Principles

Most consumer health apps optimize for engagement metrics that have nothing to do with whether you actually got healthier. Streaks, leaderboards, push notifications timed to the hours you're most vulnerable, dark patterns dressed up as motivation — the entire playbook is borrowed from gambling, not medicine. And on the back end, the same apps quietly sell your data to insurers, employers, and advertisers.

Kosei is not that. The list below is the list of things we will not build, in writing, so that future versions of us cannot quietly drift into them. If we ever break one of these commitments, it will be public, deliberate, and explained — not an accident.

And there is a positive commitment in here, not just prohibitions: when your contributions help advance science through opt-in, federated research, you share in the revenue. You are a participant in what we hope will become the world's largest distributed health study — not a data source for someone else's profit. See commitment 11.

Engagement that we earn through real outcomes is the only engagement we want. Anything else is a tax on the user, paid in attention and trust.

Why: Streaks turn a missed day into a punishment. Health is not a video game; the cost of a bad week should not be the loss of a six-month emotional investment. Streaks also reward consistency over correctness — they will happily congratulate you for ten days of behavior that is hurting you.

Kosei will measure trends and progress over time, but it will never assign a score to your unbroken chain of doing a thing.

Why: Comparing your sleep, your heart-rate variability, or your weight to other users invites exactly the wrong reaction at exactly the wrong time. Health is private and individual. Your baseline is your baseline. The only person you should be benchmarking against is yourself, six months ago.

No public rankings, no friend rankings, no anonymous percentile shaming. Aggregate research findings that the community voluntarily contributes to are different — those are described in the Privacy Policy and they are opt-in, off by default.

Why: A push notification that exists to pull you back into the app — rather than to deliver information you actually need at that moment — is a tax on your attention disguised as a feature. The vast majority of consumer-app push traffic falls into this category.

Kosei will send notifications when something genuinely time-sensitive has happened: an experiment window ending, a lab result back, a coach summary you asked to be reminded of. It will never send "you haven't opened the app in 3 days" or "5 of your friends logged a workout today" or any variant of that template, even if it would measurably increase re-engagement. The point is not to maximize re-engagement.

Why: The mobile app industry has spent fifteen years perfecting techniques to make products feel impossible to put down — variable reward schedules, infinite scrolls timed to dopamine response curves, friction added to the unsubscribe flow. The whole stack is well-documented and we are choosing not to use it.

No artificial urgency, no fake social pressure, no countdown timers on decisions that don't have real deadlines, no modal-stacked unsubscribe flows. Where we have tension between making the right thing easy and making the engaging thing easy, we will choose the right thing.

Why: The standard consumer-health business model — advertise low prices to users, sell their identifiable or re-identifiable records to insurers, employers, pharmaceutical marketers, or data brokers on the back end — is not a model we are willing to operate. The conflict of interest is too large and the consent too thin.

We will never sell, license, transfer, or otherwise hand off user-level data — records identifiable to a specific person, or reasonably re-identifiable when combined with other data — to advertisers, insurers, employers, pharmaceutical companies, data brokers, or any third party. Full stop.

Raw data from your connected providers (WHOOP, Withings, Oura, Garmin, Fitbit, Apple Health, Google Health Connect, Google Calendar, Microsoft Outlook, and any future integrations) is never shared with anyone outside Kosei, including researchers. The raw stream stays inside the trust boundary; the only thing that can leave is what you explicitly opt in to share, in the form described in the next commitment.

We continuously evaluate additional data sources based on scientific utility, security review, and member demand. Supported integrations may expand or change; connecting any source always requires your explicit authorization.

This commitment is distinct from — and stronger than — "federated research access," which is described in commitment 11. Selling user-level data is something we won't do at any price. Federated research is something we'll do only with your opt-in, only at the aggregate level, and with you sharing in the proceeds.

Why: Once an app is funded by advertising, every product decision quietly bends toward the advertiser. Engagement becomes the metric, not outcomes; attention becomes the inventory, not the user's time. This is a structural problem, not a discipline problem — even well-intentioned teams cannot fully escape it.

Kosei is funded by annual membership fees and, only with opt-in, research revenue sharing. We will not run banner ads, sponsored content, affiliate recommendations, or any other format where a third party pays to influence what you see.

Why: A health app that gets sold business-to-business to employers and insurers becomes, in effect, a surveillance tool for whoever wrote the check. Even with the best intentions on both sides, the user becomes the product and the buyer becomes the customer — and the product loses every internal argument.

Kosei will not be sold to employers, insurers, pharmacy benefit managers, or any organization that has a financial interest in the user's health outcomes — not in the consumer phase, and not until the product has earned enough trust to do so without compromising it. If that ever changes, it will be a deliberate, public decision with a written changelog and a fresh opt-in from existing users.

Why: Every analytics SDK loaded into a consumer app — Mixpanel, Amplitude, Segment, FullStory, the rest — exfiltrates user behavior to a third party that has its own incentives, its own data-retention policies, and its own track record of breaches. For a health app, the cost is asymmetric: the upside is convenience for us, the downside is your data sitting on someone else's server.

Kosei runs only first-party, server-side instrumentation. Our event taxonomy is open, the events list is published in the codebase, and nothing leaves our servers for an analytics vendor. No client-side trackers. No session replay. No fingerprinting.

Why: If leaving is harder than joining, we are not running a product, we are running a trap. The asymmetry between sign-up and account-deletion in most consumer apps is one of the more telling signals of how they actually feel about their users.

Account deletion in Kosei is a single primary action with a clear confirmation, no guilt copy, no "are you sure" stacks, no "we'll miss you" emotional friction, and no waiting period dressed up as a feature. Export your data, delete your account, walk away. We would rather lose you cleanly than keep you dishonestly.

Why: The wellness industry runs on health-adjacent claims that sound scientific but are not. Every Kosei surface that uses words like 'optimal,' 'biomarker,' 'recovery,' 'healthspan' carries a real scientific debt — we either back it up with citations to primary literature, or we do not say it.

Where the science is genuinely uncertain, we will say so. Where a claim is well-supported, we will cite it. Where Kosei is making an inference that goes beyond the literature, we will label it as inference. Confidence is a feature, not a decoration.

Why: For thirty years, the consumer health industry has treated user data as raw material to be extracted and sold. We think the right model is the inverse: the user is the participant, the user shares in the value created, and research access is granted only at the aggregate level with explicit consent. This is the part of the deal that makes the rest of it credible.

How research access actually works at Kosei. Where qualified researchers, public-health institutions, or companies want to study questions that your data could help answer, access is granted only through federated, aggregate query interfaces: queries run against the data inside our trust boundary and return only statistical answers — means, distributions, model parameters, effect sizes. User-level records never leave. Re-identifiable outputs are blocked at the query layer, not by good intentions.

Opt-in, off by default, granular. Research participation is a tiered, opt-in choice you make in settings, not a buried checkbox. You can turn off any tier at any time. Raw third-party provider data is excluded from research at every tier — it never leaves, even in aggregate.

You share in the revenue. When research access is paid for, the participants whose contributions made the answer possible share in the proceeds. The mechanism, accounting, and rate are documented in the Privacy Policy and will be reported transparently as the program matures. For the first time in consumer health, the user is on the same side of the table as the people studying the data.

The mission. If we get this right, Kosei becomes the largest voluntary, longitudinal, multi-modal health dataset ever assembled by participants who chose to be there. That is a public good — a study that advances human health while paying its participants. We treat it as an obligation, not a feature.

Implementation status (July 2026). Kosei is in founding-cohort rollout. The federated query interface, revenue-attribution accounting, and partner-vetting process are not yet built; we have zero research partners and have generated zero research revenue. We publish this commitment now so those systems are constrained by it from day one. As the program matures we will report transparently — opted-in participants, research access requests, dollars shared, technical interface status. Saying where we are is part of the commitment.

These commitments are published by Kosei and reviewed at major product milestones. Additions are easy; removals are public, deliberate, and explained.

If you believe Kosei is drifting from any of them, report concerns via the contact information in our Privacy Policy or Terms of Service.