Privacy Policy
Last updated September 6, 2026
Levvi by Remi is a private accountability coaching app. This policy explains what it stores, why, and the control you have over it. It is written in plain language on purpose. The app is currently in a private, invite-only phase used by a small, known group of people.
Who we are
Levvi by Remi is operated by Levvi by Remi LLC, 12 Sloanes Ct, Port Washington, NY 11050.
Questions about your data go to remi@levvibyremi.com.
What we collect
Your account email and password. Passwords are handled by our authentication provider (Supabase Auth) and are never stored in readable form. If you sign in with Google or Apple on the web app instead, that provider handles the sign-in step and shares your name and email with us to create your account.
Your profile: name, age (on the web app) or exact birthdate (on the phone app), and gender, plus a longer set of optional background questions (where you’re from, education, family, occupation, interests, height and weight). The optional ones are genuinely optional: you can skip any of them and the app works fully without them.
What you enter in the app: your intake reflections, life-wheel ratings and focus areas, daily check-ins and day ratings, loves list, monthly audits, declutter entries, program participation (which session you’re on, your commitments and notes), and your answers and the resulting summary from any interactive program session.
Longer-form entries: goals with their steps and target dates, your daily routine and which items you completed, sealed time-capsule entries, Growth Guide journal entries (meditation and movement, a 1 to 3 mood rating for each, gratitude, and your written reflections), and weekly and yearly reviews.
Your conversations with the AI coach: both sides of every exchange, along with technical usage figures and whether an automated safety check was triggered. If you attach a photo in the phone app, the image is stored in a private bucket scoped to your own account; only the picture you pick is uploaded, the app never opens your camera or reads your photo library on its own.
If you install the phone app and allow notifications: a push notification token, platform, and device name for each device, plus a log of nudges actually delivered to you. Whether to send you a nudge, and what it says, is decided from your own recent activity in the app, so a nudge’s text can reference your own entries.
If you connect Google Calendar in the phone app (optional): the access grant Google gives us (a refresh token, the granted scope, and which calendar to read), used only to ask Google for your free/busy blocks for today. We request read-only access and never see event titles, guests, or descriptions, only busy time ranges. You can disconnect in the app or revoke access in your Google account at any time.
A record of your acceptance of the safety disclaimer, and a log of any automated safety checks on your account, visible only to the Levvi team for reviewing how the safety systems perform.
If you join the public waitlist instead of the app: the email address and optional note you leave there.
Basic technical information: timestamps on your records and short-lived counters that prevent abuse. As of the date at the top of this policy we run no third-party advertising or product-analytics software. The phone app checks Expo’s update service on launch, which tells that service your app version and device type.
How we use it
Only to run the app for you: to show your entries back to you, track your progress, and let the coaching feature respond to what you have written. Your information is not visible to any other user, and every table in our database is restricted so an account can only ever read its own rows.
We do not use your entries to advertise to you, and we do not sell your data or share it with advertisers or data brokers. Ever. If that ever changed, it would require an update to this policy, not a silent shift in behavior.
The AI coaching feature
When you message the coach, the text you send, together with a summary of your own profile and recent activity (including any Apple Health daily values you have connected, and any photo you attach), is processed by Claude, an AI model built by Anthropic, to generate the reply. An attached photo is also checked by an automated image safety review. Before a reply is generated, the app may include short snippets from Remi Stern’s own coaching material, chosen for the life area you’re focused on, to ground the answer in her real approach.
The coach is an accountability aid, not a person and not a licensed professional. It is not medical, psychological, or crisis care. In the web app, a "Get help" link on every page gives you real crisis resources (988 Suicide and Crisis Lifeline, Crisis Text Line, SAMHSA National Helpline) at any time. In the phone app today, those resources appear when the app detects crisis-related language; if you need them and do not see them, call or text 988 directly.
Automated safety scanning
Your messages are automatically scanned for language connected to crisis situations before a reply is generated. That scanning runs in two layers: a wording check inside Levvi’s own systems, and, when that check comes back clean, a hosted AI classifier run on NVIDIA’s service that receives the text being screened. If a check is triggered, the reply for that turn may be limited to safety and crisis resources, and a record of the check is kept for the Levvi team’s safety review.
The coach’s own replies are also reviewed automatically, but that review happens after the reply has already been shown to you: it is a record for our safety review, not a filter on what you see.
None of this scanning is perfect, and it is a safety net, not a monitored crisis line. It is not watched in real time by a person, and it can miss indirect or unusual phrasing. If you are in crisis, contact 911 or the 988 Lifeline directly.
How voice input works
The microphone button uses your own device’s or browser’s built-in dictation. Your words are converted to text on your device or by your device maker’s own service, and only the resulting text is saved to your account. We do not receive or store audio recordings of your voice, and no separate transcription service processes your spoken audio today. If that ever changes, this section will name the provider before the feature turns on.
Where your data lives
Your records are stored in a managed Postgres database (Supabase). Access is row-level restricted: the app only ever returns your own rows to you, and one account cannot read another account’s data.
A small number of outside services process data to run the app: Supabase (database, file storage, and login), Anthropic (generates coach replies, and receives your message text, any photo you attach, and the profile and activity summary the coach is given), NVIDIA (the hosted classifier used by the automated safety scanning), Expo plus Apple’s or Google’s notification services (delivery of push notifications, including the notification text, and the phone app’s updates), Google or Apple sign-in (only if you use it on the web app), Google Calendar (only if you connect it), ElevenLabs (only if spoken coach replies are turned on, to turn a reply’s text into audio; off by default), and Stripe (only once paid plans are live, for payment and billing). None of them are permitted to use your information for their own advertising or resale. We do not send account emails today; if that ever changes, this list will name the mail provider first.
Apple Health
If you use the iPhone app and choose to connect Apple Health, Levvi reads six kinds of Health data with your explicit permission: steps, resting heart rate, body weight, active energy burned, sleep analysis, and mindful minutes. Levvi only reads, it never writes to your Health app, and you can turn access off in iOS Settings at any time.
Your raw Health samples never leave your device: each metric is reduced to a single daily value on-device before anything is saved to your account. Those daily values become part of the personal summary the coach works from, so they are included in what Anthropic processes when you use the coach. They are never used for advertising, never sold, and never shared beyond what it takes to run the app for you.
Retention
We keep your data while your account is active, because noticing your own patterns over months is the point of the app; we do not auto-delete old entries on a fixed schedule while your account is open.
You can ask us to delete your data at any time, and we will delete it within 30 days of that request. If you simply cancel or stop using Levvi without asking for deletion, we keep your data for up to 2 years so you can pick back up if you return, and then remove it.
Your control
You can view and change your entries in the app at any time.
You can ask us to export or permanently delete your account and everything tied to it by emailing remi@levvibyremi.com. During this private phase these requests are handled directly and by hand.
Age, California, and international use
Levvi by Remi is for users 17 and older. Setup asks your age (web) or birthdate (phone) and blocks you from finishing setup or using the app if you are under 17 (a self-attestation, not document verification). An account can exist before that step is reached; if setup is blocked, the partial account is unusable and we delete it on request. The app is not directed at children, and we do not knowingly collect information from anyone under 13; if we learn we have, we will delete it.
As a small private product, Levvi likely does not yet meet the thresholds that make the California Consumer Privacy Act formally apply; the commitments here (no sale of your information, no advertising use) hold either way.
The app is built for people in the United States and is not directed at EU or UK residents. If that changes, this policy will be updated with the commitments those laws require first.
Changes
If this policy changes in a way that matters, the date at the top will change and, where appropriate, we will tell you in the app.
Contact
remi@levvibyremi.com