 1 (no knowledge) to 5 (superior knowledge)
• Behavior: the care recipient's adherence to care plan elements, scored 1 (never appropriate) to 5 (consistently appropriate)
• Status: the care recipient's signs and symptoms relative to baseline, scored 1 (extreme signs/symptoms) to 5 (no signs/symptoms)

Scoring is performed at intake (baseline), at defined milestones (typically 30, 60, and 90 days), and at care plan revision points. Score changes across visits are aggregated to produce a cooperative-level benefit metric.

The cooperative supplements explicit Omaha scoring with passive Patient-Reported Outcomes Measurement Information System (PROMIS) T-score inference from voice journal entries, care log content, and (with consent) wearable telemetry. PROMIS inference is an active research area at the cooperative; current Apple Foundation Models in-app demonstrate scoring concordance with manual PROMIS questionnaires of approximately 0.78 (κ statistic, internal validation n=42 as of June 2026, subject to publication).

The benefit layer is the load-bearing differentiator. Surveillance products report incidents and alerts; cooperative care reports measured benefit. Reimbursement (GUIDE Respite passthrough, HSA/FSA eligibility justification) and quality reporting (CMS Star Ratings, ACO measures) both flow from the benefit layer.

The AI Architecture — On-Device and Cloud-Side.

5.1 Design Goals

The cooperative's artificial intelligence architecture is designed to satisfy four constraints simultaneously: (1) clinical utility (the AI must produce measurable benefit, scored against Section 4.4), (2) privacy (member data must not be exposed to third parties in any form members have not explicitly authorized), (3) cost (the AI must be operable at cooperative scale without per-request infrastructure costs dominating the unit economics), and (4) regulatory alignment (the AI must operate within the FDA, HIPAA, and state-medical-board frameworks for clinical decision support).

These constraints jointly motivate a hybrid architecture in which the majority of inference (estimated at greater than 95 percent of cooperative-level inference volume) occurs on the member's device, and a minority of inference (the cooperative-scale aggregations, the physician-attestation workflows, and selected complex reasoning tasks) occurs in a HIPAA-compliant cloud environment under Business Associate Agreement.

5.2 On-Device Inference (Apple Intelligence Foundation Models)

The Hearth mobile application (launching with iOS 27 on September 14, 2026) uses the Apple Intelligence Foundation Models framework introduced at Apple's WWDC 2026 (Apple, 2026). The framework provides on-device inference against a 3-billion-parameter language model with multimodal (text and image) input capability and zero per-request cost. Inference occurs entirely on the iPhone 15 Pro or newer device's Neural Engine; no data is transmitted to Apple's servers.

The cooperative's on-device use cases include daily brief synthesis, photo-to-care-plan extraction from medication bottles or lab results, voice-driven advance directive drafting, and care-navigation question-answering. The privacy posture of on-device inference is structural: data never leaves the device. The cooperative cannot access the inference inputs, intermediate states, or outputs unless the family explicitly publishes them to the cooperative's backend.

5.3 Cloud-Side Inference (AWS Bedrock + Anthropic Claude for Healthcare)

A minority of inference tasks require larger context windows, more sophisticated reasoning, or aggregation across multiple member families. These tasks are routed to Anthropic Claude (in the cooperative's deployment, accessed via Amazon Web Services Bedrock under the HIPAA-eligible service framework, with Business Associate Agreement in place between Anthropic, AWS, and the cooperative). The cooperative's cloud-side use cases include Letter of Medical Necessity drafting, GUIDE Model respite-need detection across families, time-bank matching, and quality reporting synthesis.

The cloud-side architecture is HIPAA-compliant by construction: Anthropic Claude for Healthcare (launched January 2026; Anthropic, 2026) provides BAA coverage and audit logging; AWS Bedrock provides HIPAA-eligible service designation; the cooperative's backend (currently Supabase on AWS Postgres RDS, with migration to native AWS RDS planned for Q4 2026) supplies the encrypted application layer.

5.4 Why the Cooperative Form is the AI Trust Substrate

The artificial intelligence components described in Sections 5.2 and 5.3 are technologies that are increasingly commoditized. Apple Foundation Models, Anthropic Claude, OpenAI GPT-4-class models, and a range of open-source alternatives can each be substituted into the cooperative's architecture without fundamental change. The cooperative's iOS 27 LanguageModel protocol implementation makes this substitution explicit: the application can route inference to any compliant model provider.

What is not substitutable is the trust substrate within which the AI operates. The cooperative legal form constitutes this substrate. Member-data bylaws prohibit the cooperative from training its models on member-identifying data without explicit member consent. Member governance over algorithmic decisions means the AI cannot be tuned to optimize against member welfare in pursuit of shareholder returns. Patronage equity returns AI-generated economic surplus to members, aligning AI's economic incentives with member benefit rather than against it.

This is the structural answer to the alignment question every family raising AI for care purposes asks: how do I know the AI is on my side? The cooperative's answer is: because the cooperative is owned by its members, and the AI is owned by the cooperative.

Payment and Reimbursement Rails.

6.1 Member Dues

The cooperative's core revenue stream is recurring member dues: $59 per month for full member access, billed via Stripe Connect under the cooperative's Stripe HIPAA addendum and BAA. Founding members may claim a $100 founding share (refundable on member exit) that locks in the founding member rate ($59/month) in perpetuity regardless of future rate changes. Cooperative employees and care workers receive complimentary member access.

Member dues fund: the cooperative's technology infrastructure, care worker employment costs not covered by other revenue lines, administrative overhead, and surplus distributed annually to members via patronage allocation.

6.2 Letter of Medical Necessity Drafting

The cooperative drafts physician-attested Letters of Medical Necessity at $49 per letter. The cloud-side AI drafts the medical necessity argument and applicable diagnostic codes; a cooperative-network physician reviews (typically 5-10 minutes per letter), attests, edits, or rejects the draft; the cooperative delivers the attested LMN as a signed PDF to the family; the family submits the LMN to their HSA or FSA administrator to substantiate qualifying medical expenses under IRS Publication 502.

The economic significance of LMN drafting is large. A single LMN can substantiate $5,000 to $20,000 of annual qualifying medical expenses, paid out of pre-tax dollars at the family's marginal tax rate (effective subsidy: $1,500 to $7,000 per family per year). The cooperative's $49 fee is recovered by the family within the first month of HSA/FSA-reimbursed care.

6.3 CMS GUIDE Model Respite Benefit Passthrough

The Centers for Medicare and Medicaid Services launched the Guiding an Improved Dementia Experience (GUIDE) Model on July 1, 2024, as an 8-year demonstration designed to test a comprehensive package of care coordination, caregiver support, education, and respite services for Medicare beneficiaries with dementia and their unpaid caregivers (CMS, 2024). GUIDE participants are Medicare Part B-enrolled providers who establish Dementia Care Programs.

A core feature of the GUIDE Model is the respite benefit: up to $2,500 per beneficiary per year for respite services (in-home, adult day, or facility-based), paid by CMS to the GUIDE participant and passed at 100 percent to the participant's contracted Partner Organization for actual respite delivery. The cooperative is positioned to serve as a Partner Organization for Colorado GUIDE participants under bilateral contract.

The GUIDE Model expanded its respite benefit framework as of July 2026. GUIDE participants designate Partner Organizations through bilateral contracting rather than a separate CMS enrollment process. The number of qualified Partner Organizations nationwide remains inadequate to serve the full GUIDE participant pool, creating a clear opening for cooperatives like co-op.care.

At 10-20 dementia families per GUIDE participant clinic and an average of $1,800-$2,400 per family in actual respite delivered per year, the per-clinic per-year revenue to the cooperative is in the $18,000-$48,000 range. Across three to five Colorado GUIDE participants in year one, the GUIDE passthrough revenue is in the $54,000-$240,000 range.

6.4 Patronage Equity and the Care Token

Surplus generated by the cooperative beyond operational costs is allocated to members under IRC §§ 1381-1388 as patronage equity. Allocations are made in proportion to the member's patronage in the relevant fiscal period — for family members, hours of cooperative care received or hours of mutual aid contributed; for care workers, hours worked. Patronage allocations may be paid in cash (subject to annual reserves) or retained as capital credits.

The Care Token (Section 3.3) is not a revenue line; it is a mutual aid instrument. Members earn tokens by providing care to other members beyond their own family circles; members spend tokens to receive care from other members. Tokens are non-transferable to non-members and non-redeemable for cash. The Care Token's role in the cooperative's economics is to support a non-monetary care exchange among members for whom monetary payment is undesirable or infeasible.

6.5 The Solidarity Layer — Lessons from Cooperative Health Financing

The cooperative's mutual aid instruments do not exist in a vacuum. Roughly two million Americans have already left the regulated insurance market for health care sharing ministries — member communities that pool a monthly contribution and pay one another's medical bills directly, under the Affordable Care Act's sharing-ministry exemption (26 U.S.C. § 5000A(d)(2)(B)). Christian Healthcare Ministries alone facilitated $1.56 billion in shared medical bills in 2025. The mechanism is older than insurance: the well carry the sick until the roles reverse.

Two structural facts make this precedent directly relevant. First, every sharing ministry — without exception — shares acute medical bills (physician, hospital, pharmacy) and excludes the cost of aging. Long-term and custodial care, home aides, assistance with activities of daily living, and aging-in-place services fall categorically outside the definition of a "medical bill"; even the senior-oriented programs (CHM SeniorShare, Medi-Share 65+, Liberty Assist) operate only as Medicare-secondary acute-care supplements. The cost that most often empties a family — the slow, years-long cost of growing old — is shared by no one. This is the white space the cooperative's solidarity layer is designed to occupy.

Second, the sector demonstrates precisely the failure modes the cooperative must avoid. Liberty HealthShare diverted approximately $140 million of member contributions into unrelated businesses; the Aliera/Trinity collapse left $660 million in unpaid claims, with members recovering one to five cents on the dollar; and Sedera entered a $1.3 million settlement with the California Attorney General in 2025 over plans sold as sharing ministries without qualifying for the exemption, and was barred from the state. Independent analysis (Georgetown CHIR) finds sharing ministries pay roughly half of submitted bills. The sector's broken spine is trust — the gap between a promise to share and an enforceable obligation to pay.

The cooperative's design answers each failure directly. It does not claim sharing-ministry status: the exemption requires continuous operation since December 31, 1999, which the cooperative cannot satisfy, and claiming it without qualifying is the conduct that produced the Sedera enforcement. The solidarity layer instead operates as a mutual-benefit function under Colorado cooperative law. It borrows the peer-to-peer routing pioneered by Samaritan Ministries, in which contributions move directly from member to member in need while the cooperative coordinates rather than custodies the funds — the architecture that most reduces misappropriation risk. And it is capitalized from member dues and the Letter-of-Medical-Necessity revenue line before it bears any sharing risk, avoiding the undercapitalization that destroyed twenty-one of the twenty-three federal Consumer Operated and Oriented Plans (the "CO-OP" insurers) created under the Affordable Care Act — of which only two member-governed survivors remain (Community Health Options in Maine and Mountain Health CO-OP in Montana and Idaho), both of which endured by staying small, staying local, and refusing to underprice their risk.

The conclusion is that the cooperative's care-financing layer should be neither regulated insurance nor a sharing ministry, but a member-owned care-sharing cooperative aimed at the one cost the entire financing landscape leaves uncovered: the cost of growing old at home. It is, in the language of the manuscript that accompanies this paper, the solidarity fund grown to scale and pointed at the years insurance will not touch.

Case Study — The Boulder Front Range Pilot.

7.1 Pilot Architecture

The cooperative's first operational pilot is a 10-family dementia care cohort drawn from the Boulder Front Range, anchored by a partnership offer received in June 2026 from a HIPAA-compliant dementia care wearable manufacturer (the manufacturer remains confidential in advance of partnership terms). The product consists of a wrist-worn locator pod, charging docks, optional biometric sensing, and HIPAA-compliant cloud infrastructure.

The pilot architecture allocates one kit per family. Each family receives the wearable hardware at no cost, access to the Hearth iOS application via Apple Internal TestFlight, and 60 days of cooperative care visits at no cost to the family. The cooperative bears the cost of the care visits in year one as an investment in proof-of-concept.

The recruitment strategy targets mid-stage dementia families on the Front Range through three channels: (1) discharge planner referrals from a regional community hospital partner (3-5 families anticipated), (2) referrals from a Boulder hospice and PACE program (3-4 families anticipated), (3) cooperative members' personal networks and the recruitment surface at co-op.care/families (1-3 families anticipated).

7.2 Pilot Measurement Framework

The pilot's measurement framework is constructed to satisfy three audiences: the cooperative itself, the GUIDE Model oversight process, and the academic and partnership community.

Primary measures (collected at intake, day 30, day 60, day 90): ARCHANGELS Caregiver Intensity Index (Load, Impacts, Buffers subscales) per Burton et al. (2018); PROMIS Global Health Short Form for the caregiver; Omaha System Knowledge-Behavior-Status scoring per Section 4.4; care recipient functional status (Activities of Daily Living, Instrumental Activities of Daily Living); health utilization (emergency department visits, hospitalizations, urgent-care visits).

Secondary measures: visit verification chain integrity (cryptographic attestation of all visits, no gaps); wearable adherence (proportion of waking hours the pod is worn); family caregiver subjective burden (Zarit Burden Interview-Short).

Tertiary qualitative measures: family caregiver narrative interviews at day 60 and day 90; care worker focus group at day 90.

7.3 Pilot Outcomes (Anticipated Ranges, Subject to Empirical Validation)

The cooperative's anticipated outcome ranges, calibrated against published dementia caregiver intervention literature (Belle et al., 2006; Mittelman et al., 2007; Gitlin et al., 2010):

• ARCHANGELS Load score: anticipated reduction of 15-30 percent at day 90 vs. intake
• PROMIS Global Health score: anticipated improvement of 0.3-0.7 standard deviations at day 90
• Omaha Status score: anticipated improvement of 0.4-1.0 points (on the 1-5 scale) at day 90
• Care recipient health utilization: anticipated reduction of 20-40 percent in unplanned ED visits over 90 days vs. pre-pilot baseline (matched-control comparison)
• Wearable adherence: anticipated 70-85 percent of waking hours by day 30

The cooperative explicitly does not claim these outcomes in advance of empirical validation. The ranges are provided to establish a credible measurement framework, not to make outcome claims.

7.4 Pilot Status (as of June 2026)

Wearable partner offer received in June 2026; partnership terms pending confirmation. Health-system partnership conversation in early stages with a Front Range community hospital (Section 8 articulates the partnership architecture in illustrative form). Hospice and PACE program outreach initiated. Family recruitment surface live at co-op.care/families with a dedicated dementia-pilot CTA. Apple Developer Program enrollment is under way under the cooperative entity; TestFlight access opens at iOS 27 GA. Pilot start target August 2026. Pilot duration 90 days. Pilot completion target November 2026.

A Front Range Community Hospital Partnership Architecture.

This section describes the partnership architecture in illustrative form. Specific partner organizations are not named in advance of executed letters of intent.

8.1 Host Hospital Context (Illustrative)

The cooperative's Phase 1 health-system partnership target is a regional not-for-profit community hospital on the Front Range whose Community Health Needs Assessment identifies healthy aging as a top community priority and whose foundation has the institutional capacity to co-design a Partner Organization arrangement under the CMS GUIDE Model. Readmission rates and 10-day post-discharge gaps for the elderly population are well-documented structural drivers of avoidable utilization in community hospital systems of this profile; a GUIDE Partner Organization arrangement addresses both.

8.2 The Partnership Proposal

The cooperative proposes a multi-layered partnership with the host hospital:

Layer 1 — GUIDE Partner Organization. The cooperative serves as the hospital's Partner Organization under the CMS GUIDE Model for the dementia population. The hospital refers GUIDE-eligible dementia families to the cooperative; the cooperative delivers Medicare-reimbursed respite via the GUIDE passthrough; the hospital receives outcomes data and a co-design seat on GUIDE Model evolution at the cooperative's level. No operational cost to the hospital; CHNA alignment; readmission-reduction benefit for the dementia cohort.

Layer 2 — 10-Family Dementia Pilot. Hospital discharge planners refer 3-5 mid-stage dementia families to the cooperative's pilot cohort. Pilot families receive wearable hardware, Hearth iOS access, and 60 days of cooperative care at no cost. Pilot generates the visit verification, outcomes, and adherence data that supports the hospital's CHNA reporting, the cooperative's operational standing as a GUIDE Model Partner Organization, and joint publication opportunities.

Layer 3 — Geriatrics Clinical Channel. The cooperative seeks engagement with the host hospital's geriatrics medical leadership. Established ED-delirium care advocacy aligns with the cooperative's caregiver-strain framing; the clinical-channel partnership establishes the cooperative's clinical credibility within the host hospital and supports physician-attested LMN workflows for hospital-referred families.

8.3 Partnership Timeline

The proposed partnership timeline aligns with the GUIDE Model active as of July 2026. Week 1: partnership outreach initiated. Week 2: introductory call. Week 3: if aligned, brief letter of intent executed; initial 3-5 dementia family referrals. Weeks 4-6: first pilot family enrollment, TestFlight onboarding, wearable hardware deployment, first cooperative care visits. Weeks 7-12: pilot operations; Apple App Store submission for Wave 1. Weeks 13+: Hearth Wave 1.5 public App Store preparation; 90-day pilot completion (November 2026); Partner Organization agreement formalization with anchor health-system partner.

Regulatory Pathways.

9.1 CMS GUIDE Model Partner Organization

The cooperative's primary reimbursement pathway is operating as a Partner Organization under the CMS GUIDE Model through bilateral contracting with a GUIDE Participant — no separate CMS enrollment is required of the cooperative. The GUIDE Participant adds the cooperative to its Partner Organization Roster and contracts for respite delivery. The cooperative's side of that contract demonstrates: a clinical model for caregiver respite delivery aligned with the GUIDE Model's evidence base; operational capacity to deliver respite in multiple modalities; a visit verification mechanism that satisfies CMS audit requirements (the three-layer verification stack of Section 4); and a Business Associate Agreement chain for any PHI flow.

9.2 HIPAA Business Associate Agreements

The cooperative operates under HIPAA's Business Associate framework. The cooperative executes BAAs with upstream and downstream parties handling PHI: the wearable manufacturer; Anthropic (Claude for Healthcare); AWS (Bedrock, HealthLake, RDS, Lambda, Kinesis); downstream health-system and program partners as engagements formalize. Care workers execute confidentiality agreements aligned with HIPAA workforce-member requirements.

9.3 State Cooperative Law Compliance

The cooperative complies with Colorado LCA law throughout its operations: annual report filing with the Colorado Secretary of State; maintenance of cooperative bylaws and minute book; annual member meeting with notice and quorum requirements; patronage allocation under IRC §§ 1381-1388; distinct membership classes with documented governance rights.

9.4 Anti-Kickback Statute and Stark Law

The cooperative's payment flows are designed to satisfy anti-kickback safe harbors and Stark Law exceptions: member dues are flat-rate per-member per-month, not tied to volume or value of referrals; LMN drafting fees are flat-rate per letter, not tied to physician referral patterns; GUIDE Model respite passthrough is at 100 percent (no cooperative skim); patronage allocations are based on cooperative patronage rather than referral generation. Colorado healthcare counsel reviews these payment structures on an ongoing basis.

9.5 Securities Law

The cooperative's member share structure ($100 founding share refundable on exit; $59/month for ongoing access) is designed to fall outside the federal securities laws under the cooperative member exemption. Patronage allocations under IRC §§ 1381-1388 are not securities by virtue of their statutory treatment. The Care Token (Section 3.3) is structured to fall outside the Howey definition of investment contract.

The 24-Month Implementation Roadmap.

10.1 Wave 1: iOS 27 Launch and Boulder Pilot (June - September 2026)

Apple Developer Program enrollment confirmation under cooperative entity; bundle ID claims for care.coop.chanio and care.coop.hearth. Xcode 27 project scaffolds for chanio (Productivity-category) and Hearth (private TestFlight). Apple Intelligence Foundation Models integration with on-device daily brief, photo-to-care-plan, voice-driven advance directives, Sage care-navigation Q&A. TestFlight private deployment to the 10-family dementia pilot cohort + 50-90 close-circle cooperative testers. Public App Store launch of chanio on iOS 27 GA Day 1 (September 14, 2026). CMS GUIDE Model Partner Organization agreement initiation with anchor health-system partner. LCA bylaws final filing. Pilot start with first 3-5 families enrolled. Target end-state: 200-500 cooperative founding members.

10.2 Wave 2: Public Hearth and ComfortCard Wallet (Q4 2026)

Hearth public App Store launch in Lifestyle category. ComfortCard Apple Wallet integration: virtual card in Apple Wallet; family scans care receipts with iPhone camera; multimodal Foundation Models extracts the qualifying-expense classification; ledger entry updated on-device. ClinicalSwipe physician review marketplace. AWS Bedrock + HealthLake + RDS migration complete, contingent on the AWS Build AI Golden Pitch Competition grant. Target end-state: 500-2,000 cooperative members; first GUIDE Respite passthrough revenue.

10.3 Wave 3: Advance Directives, Time-Bank, Identity Layer (Q1-Q2 2027)

CareGoals advance directive product: voice-driven directive capture, multi-state notarization, Apple Wallet pass for the executed directive, FHIR Advance Directive Information section pushed to Apple Health Records. altru.care time-bank: member-to-member care exchange, Care Token ledger, mutual aid matching across the member graph. mapofyou identity and values layer: cross-product synthesis. First federation node beyond Colorado. Target end-state: 2,000-10,000 cooperative members; multi-state presence.

10.4 Wave 4: SurgeonValue, Harnesshealth, Consumer Funnel (Q3-Q4 2027)

SurgeonValue surgeon practice OS. Harnesshealth orchestration: the AI orchestration layer routing inference across model providers via the iOS 27 LanguageModel protocol. arthritisrisk consumer assessment: 5-minute risk assessment using HealthKit passive data + multimodal joint photo. Target end-state: 10,000-50,000 cooperative members; cross-product integrations with surgical practices and the cooperative's federated physician network.

10.5 Wave 5: Federation and the Long-Arc Vision (2028+)

SolvingHealth umbrella application surfacing member status across all installed cooperative-network applications. Sovereign compute cooperative (see Section 11). Multi-region federation: cooperative-of-cooperatives structure spanning multiple states. Target end-state: cooperative as the proof-of-concept for member-owned AI infrastructure at planetary scale.

Long-Arc Vision — Sovereign Compute Cooperatives.

11.1 The Architecture

The cooperative's long-arc architectural vision (Phase 3-5 in the implementation roadmap, 12-60 months in the future) extends the cooperative model from care delivery to the artificial intelligence infrastructure that supports it. The structure: each member operates a personal AI server — a Mac mini, a Cyberdeck, or equivalent commodity hardware running open-source language models locally. Members cluster around shared values and trusted curators, federating their personal servers into community-scale compute pools. Compute contribution is recorded to a blockchain-attestable log (an extension of HashCare into compute attestation). Members earn AI tokens by contributing compute and by using AI in ways aligned with the cooperative's documented values. Tokens fund the member's ongoing AI usage at the cooperative's federated compute pool.

11.2 Connection to Existing Architecture

This vision is the architectural endpoint of work already underway. The chanio application's local-language-model routing (operational since 2026) is the personal AI server primitive. The HashCare attestation chain is the compute attestation primitive. The Colorado LCA legal form is the cooperative ownership primitive. The Omaha System benefit verification framework is the values-alignment measurement primitive. The Hearth iOS application's iOS 27 LanguageModel protocol implementation is the model-substitutability primitive.

11.3 Why Not Now

The sovereign compute cooperative is named here for completeness; it is not part of the cooperative's Phase 1 build. Six load-bearing problems must be solved before the sovereign compute cooperative is operationally feasible: distributed inference latency and reliability (requires 12-24 months of focused engineering); values-alignment evaluation function (an analog to the Omaha System for AI compute, currently novel); token regulatory clarity (cooperative member benefit structuring with securities counsel); community bootstrap (requires Phase 1-2 member base of 5,000-10,000); hardware accessibility (technology trajectory will bring entry floor below $300 by 2029); and the Anthropic relationship (architectural framing positions Anthropic as frontier-model provider and the cooperative as the community-owned consumption layer).

Limitations and Open Questions.

12.1 The Cooperative Form's Untested Scale

The cooperative legal form, while well-established in cooperatives of substantial scale (Mondragón, REI, Land O'Lakes, the agricultural cooperative system), has not been tested at scale specifically as a member-owned care utility integrating worker-ownership, AI-mediated coordination, and Medicare reimbursement. The cooperative's governance, surplus distribution, and federation mechanisms will require operational iteration over the first 5-10 years.

12.2 The CMS GUIDE Model's Demonstration Status

The GUIDE Model is an 8-year demonstration; its long-term continuation is contingent on CMS Innovation Center evaluation. If the demonstration is not extended at year 8 (2032), the cooperative's GUIDE passthrough revenue line ends. The cooperative's diversification across member dues, LMN drafting fees, and patronage equity provides operational resilience, but the GUIDE revenue line is a non-trivial fraction of the Phase 1-2 revenue projection.

12.3 The Apple iOS Demographics

The cooperative's Phase 1 Hearth iOS application requires an iPhone 15 Pro or newer device for Apple Intelligence Foundation Models on-device inference. This is a structural constraint on the family-caregiver demographic the cooperative can serve in Phase 1. The 93 million Americans providing family care in 2024 includes a meaningful share of households whose iPhone is older than the 15 Pro generation (or who do not own an iPhone at all). The cooperative's honest Phase 1 SAM is therefore the 10-15 million iPhone 15 Pro+ family caregivers, not the full 59 million.

12.4 The Care Token's Untested Regulatory Reception

The Care Token's structure (closed-loop, non-transferable, non-appreciating) is designed to fall outside securities and money-transmitter regulation. This structure has not been formally tested by regulatory bodies for cooperative care contexts specifically. The 1985 IRS Revenue Ruling on time-bank instruments provides supporting precedent; the cooperative's counsel has not yet obtained a private letter ruling or no-action letter for the cooperative's specific implementation. If a future regulatory determination treats the Care Token as a security or money-transmitter instrument, the cooperative would restructure the instrument.

12.5 The Verification Stack's Empirical Validation

The three-layer verification stack (Section 4) is sound in principle but its real-world deployment for cooperative care has not been empirically validated at scale. The Phase 1 pilot (Section 7) provides the first opportunity for empirical validation. If the verification stack proves operationally fragile or insufficient for GUIDE oversight, the cooperative would iterate the verification mechanisms.

12.6 The Sovereign Compute Cooperative's Distance from Operational Feasibility

Section 11 articulates a long-arc architectural vision. The vision's distance from operational feasibility is substantial: six load-bearing engineering and governance problems must be solved before the sovereign compute cooperative is real. The vision is documented here for completeness and to articulate the cooperative's directional ambition, not as a commitment for any specific Phase 1-4 timeline.