The Kiribati Health Commons

An open-source, AI-managed commons — the programs, people, ideas & health record of Kiribati, searchable in plain language (working title)

Nobody — not the ministry, the donors, or the NGOs — actually knows what has been tried on these islands, who did it, or how to reach them. The Commons preserves the who's, the what's, the how's, and the what-could-be's — for everyone working in Kiribati, forever.

Overview

Kiribati's health actors — government ministries, NGOs, churches, and visiting teams — operate blind to one another, and blind to everyone who came before them. When Oceans of Gratitude began its work, we tried to build on programs that had operated on these islands decades earlier — and spent weeks chasing disconnected phone numbers and dead email addresses of organizations that no longer existed. Their knowledge, their relationships, their hard-won lessons: gone. This is the deepest failure in Kiribati's health landscape — not missing data, but severed institutional memory. Every organization keeps its own records in its own format; staff rotate home and their contacts and ideas leave with them; official statistics capture only the fraction of reality that ever reaches a clinic; and every new effort starts from zero.

The Kiribati Health Commons replaces this with shared national infrastructure: an open-source platform, maintained by AI, that institutionalizes and preserves the who's, the what's, the how's, and the what-could-be's — searchable in plain language, English or Kiribati, with real-time answers grounded in everything ever recorded, and actively coordinating the efforts of everyone working today. Oceans of Gratitude brings the assets that make this buildable where others would fail: the founding dataset (patient records and the island's first surgical registry from our April 2026 mission of ~900 patients), and years of earned trust with the local institutions whose adoption decides whether any system lives or dies.

Four Commons, One System

  • The Program Registry — who, what, how, and who to call. A living directory of every NGO, ministry program, and care effort operating in Kiribati — and every one that came before: what they did, what they found, what worked, and current points of contact for each. No successor organization ever again spends weeks hunting dead phone numbers; the institutional memory of a nation's health work, preserved and current.

  • The Project Hub — coordination, not just memory. A live view of every current effort: who is working on what, where, right now. Normally disjointed efforts from separate NGOs coordinate here — the AI flags overlapping projects before money is wasted, surfaces complementary ones before opportunities are missed, and keeps the whole network of service providers and supporters in the loop, working with and off each other instead of in parallel ignorance.

  • The Living Diary — ideas that outlive their authors' postings. A shared space where NGO workers and clinicians record their ideas for programs, improvements, and efficiencies — the what-could-be's that today evaporate when a posting ends. Anyone wanting to take action need only search: the AI surfaces relevant ideas, prior attempts, and the people connected to them.

  • The Clinical Record — gated, separate, and sovereign. Patient data lives in a separate database with HIPAA-like access procedures: approved Kiribati clinicians only, explicit access gates on every patient-data section, identified views restricted to treating providers, full audit logging, and de-identified aggregate views feeding the open layers for planning.

Across all four, AI does the work no island program can staff: structuring, de-duplicating, and translating inputs (Kiribati ↔ English); maintaining registries and the contact directory; flagging follow-ups; and answering plain-language queries in real time with data and context. "What's been done for diabetes screening on Fanning in the last two years, and who ran it?" "Who is the current contact for the church clinic program on Washington?" "Show every patient flagged for follow-up since the April mission." It records, it answers, it remembers, it connects — and it coordinates.

How It's Built

  • Field layer: offline-first mobile and tablet clients — designed for atolls where connectivity is intermittent; structured forms plus free-text and photo capture; local storage with opportunistic sync.

  • AI layer: AI-driven ingestion (structuring, de-duplication, translation), natural-language query and report generation, follow-up flagging, directory maintenance. Model-agnostic by design; inference costs are modest at national scale for a population of 133,000.

  • Data layer — two tiers by design: an open tier (program registry, project hub, living diary, de-identified aggregates) and a physically separate clinical tier holding patient data behind HIPAA-like gates. Records aligned to FHIR/WHO interoperability standards; redundant hosting with an on-island cache so the record survives connectivity loss.

  • Access layer: role-based access control — clinical-tier access limited to approved Kiribati clinicians through formal approval procedures; open-tier access for ministries, NGOs, and providers; full audit logging of every query on both tiers.

Data Governance & Sovereignty

The data belongs to Kiribati. The Commons is designed for formal partnership with the Ministry of Health, and its two-tier architecture makes governance structural rather than aspirational: patient data is never merely access-controlled within a shared system — it lives in its own gated database, under HIPAA-like procedures, opened only to approved Kiribati clinicians, with consent captured at intake, de-identification as the default for everything leaving the clinical tier, and audit trails on every access. Open-source code means any party — including the government it serves — can inspect exactly how the gates work; it also means no vendor lock-in, no license fees, and a platform that cannot die with a grant cycle, because anyone can run it.

The Path to National Scale

  1. Seed (now) — Ingest April 2026 mission records and the cataract registry into the clinical tier; build the program registry and contact directory from our existing island network; open the query interface.

  2. Clinic pilot — Live capture at Kiritimati clinics and the December 2026 dental mission; clinician approval procedures stood up; living diary opens to practitioners.

  3. Partner onboarding — Ministry of Health, NGOs, and church providers join the open tier; shared registries (hepatitis screening, immunization, cataract) go live.

  4. National — Fanning and Washington capture; Tarawa and outer-island expansion in partnership with the ministry.

Why This Succeeds Where Systems Fail Here

Health-data projects in the Pacific fail predictably: designed in distant capitals, dependent on constant connectivity, staffed around administrative burden no island can sustain, and abandoned when funding ends — taking their contacts, lessons, and ideas with them. The Commons inverts each one. It is offline-first because the atolls are. AI absorbs the administrative load instead of assuming clerks who don't exist. Open-source removes the license-fee death spiral. It is built to outlive every organization that contributes to it — including ours: when any program ends, its who's, what's, how's, and what-could-be's remain, searchable, for whoever comes next. And adoption — the thing no architecture can buy — rides on the earned trust that brought nearly 20% of an island's population to our first mission.

A Bureaucracy Hack, By Design

The Commons follows the playbook Marina Nitze and Nick Sinai codified in Hack Your Bureaucracy, applied to a national health system. A formal ministry IT modernization would take a decade, die twice in procurement, and arrive obsolete; instead, the Commons starts where permission already exists — our own mission data and registries — delivers something people actually want on day one, and lets adoption, not mandate, carry the change. It makes yes easy: free, open-source, no procurement, no license, no vendor to negotiate with. It treats the bureaucracy as people, because it is one — the program registry and contact directory are stakeholder mapping made permanent, so the next reformer/provider/innovator inherits the who and not just the what. And it hands every future bureaucracy hacker in Kiribati the tool the authors would prescribe: before you fight the system, search what has been tried, find who did it, what they learned, and start from there. We did not have that tool when we started. Everyone after us will.

What This Gives Government & Global Partners

For the Government of Kiribati, the Commons delivers functionality no ministry in a nation of 133,000 could ever have dreamed of offering: real-time national registries, a live provider and program directory, plain-language health intelligence, and follow-up tracking — capabilities that would cost a wealthy country's health system millions, delivered at effectively no cost and owned by Kiribati itself. For the WHO, the World Bank, UNICEF, and international NGOs, it offers something they have never had here: a live, ground-truth picture of the issue and care landscape in Kiribati — what the burden actually is, who is working on it, what has been tried, and what it produced — replacing episodic assessments from teams that rarely leave the capital. Better visibility means better-targeted funding; better-targeted funding means the aid paradox begins to close.

Built to Scale Beyond Kiribati

Nothing in this architecture is specific to Kiribati except its founding data. The same failure — siloed actors, severed institutional memory, statistics blind to most of the population — defines dozens of small island developing states and low-resource health systems worldwide. Because the Commons is open-source, any country, ministry, or coalition can fork and deploy it: the offline-first field layer, the AI management, the two-tier privacy architecture, and the registry–hub–diary structure all travel. Kiribati is the proving ground; the product is a template. A funder here is not buying one nation a database — they are underwriting a replicable model for institutional memory and coordination in the world's hardest-to-serve health systems.

Program Funding (Annual)

  • Open-source platform deployment, hosting & AI tooling — $5,000–$10,000

  • Field tools — tablets & connectivity for clinics and missions — $3,000–$6,000

  • Documentation coordinator (part-time, island-based) — $4,000–$8,000

  • Partner onboarding & training — ministries, NGOs & providers — $3,000–$6,000

  • Total annual program — $15,000–$30,000

Support This Program

  • $10,000 deploys the platform — open-source infrastructure and AI tooling for the whole country.

  • $1,000 funds a month of island-based coordination.

  • $30,000 funds the entire program for a year — the backbone every health program in Kiribati, and eventually beyond, can build on.

  • Beyond funding: this project has specific, high-leverage needs — pro-bono engineering time, cloud and AI inference credits, and technical advisory on architecture and data governance. If that's you, contact us.

Donate Now → All contributions are tax-deductible. To direct your gift to this program, note the program name with your donation or contact us.

Program Updates

The latest from the field. Check back here as this program grows.

2026 — The founding dataset exists. Our cataract candidate roster — to our knowledge the first patient registry of its kind on the island — is live and growing, and becomes the Commons' founding dataset alongside records from the April 2026 mission of ~900 patients.