Of the community.
By the community.
For the community.
A health system run by the people who live in it. Care from neighbors. Membership owned by everyone who contributes. Physician-attested. Built in public on the backs of AI.
"Behavior is where it starts. Ownership is what makes it stick."
The thesis, in one paragraph.
The existing health system is built around hospitals, EHRs, payers, and insurers — none of which are the patient. The new infrastructure is built around the patient: their wallet, their identity, their vault, their context, their physician of record, their family timeline, their LMN, their attestation. Everything else is plumbing.
The infrastructure for personal and family health should be owned by the families who use it and the neighbors who deliver it. Not corporately owned. Not subscription-rented. Not investor-skimmed. Cooperative legal form (LCA), patron-member share, governance vote, patronage dividend. The network is structurally what it claims to be.
AI educates everyone — Sage answers the question, chanio compounds your understanding over time, every assessment surface explains what's happening in your body. But AI alone is not safe. The safety layer is physician attestation: every clinically-consequential AI output is reviewed and signed by a licensed physician, cryptographically receipted, audit-defensible. AI educates; physicians attest; you act with confidence.
And healthcare doesn't start with a doctor visit. It starts with what you do every day — what you eat, how you move, what you spend on, what you say to your AI, what you save to your wallet. The network meets you at behavior — and converts behavior into ownership: an attested record you own, a wallet you own, a member-share you own, a dividend you can accrue.
Why now.
The capability of healthcare is collapsing in cost and exploding in power — enough, in ARK Invest's framing, to begin turning sickcare into healthcare. Their Big Ideas 2026 projects the cost to sequence a whole genome falling roughly ten-fold to about $10 by 2030; the cost of frontier AI dropping 91% in eight months in 2025 alone; and roughly a third of FDA-cleared diagnostics becoming AI-powered by 2030, approaching all of them over time. AI is being adopted faster than the internet ever was. The deepest map of your own biology will soon cost about the same as lunch.
When intelligence gets that cheap, the AI is no longer the scarce thing — everyone will have it. What stays scarce is who owns the data, who is accountable when it is acted on, and who owns the infrastructure underneath. The existing system's answer to all three is: not you.
When the capability is nearly free, we own the part that stays scarce — a trusted relationship, a physician who is accountable, and an asset your family owns.
It is also a fork in how care gets delivered. ARK values the household-care opportunity at roughly $13 trillion — built on the 2.3 hours a day of unpaid care work families already do — and expects humanoid robots to reach human-level task performance around 2028. Every robotics company is now aiming a machine at exactly this work. The question is not whether care gets help from machines. It is whether the people who give care are replaced by a corporation's robot — or own the machine that helps them, as members of the cooperative. We intend to own it.
Five pillars.
Community ownership.
Cooperative legal form (Limited Cooperative Association). Patron-member share. One member, one vote. Patronage dividend allocated to members based on their actual use. The contributors are the owners. The owners cannot be bought out by an investor or a PE firm because the equity belongs to the members structurally — not contractually.
New infrastructure (not legacy retrofit).
Hospitals, EHRs, payers — that's the plumbing of the existing system. The new infrastructure is built around the patient: a wallet that holds their identity (Apple Wallet, ER-readable), a vault that holds their record (local-first, AI-readable), a ledger that signs every clinical output (cryptographic, audit-defensible), an agent layer that runs on their behalf across seven layers of life. Anyone can build on top. Nobody can take it away.
AI as universal educator + physician attestation as safety.
AI is the great democratizer of health knowledge. Sage answers questions. chanio teaches Socratically over time. Every assessment surface explains what's happening in your body, in plain language, anchored in evidence. But AI alone is not safe — and that is measured, not asserted.
On psychosis-bench, bare language models showed a strong tendency to perpetuate a user’s delusion at every turn rather than challenge it — offering a safety intervention in only about a third of the turns that warranted one. A harness helps enormously: a physician-built neurosymbolic stack took a small model from 30.2% to 95.8% on that same test. And it still tells its users it “never gives definitive diagnoses” — because a harness can make a model safer, but it cannot make it accountable. Accountability has an address. When China ran an open model across 300+ hospitals, Nature Medicine found it produced “plausible but factually incorrect outputs” — and that the liability landed on the hospitals and the doctors, not the product. It lands on whoever signed. So we make signing the design: licensed physicians review and sign every clinically-consequential AI output. Hashcare-receipted. IRS-defensible. FHIR-shareable. AI educates. Physicians attest. You act.
Behavior first, ownership next.
Healthcare doesn't start at the doctor's office. It starts with daily behavior — your food, your sleep, your movement, your spend, your conversations. The network meets you at behavior, then converts behavior into ownership: every action becomes an attested record, every record lives in your wallet, every wallet is a member-share in the cooperative. Compounding into ownership is what makes the behavior stick.
Scale through reciprocity, not extraction.
10 hours of mutual aid per member per year, multiplied by membership at planetary scale, multiplied by cooperative-ownership equity for every contributor. The structure of the most valuable company that has ever existed — because the value created (billions of hours of mutual aid coordinated, attested, and rewarded) flows back to the people who created it. There is no leak. Investor-owned companies cannot replicate this without dissolving their existing equity. Once the members own the cooperative, no one can buy it from them.
External validation.
June 10–15, 2026 · five voices, five days, one thesis
The week we drafted this, the largest frontier AI lab and the most-read AI columnists published our pillars back to us in their own vocabulary. We didn't change the manifesto. We just left it next to what they said.
“Equity-sharing mechanisms giving workers partial ownership in AI enterprises … sovereign wealth models … pre-distributive capital accounts.”
— Anthropic, An economic policy framework for the AI exponential (June 11, 2026) · $350M committed.
“I don’t prompt Claude anymore. I write loops, and the loops do the work. My job is to write loops.”
— Boris Cherny, head of Claude Code, Anthropic, quoted in The New Stack, “Loop engineering” (June 10, 2026). The architecture we’ve been building — scheduled, isolated, verified, persistent — just got an industry name.
“Smart judgment is arguably the most important talent right now.”
— Jim VandeHei, Axios C-Suite, “Why loops are the next level of your AI use” (June 15, 2026). Judgment is what physicians bring. Loops scale it.
“The companies building this technology have a responsibility to make sure the benefits are fully realized and widely shared, and to invest directly in the workers absorbing the change.”
— Anthropic, announcing Claude Corps (June 11, 2026). $150M for fellows embedded at the organizations doing the work. The behavior → ownership compound, restated.
“AI sovereign wealth funds funded by investment stakes in AI-driven productivity … new ways of sharing this broadly.”
— Anthropic, policy framework, on the >25% unemployment scenario. The structure has a name in policy. We have one running in care.
July 2026 · and then the care people said it too
The AI labs validated the structure. These three matter more. They are the people who ran American public health and American clinical AI, and they say the thing we are building is not merely possible — it is already technically possible, and the only reason it does not exist is the money.
“The technology exists to do essentially everything that we do now in that first primary care setting of healthcare in brick and mortar — and you can put it in the home. This is not science fiction.”
— Susan Monarez, PhD — former Director of the CDC; helped establish BARDA and ARPA-H. She also names the enemy plainly: “brick-and-mortar health care is one of the cruelest monopolies … if you cannot get there physically, you are denied care.” Public panel on prevention and civic health, July 2026.
“There’ll be a primary care visit in the future where you can have an agent personally at home that’s supervised by a doctor — but the doctor now can see a panel of 500, 1,000 patients as opposed to the 100 and 200 that we used to see.”
— Girish Nadkarni, MD — Chair of the Windreich Department of AI and Chief AI Officer, Mount Sinai Health System. The attested-agent architecture and its leverage math, from one of the first academic AI departments in American medicine. Same panel.
“For every dollar spent on public health, we can save $560 in health care costs … we spent $1 on water fluoridation and saved Medicaid $14 in children’s dental bills. You don’t get those kind of returns on Wall Street.”
— Nirav Shah, MD — former New York State Health Commissioner (citing Trust for America’s Health). He is right that you don’t get them on Wall Street. That is exactly the problem — and the reason for our legal form.
And it is not a theory. The proof has been running for twenty-five years — owned by the people who receive the care.
“The whole health care system created, managed and owned by Alaska Native people … a primary focus on building and maintaining relationships.” The patients are not called patients. They are called customer-owners.
— Southcentral Foundation’s Nuka System of Care (Gottlieb, 2013 — cited 238×). Emergency and urgent-care visits down ~50%, hospital admissions down ~53%, satisfaction ~97%, sustained for over a decade. A different context — a tribal health system on its own funding rails — but the same claim, proven: when the people who receive the care own it, outcomes go up and cost goes down.
The validation is not the win. The win is that the words we use are now the words a $500M-committed lab uses, in the same week — and that the model we are building has a quarter-century of peer-reviewed operating data behind it. The next sentence in this manifesto gets written in our own operating data — the first attested LMN, the first member-owned care visit, the first physician signature in a hashcare receipt.
Why the money can’t do it.
Here is the sentence that explains why a health system run by neighbors does not already exist. A former director of the CDC pitched exactly this — primary care, in the home, for 330 million Americans — to an investor she respects deeply. His answer:
“You cannot make me a 10x return of my investment — $500 million or more — in 18 months. You can’t do it.”
He was not being cynical. He was being accurate. Prevention returns $560 on the dollar — over decades, diffusely, to the commons. A fund has ten years and owes its partners a multiple. Those are not the same clock. Every honest venture investor in the country would give her the same answer, because the mismatch is structural, not moral.
So the blocker on home-based prevention was never the technology. The former CDC director says the technology is here. The chief AI officer of a major health system says the delivery system is here. The blocker is that the only capital in the room needs a clock that prevention cannot meet.
A cooperative does not need a 10x exit in eighteen months. Its owners are the patients — and their return horizon is their own life.
That is the entire argument for our legal form, and it is not an ideological one. It is asset-liability matching. Prevention’s payback period and cooperative capital’s patience are the same curve. We are not a kinder way to build this. We are the only clock that matches.
The investor had a second objection, and it is also correct: “This will work in the Bay Area. In Kendall Square. Maybe parts of New York City. You cannot scale this.” True — for a company. In 1935, nine in ten American farms had no electricity, because no utility could earn a return stringing wire to sparse houses. The farmers formed cooperatives and wired it themselves, and more than 900 of those co-ops still run that grid today. You do not scale prevention. You federate it.
The math.
If 1 billion people each contribute 10 hours per year of mutual aid through this network — at fair-market labor value (~$26/hr) — that is $250 billion per year of coordinated human care, owned by the contributors themselves. It produces three compounding revenue streams from the same labor unit:
- Patient pays ($19/mo retail, $0 for contributors who trade 10 hours).
- Payer pays ($5–15 PMPM for the SDOH attestation feed produced by every cared-for shift).
- Employer pays ($5–7/employee for the wellness benefit running on the same labor pool).
"The same shift pays twice."
The labor doesn't compete with care; it is care. The data doesn't extract from the patient; it belongs to the patient. The membership doesn't gate access; it grants ownership. Every step of the loop closes back on the contributor.
One sentence.
Cooperative health infrastructure, built on AI, attested by physicians, owned by the people whose lives it runs in.
Find out what's possible.
It starts with one question.
10 hours of mutual aid a year covers your ComfortCard membership, and gives you a share of the cooperative — vote, dividend, governance. Not a subscription. Not a club. A new layer of infrastructure for the people you love, owned by the people who built it.
Built in public, on the backs of AI, attested by physicians, owned by anyone who shows up.