Privacy Policy

Last Updated: November 13, 2026

Effective April 1, 2026
Starting with app version 1.2.7.
  • AI permission wording changes in version 1.2.7. The Terms and Privacy Policy agreement you tick before Get Started or Sign In also allows the App to send what you give its AI features (photo scans and scan corrections, voice logging and recipe import) to the outside AI services named in Section 6, for analysis. There is no separate question before these features first send, and they send only when you use them. Users who accepted earlier wording are not asked again; that earlier wording did not include this AI-sharing permission. The Bodydex Assistant and document import keep their own separate consent (Section 6.6).
  • Reports always go through our server. Support messages, bug reports and food-data reports travel through our own server on their way to our support database; there is no direct-database fallback. Messages saved to the outbox after a temporary delivery failure are retried. Oversized attachments and permanent validation errors require correction before resubmission. Once a queued message is more than 14 days old, another failed delivery attempt can stop automatic retries; the App marks it failed and keeps it available on the device (Section 3.3).
  • More nutrients written to Apple Health and Health Connect. With your permission, the App also writes vitamins, minerals and fat types from your food log to your health app, on your device (Section 5.2).
Starting with app version 1.2.6.
  • Setup screen counts. During setup the App adds to simple counts of how often each setup screen is reached. The counts carry no installation or account identifier and are kept only as totals (Section 3.2).
  • Genuine-app check. During setup, after the birthday question, the App contacts our server at attest.bodydex.app so that it can show it is a genuine copy of Bodydex. If that check is switched on, Apple or Google is asked to confirm it (Section 3.2).
  • Ad measurement, only if you agree. From app version 1.2.6 the App sends no install message to Meta. Instead it contains optional ad-measurement tools from Meta, from our measurement partner Tenjin and, on iPhone, Google Analytics for Firebase. They are switched off unless we switch them on for new installs, and even then they start only for a new install in the US, for someone 18 or older who agrees on the "Ad measurement" screen during setup. They run during setup and briefly afterward, then stop for good on that installation (Section 8.1).
  • The ad-measurement settings request. During setup, a new install in the US asks our own server for the current ad-measurement settings. The request carries no advertising, account or installation identifier (Section 3.2). It may also carry counts of a few fixed error codes, with no identifier.
  • The birthday question. If the birthday entered in setup shows an age under 13, the App first asks whether it is right. Once it is confirmed, the App stops and asks our server to delete what setup created (Section 12).
None of these 1.2.6 changes includes your logs or your health measurements. The counts and the genuine-app check are not sent when the country or region of your App Store or Google Play account is one listed in Section 11.6, or cannot be read.

1.Introduction

Bodydex ("the App") is operated by Daniel Bobunov ("we," "us," or "our"), an individual developer. This Privacy Policy explains how we collect, use, disclose, and safeguard your information when you use the App.

We are committed to protecting your privacy. Bodydex is built on a local-first architecture — the vast majority of your personal data is stored exclusively on your device and is never transmitted to our servers. Data leaves your device in the cases this policy describes: when you use a feature that needs online processing (such as AI food scanning or the Bodydex Assistant); when you choose to send something (for example, a bug report or a support message); and automatically, for a limited set of purposes: checking your subscription; in app versions before 1.2.6, the one install message described in Section 8.2; and from app version 1.2.6, the setup screen counts, the genuine-app check and the ad-measurement settings request described in Section 3.2. From app version 1.2.6, ad measurement (Section 8.1) sends data to ad companies only if you agree to it during setup.

This policy describes how the App handles your information. Where we rely on your consent — for the Bodydex Assistant, for document import (Section 6.7), for Apple Health / Health Connect access, and for the consumer health data those features send (Section 13) — we ask for it separately in the App, before the feature is used, and you can withdraw it as described in Section 11.7. Using the App is not itself consent to any of those features. When you tick the Terms and Privacy Policy agreement before Get Started or Sign In in app version 1.2.7 or later, you also allow Bodydex to send what you give the AI features — photo scans and scan corrections, voice logging and recipe import — to the outside AI services named in Section 6, for analysis; there is no separate question for these features (Section 6.6). Ad measurement (Section 8.1) has its own consent: the "Ad measurement" screen in setup and, on iPhone, Apple's tracking prompt. The App asks no question about the install message in Section 8.2, the setup screen counts, the genuine-app check or the ad-measurement settings request in Section 3.2: they are sent as this policy describes, and this policy is where we tell you about them.

2.Local-First Architecture

Your data belongs to you. Bodydex stores all core data in a local database on your device, protected by your device's built-in encryption at rest (iOS Data Protection / Android File-Based Encryption). We do not maintain a synchronized server copy of your tracking database, and we cannot access the records in the App's database on your device. We do retain the operational records this policy describes — for example subscription, rate-limiting and support records. What reaches our servers is what this policy describes: data from online features you choose to use, and limited data the App sends automatically: in app versions before 1.2.6 the install message in Section 8.2 and, from app version 1.2.6, the setup screen counts, the genuine-app check and the ad-measurement settings request in Section 3.2. Ad measurement (Section 8.1) runs only if you agree to it, and what it sends goes from your phone to the ad-measurement companies, not to our servers, apart from error counts and deletion requests (Sections 3.2 and 8.1).

The following data is stored on your device. It is not transmitted to our servers except as described elsewhere in this policy — for example, the AI-feature payloads in Section 6, the feedback and support submissions in Section 3.3, and the parts of it that features you choose send for you:

  • User profile information (name, email, physical attributes, goals)
  • All food logs, meal entries, and daily nutritional summaries
  • Weight logs, body measurements, and progress photos
  • Lab results — both the ones you enter yourself and the ones read from your own report that you chose to import, whether you photographed it or chose its PDF (Section 6.7) — with their dates, the lab, and the reference range printed on that report
  • Medication, supplement, and peptide schedules and adherence history
  • Recipes and custom food items
  • Exercise and workout logs
  • Dashboard layout preferences and app settings
  • Bodydex Assistant conversation history and memory
  • Notification preferences and schedules
  • Behavioral data the App keeps for its own suggestions (food input patterns, how often you use features)

This list is about the records the App keeps on your device. The setup screen counts described in Section 3.2 are a separate, limited record of how often each setup screen is reached. They include none of the logs, profile details or other records listed above, and nothing you entered.

Note: As Bodydex evolves, we may introduce optional cloud backup or synchronization of your tracking records, and social features. If and when such features are introduced, this Privacy Policy will be updated accordingly, and backing up or synchronizing your tracking records to a server will require your explicit opt-in consent.

3.Data We Collect

3.1 Data You Provide

When you use the App, you may voluntarily provide the following information, which is stored locally on your device:

  • Account Information: Name, email address (via Apple or Google Sign-In)
  • Physical Profile: Height, weight, age/date of birth, biological sex, activity level
  • Health Goals: Target weight, dietary goals, goal speed, diet type, motivations, obstacles, cooking frequency
  • Food Data: Food log entries, meal photos (for AI scanning only), recipes, custom food items
  • Health Tracking: Weight logs, body measurements, progress photos
  • Lab Results: Marker names, values, units, the date on the report, the lab, and the reference range it printed — typed in by you, or read from your own report that you chose to import as a photo or as a PDF, and then reviewed and saved yourself (Section 6.7)
  • Protocol Data: Medications (including optional prescribing doctor name), supplements, peptides, dosage schedules, adherence logs
  • User Content: Notes, recipe descriptions, coach conversation messages
  • Preferences: App theme, notification settings, dashboard layout, preferred units
  • Feedback: Feature requests, bug reports, and app satisfaction surveys (when voluntarily submitted)

Third-Party PII: The App allows you to optionally enter a prescribing doctor's name for medication records. The name you type there is stored in the App's database on your device. Copies of your records that you make or send (an Export Data file, a device backup, a record you ask the Assistant about) are handled as this policy describes for those features. We are not responsible for the accuracy of third-party information entered by users.

3.2 Data Collected Automatically

  • Device Identifier: A stable, app-scoped device identifier (iOS Vendor ID or a generated UUID) kept on your device. The App uses it to anchor your local profile and for abuse prevention, which is why it survives Delete All Data (Section 9); it is not sent to our servers as is. It is different from the pseudonymized device hash described in Section 3.3 — a one-way hash that accompanies feedback and support submissions, used for anti-spam limits and to associate support tickets with this installation. Both are pseudonymous — not intended to directly identify you — and neither of these two is used to track you across other companies' apps or websites. The ad-measurement tools of Section 8.1 use other identifiers, and only if you agree to ad measurement.
  • Subscription Status: Your subscription tier (Free or Premium, including trial status) is verified through RevenueCat to determine feature access. We store a cached status locally; the server verifies it periodically. From app version 1.2.7, during cancellation, if a configured store discount is available, the App may send our server your platform, plan and offer code with account authentication to check availability. Retention discounts are currently disabled, so this check creates no offer-history record; your selected cancellation reason is not sent.
  • IP Address: Cloudflare and our server infrastructure process your IP address and ordinary request metadata for delivery, security, abuse prevention, and operational logging, under their retention controls. We do not attach your IP address to your tracking data.
  • Pseudonymized Service Usage: When you use AI-powered features (food scanning, voice logging, the Bodydex Assistant, food search, recipe import), we record a pseudonymized event noting which feature was used and whether the request succeeded, errored, or hit a rate limit. For recipe imports that fail, the event additionally notes the source platform (e.g. TikTok, Instagram, YouTube) and a technical failure reason so we can detect and fix import breakages; the shared link itself is not stored in these events. The event is indexed by a one-way SHA-256 hash of your account identifier so we can compute aggregate metrics (daily active users, error rates, cost forecasts) without storing your raw identifier. These events do not contain food data, photos, message contents, or voice transcripts, and use a hashed account identifier instead of the raw identifier. The identifier is pseudonymous: it can link events from the same account and may remain personal data.
  • Google's components in the Android App: The Android App contains Google's own software for purchases (Google Play Billing) and for barcode scanning (ML Kit). These components record technical usage data about their own operation and send it to Google by themselves, under Google's terms. It is not sent to our servers, and we cannot list exactly what it holds. In local-only mode and after a confirmed under-13 birthday (Section 12), the App tries to empty what is waiting on the device before it is sent.
  • Setup screen counts (from app version 1.2.6): while you go through setup — including a setup started on an earlier version and finished on 1.2.6 or later — the App adds to simple counts so that we can see how often each setup screen is reached and where setups stop. No question is asked about them and there is no setting in the app to turn them off; they stop when setup ends. Relaunching the App or restarting setup can add to the same counts again, so they are counts of screen arrivals, not of unique people. What the App sends to our statistics server (funnel.bodydex.app) is only numbers to add to running totals — for example, "the birthday screen was reached once" — with the platform, the App version, the version of the setup flow, whether this run of setup was a fresh start, a resumed setup or a restart, and, on Android, whether the device identifies itself as an automated Google Play test device. From app version 1.2.7, on iPhone, the same field also marks a test copy of the App installed through TestFlight, for Apple's App Review or by us during development; the App reads this on the phone from the type of its App Store receipt, and an install from the App Store sends no such mark. The counts cover: each setup screen reached, from the first screen on — including the sign-in, Health Sync, body measurements, target weight and birthday screens — but never what you entered on it; the outcome of each Apple or Google sign-in attempt during setup (succeeded, cancelled or failed); what happened on the setup subscription screen (for example, a trial started, a purchase, a restore, an error, or the screen was closed or skipped); whether setup was finished; whether the prompt to resume an unfinished setup was shown and answered; and, if the Ad measurement screen was shown, which answer was chosen there (on iPhone, that Continue was tapped and then whether tracking was allowed; on Android, Allow or Don't allow). No installation or account identifier is sent with the counts, and no time or order from your device. After the birthday question, a counts request can also carry one app pass from the genuine-app check described next: a random value that is used once and names neither you nor your phone. We store screen increments as hourly totals, not individual screen-event records. Separate operational records note each request's outcome (for example accepted, rejected or rate-limited) and how many increments it carried, without an installation identifier. The hourly totals are deleted from our live database about 45 days after the hour they cover (recovery copies of that database expire within a further 30 days); daily totals made from them are kept indefinitely. A total can be small — late at night one person's setup may be the only one in an hour. These totals contain no installation or account identifier, and we do not attempt to identify people from them. Your IP address reaches Cloudflare with each request, as with any internet request; our server uses it only to deliver the request and as the key of a rate limit against abuse, and does not write it to the counts, to those operational records or to any log of its own. Counts start on the first screen of setup, before you accept the Terms, because they carry no account or installation identifier. They are not sent in local-only mode (Section 12); after the birthday question shows an age under 13 (counts not yet sent are then discarded); when the country or region of your App Store or Google Play account is one listed in Section 11.6 or cannot be read; or once setup is finished.
  • Genuine-app check and app passes (from app version 1.2.6): so that we can tell real setups from made-up counts sent by a script, the App can show our server that it is a genuine copy of Bodydex on a real device. No question is asked about this check. It runs during setup only, after you have answered the birthday question; nothing is sent for it before that answer, and nothing after setup ends.

    What the App sends first. The App sends a request to our own server at attest.bodydex.app, asking whether the check is switched on. That request holds: a format number, the App version, the platform, on Android whether the device identifies itself as an automated Google Play test device, which of our servers an app pass would be for, whether the App is asking to register or to fetch app passes, and that it is for setup. It holds no installation code, no account identifier and nothing you entered. Like any internet request it reaches Cloudflare with your IP address. If our server answers that the check is switched off, nothing more happens; the App may repeat this request later during setup.

    If the check is switched on. The App asks Apple (App Attest, on iPhone) or Google (Play Integrity, on Android) to confirm that it is the genuine App on a genuine device, and sends that confirmation to attest.bodydex.app. Apple and Google handle that request under their own terms: they learn that Bodydex asked for the check from your device. On iPhone the confirmation includes a key identifier that App Attest creates for the installation; the App sends it each time it fetches app passes, and our server uses it and does not store it. Our server then hands the App a small number of one-use app passes. One app pass can travel with a setup-counts request, so that our statistics server can tell that the request came from the genuine App. An app pass is a random value, different every time. It carries nothing that identifies you or your phone, and we keep no list of which phone received which app pass.

    What our servers keep for the check, when it is switched on. For each iPhone installation that fetches app passes, a counter under a one-way code that changes every day; it is removed on a schedule, normally within about two days (a failed cleanup can delay this), and the recovery copies of that database expire within a further 30 days. A one-way record of each one-time code our server handed out for the check and saw used, for up to about a day. A one-way record of each app pass that has been used, for about two days. Totals with no code at all: daily totals of the check's outcomes, kept for 400 days, and hourly totals of the counts that arrived with a valid app pass, deleted with the other hourly totals. On your device the App keeps app passes until they are used or expire, and on iPhone the key identifier; the next paragraph says when it clears them earlier. The App also keeps verification and retry information for the check on the device, and may send attest.bodydex.app a technical failure code if the check cannot finish.

    When it does not run, and what it is not used for. The check does not run in local-only mode, after the birthday question shows an age under 13 (Section 12), or when the country or region of your App Store or Google Play account is one listed in Section 11.6 or cannot be read; in the first two cases the App also deletes the app passes and the key identifier it holds, and Delete All Data deletes them too. The check is not used for advertising, for a profile of you, or to decide what you can do in the App. Its result changes nothing you see, and it never affects support messages or bug reports.
  • Ad-measurement settings request (from app version 1.2.6): during setup, a new install whose App Store or Google Play account and phone region are in the US asks our own server (main-api.bodydex.app) for the current ad-measurement settings: whether ad measurement is switched on for new installs and for which ad companies, and whether the network location of the request is in a place where it is offered. It may ask again while setup is open, and once as setup ends. The request carries no advertising, account or installation identifier and nothing you entered, and it goes to no other company. Like any internet request it reaches Cloudflare with your IP address; our server uses the IP address as the key of a rate limit against abuse and the network location only to answer "offered", "not offered" or "unknown", returns neither your country nor your state, and writes none of it to a log or a record. The App keeps the answer in memory only. Apart from that last request, it is not sent once setup has ended, and never after you have declined ad measurement, in local-only mode, or after the birthday question shows an age under 13. The response helps decide whether ad measurement can run and, on Android, whether the Ad measurement screen of Section 8.1 is shown; on iPhone, it does not decide whether that screen appears. The request is not ad measurement itself. Error counts (from app version 1.2.6). The request can also carry counts of a few fixed error codes, for example that the answer could not be read or, if you agreed to ad measurement, that a tool failed to start; never an identifier, an error message or anything you entered. If a tool fails to start, the App may send one extra request with them shortly afterward and one as setup ends. Our server keeps these counts, and how each request was answered, as totals with no identifier and without your IP address, for about three months.

3.3 Feedback & Support Submissions

When you submit a bug report, feature request, comment, or support request in the app (including replies in a support conversation), we collect:

  • The text you typed (title and description)
  • Any screenshots you chose to attach (optional)
  • Any photos of a product label you chose to attach to a report of incorrect food data (optional). They are stored with the report on our backend and in our internal support tooling so we can check the correction against the label. The backend copy follows the 180-day limit below; copies in our internal support tooling are kept as long as needed to resolve the report
  • Your email address, only if you typed one (optional, used for follow-up)
  • App version, platform, operating system, and OS version (auto-attached for diagnostics)
  • Your subscription status and the store it was purchased through (auto-attached, so we can prioritize and troubleshoot billing issues)
  • A pseudonymized device hash (one-way SHA-256, used for anti-spam and to securely associate support tickets with this installation)
  • Your account identifier, only if you are signed in (so we can match feedback to your account if you ask us to)
  • A push notification token, only if you have allowed notifications for the App (used solely to notify you when we reply to your submission; the notification is relayed through Expo's push notification service to Apple or Google for delivery; while you are signed in, if you allow notifications after sending, or your phone's push token changes, the App also adds the current token to your recent open submissions)

Support messages, bug reports and food-data reports are sent whenever you send them. The country list in Section 11.6 and the genuine-app check in Section 3.2 do not apply to them. From app version 1.2.7 they are always sent through our own server, which runs on Cloudflare (Section 8), on their way to our backend; there is no direct-database fallback. Messages saved to the outbox after a temporary delivery failure, for example because the phone is offline or our server cannot take one at that moment, are retried. Oversized attachments and permanent validation errors require correction before resubmission. Once a queued message is more than 14 days old, another failed delivery attempt can stop automatic retries; the App marks it failed and keeps it available on the device. A message still waiting is sent one more time when you choose Sign Out, Reset & Start Over or Delete All Data; Delete All Data then removes waiting messages, support form drafts and screenshot copies from the device, sent or not. What it asks our servers to delete, and what stays there, is described in Section 11.2. Support messages are off in local-only mode, where the App says so, and after a birthday showing an age under 13 has been confirmed (both in Section 12).

Submissions are stored on our backend (Supabase, US region); that copy is scheduled for automatic deletion after 180 days. If we cannot safely verify ownership of an older attachment, including where there are too many attachment references for automatic verification, we retain the report, our replies and its attachments beyond 180 days until the issue is reviewed and deletion can safely be completed. A closed report normally leaves the list in the in-app inbox 30 days after it was closed, once you have read our replies; that does not delete the stored report. So that reports actually get seen and fixed, submissions (including any screenshots you attach) are also delivered to the developer through an operator notification service (Telegram Bot API, encrypted in transit) and retained in our internal support tooling for as long as reasonably necessary for support, security, or legal purposes. If you have allowed notifications for the App, we use your push token to notify you when we reply. Switching notifications off for Bodydex in the phone's settings stops notifications from being shown; it does not remove the push registration. If you had allowed notifications, the notification software inside the App can still renew that registration with Expo, and on Android with Google, when the App starts. Feedback is used solely to improve Bodydex — it is never sold, never used for marketing, and never shared beyond the delivery services named here.

The "Trouble signing in?" form (from app version 1.2.6). The sign-in screen has a form for people who cannot sign in. This form does not require sign-in, and its submissions are not stored under your Bodydex account identifier; the email address and message you enter can still identify you. When you send it, we collect the email address and the message you type (the address is required on this form, so that we can reply), the app version, the platform and operating system version shown on the form, and random codes made for the form. Our server uses the codes so that a message is not stored twice, to apply the daily limit for one form, and so that the App can later ask us to delete the message. We use all of this to help with the problem and to stop misuse of the form. No screenshot, subscription status, device hash, account identifier or push notification token is attached.

The message is stored on our backend (Supabase, US region) in a table of its own, together with a keyed one-way value made from the form code, which our server uses to apply the limit for one form and to find the message when the App asks for its deletion. It is shown in our internal support tooling, which reads it from that table and keeps no copy of it. It is not delivered through the operator notification service: that service receives only a notice that a message arrived, with a count and no address or text. A daily cleanup on our server removes each message once it is 28 days old, so it is normally gone from our support system within 30 days (a failed cleanup can delay this). We reply by ordinary email from our support address; that email conversation stays in our support mailbox, and what you write back to us is kept as described under "When you email us" below. We never use the email address you type into this form for marketing and never sell it.

To stop misuse of the form, our server limits how many messages can be stored in a day from one network and from one form, and how many can be stored in an hour and in a day in total. For the network limit it keeps, for each stored message, a network-limit marker in Cloudflare KV under a keyed one-way value made from the sending network and the day. A network-limit marker is gone within 25 hours. It is never the network address, and it is never stored beside the message. Separately, Cloudflare Durable Objects hold admission records containing a random reference, time, status and a keyed value derived from the form code, to enforce submission limits. They contain no network-derived value, email address or message text. Cleanup normally removes them within about 27.5 hours; failures can delay removal, and Cloudflare's recovery copies of that storage can keep them for up to a further 30 days. Deleting a form message does not remove these records or the network markers. As with any request, your IP address is processed to deliver the message and to apply a short rate limit (Section 3.2); it is not stored with the message. We also keep technical outcomes, error codes and aggregate counts to operate the form; these contain no email address or message text.

On your device the App keeps only the form's random codes with a reason word, a few times and counters and a yes / no, so that it can ask for the deletion described in Section 11.2, and the marker described next. After Delete All Data, to help finish your deletion request if the App closes unexpectedly, the App temporarily keeps a small deletion marker in the phone's secure storage. It contains no email or message. The App tries to remove it once the request is saved on this device. On an iPhone it may survive uninstalling the App. The form does not save your email or message to device storage: they are held in the App's memory only and are gone when the App is closed. A message from this form therefore does not wait on the device and is not tried again by itself. If the form cannot send, it says so and offers email instead: it can open your mail app with your message and a line naming the app version and phone system already in it, addressed to our support address, and nothing goes by email unless you send it there. In local-only mode (Section 12) the form sends nothing and shows a notice in place of its fields.

When you email us. Email you send to one of our addresses at bodydex.app (such as our support, press, privacy or legal address) is received by Cloudflare Email Routing and delivered to our mailbox at Google (Gmail). A copy is also kept in our private internal support system, which runs on infrastructure we operate. That copy is the email as it arrived, including any attachments, with metadata such as the sending and receiving addresses, the subject, the time it arrived and the sender authentication results added by our mail service (SPF, DKIM and DMARC). We use it to answer you, to track your request until it is handled, including privacy requests and the deadlines that apply to them, and to keep a record of how we handled it. Our system sorts email into categories, such as privacy requests or suspected spam, with fixed rules, not an AI service. While the copy is on its way to our system, Cloudflare holds it in its R2 storage as our processor; our system normally collects it and removes it from there within minutes, and a copy left there is scheduled for deletion after 14 days. Our system also sends us alerts, through the operator notification service in Section 8, about email that is waiting too long, is close to a reply deadline or did not reach our mailbox. An alert can name the address the email was sent to, its category, its reference number in our system, its age, its status and any deadline, but never the sender, the subject or the text.

In our system we keep, in this order of priority: email sent to our legal address, and receipts and billing records from our vendors, for 7 years; email that is a privacy request, a data request or an account-deletion request, or was sent to our privacy address, for 3 years, as a record of how each request was answered; otherwise, newsletters and email our system marks as suspected spam for 90 days; and all other email, such as email to our support or press address, for 3 years. Each period runs from when we marked the email handled or, if we have not, from its last change of status or, failing that, from when it arrived. Email about a request that is still open or waiting for a reply is not deleted because of its age. The copy delivered to our mailbox at Google is used for the same purposes and kept for as long as we need it for them. We never sell the email you send us and never use it for advertising or marketing. To ask for your email to be deleted sooner, write to (Section 11.9).

3.4 Data from Third Parties

  • Apple / Google Sign-In: Your name and email address as provided during authentication, delivered to us as identity tokens and profile claims — we never receive or store your Apple or Google password. A stable app account identifier is created from this sign-in and linked to your subscription status. Apple's "Hide My Email" feature is supported.
  • Apple HealthKit / Google Health Connect: If you grant permission, we may read and/or write health data as described in Section 5.
  • Age group from your app store: Where the law requires it, the App Store or Google Play tells the app the age group of the account on your device. It is used only to apply legally required age limits to protocol tracking, the research library and Community Stats sharing, and, if it shows an account under 13, to switch on the local-only mode described in Section 12. The age group itself is kept in memory only while the app is running, and is never stored, sent to our servers or shared; the one thing stored is the yes/no local-only setting described in Section 12.
  • Apple Ads attribution (iOS): If an installation followed an Apple Ads (App Store search) ad, Apple returns that ad's campaign, ad group and keyword to our subscription processor RevenueCat, as described in Section 8.3. From app version 1.2.6 the App does not ask Apple for this when the country or region of your App Store account is one listed in Section 11.6.
  • Your store account's country or region (from app version 1.2.6): when the App starts, it asks the App Store or Google Play, through our subscription processor RevenueCat's software in the App, for the country or region of the store account on your device. It is used on your device to decide whether the Apple Ads attribution of Section 8.3 runs (iPhone), whether ad measurement in Section 8.1 may run (only for a US store account), and whether the setup screen counts, the genuine-app check and the ad-measurement settings request in Section 3.2 are sent (Section 11.6). The App keeps it in memory while it is running and does not store it. It sends it to our server only with a food search while you are signed in with the Bodydex Food Core beta turned on (Section 8).

4.How We Use Your Data

We use the data described above to:

  • Calculate personalized calorie and macronutrient targets
  • Track food intake, nutrition, and dietary adherence
  • Manage medication, supplement, and protocol schedules with reminders
  • Provide AI-powered food photo analysis and nutritional estimation
  • Deliver AI nutrition coaching responses (when initiated by you)
  • Sync nutrition and fitness data with Apple HealthKit or Google Health Connect (with your permission)
  • Deliver push notification reminders (meal logging, medications, etc.)
  • Verify subscription status and enforce tier-based feature limits
  • Prevent abuse of free-tier rate limits via pseudonymous identifiers
  • Monitor service quality and forecast infrastructure costs via pseudonymized, aggregate usage metrics (see Section 3.2)
  • Understand where new users stop during setup, and improve setup, sign-in and the subscription screen, using the setup screen counts (see Section 3.2)
  • Tell real setups from made-up counts, using the genuine-app check (see Section 3.2)
  • Measure whether our own ads lead to installations, trials and purchases: from app version 1.2.6 through ad measurement, only if you agree to it during setup (Section 8.1); in app versions before 1.2.6 through the one install message sent through our server to Meta (Section 8.2); and on iPhone through Apple's own ad measurement (Section 8.3)

We do not sell or rent your personal data for money, and we disclose no personal data to data brokers. We disclose personal data to advertising companies only in the two ways described in Section 8: ad measurement, if you agree to it (Section 8.1), and the install message of app versions before 1.2.6 (Section 8.2). Giving identifiers and these events to advertising companies may count as a "sale", "sharing" or "targeted advertising" under some US state laws (Section 11.5). We do not give the contents of your logs (food, body, medications and the rest) to advertising companies. Our sole revenue model is subscription fees.

5.Apple HealthKit & Google Health Connect

HealthKit and Health Connect data is handled with the highest level of privacy protection. The App does not copy your health records to a server of ours to hold them: we do not maintain a synchronized server copy of your tracking database. What the App reads from a health platform goes one of two ways, and Section 5.1 says which for each record type: some of it is written into the App's own database on your device (a weigh-in into your weight log, a drink into your water log; and a workout recorded by another app as a summary that feeds the muscle-recovery view, not as an entry in your workout history), and the rest is read on demand and not stored — the vitals readings, fetched for the dates a chart or a question covers, and your daily step and active-energy totals, read when a screen shows them; all are shown to you and held only in memory while the screen is open (the one exception: the home-screen widget keeps today's step and active-energy totals on your device, replaced each time it refreshes). Three things can then leave the device, all described in Section 6.2 and none of them without your consent to the Bodydex Assistant: values that were imported into your ordinary logs (the Assistant reads those as the log rows they became); a recovery percentage per muscle group, if the Assistant is asked about training and you allowed the fitness category — a single derived number that the workout summaries above feed into, never the external session itself; and, only if you tick the separate Health vitals option, vitals readings read on demand for the dates your question is about. Nothing is sent to a health platform except what Section 5.2 lists. Apart from the Assistant, values imported into your logs leave the device only in a copy you choose to make: a Profile › Export Data file, and, on iPhone, your device's own iCloud or computer backup, which includes the App's database.

5.1 What We Read (with your permission)

Read into your logs. These become ordinary rows in the App's own database — a weigh-in in your weight log, a drink in your water log — and from that point the App treats them like anything you typed in yourself:

  • Body weight and height
  • Water intake
  • Biological sex and date of birth (iOS only, for profile auto-fill during onboarding)

Read on demand, not stored: daily step count and active energy burned (calories). The App reads today's totals (and, for Trends, the days a chart covers) when a screen shows them; they are not written into your logs or the App's database. The only copy kept is today's two totals for the home-screen widget, on your device, replaced each time it refreshes and cleared when you turn the read off.

Read as a summary for muscle recovery. Workout and exercise sessions recorded by your other apps — a watch, a running app, another gym app — are read so the App can tell how recently each muscle group was worked. What is stored is a summary of each session: when it started and ended, how long it lasted, the activity type, the muscle groups it maps to, a workload estimate, the name of the app that recorded it, and an energy figure where the platform supplies one. These summaries are not entries in your workout history — your workout history holds the workouts you logged in Bodydex. They appear in the muscle-recovery view and as a count on the fitness dashboard, and you can turn the whole thing off at Profile › Apple Health / Health Connect › Workouts.

Read on demand to chart a vital against your doses. Each of these is requested on its own, at the moment you add that chart, and is read only for the dates being charted or asked about. They are not written into your logs:

  • Resting heart rate
  • Heart rate
  • Heart rate variability
  • Blood pressure (systolic and diastolic)
  • Blood glucose
  • Blood oxygen saturation
  • Respiratory rate
  • Sleep sessions (nightly total)
  • Body fat percentage
  • Lean body mass

The full list of the vitals charts you have enabled is in the App at Profile › Apple Health / Health Connect › Vitals for protocols. Sending any of these readings to the Bodydex Assistant additionally requires the separate Health vitals tick described in Section 5.3.

Heart rate in particular is read on demand only: for the Trends heart-rate chart and for a vitals chart, for the dates that chart covers. It is never written into your logs or the App's database. When the Bodydex Assistant draws a vitals chart for you, the App keeps the last few of those charts' readings in its memory, never in storage, so the chart can be shown again; they are gone when the App is closed. Separately, when a workout is saved to Apple Health through the App's older save method and Apple Health already allows Bodydex to write heart rate, the App attaches a copy of the heart-rate samples Apple Health holds for that workout's time to the workout; those copies stay inside Apple Health.

5.2 What We Write (with your permission)

  • Dietary energy consumed (calories)
  • Dietary macronutrients (protein, carbohydrates, fat, fiber — and, where the platform supports them, sugar, sodium, and cholesterol)
  • The minerals chromium and molybdenum, for foods whose data states them (on Apple Health, only if you allow those two types)
  • From app version 1.2.7, with your permission and when values are available, Bodydex also writes vitamins A, B1, B2, B3, B5, B6, B7, B9, B12, C, D, E and K; calcium, iron, magnesium, phosphorus, potassium, zinc, selenium, copper, manganese and iodine; and saturated, monounsaturated and polyunsaturated fat (on Apple Health, only the types you allow). Health Connect also receives trans fat, and total unsaturated fat when both the monounsaturated and polyunsaturated values are known. These writes occur on your device; they do not upload these records to Bodydex's servers.
  • Body weight measurements
  • Water intake
  • Workouts you finish in Bodydex, on both Apple Health and Google Health Connect: the session with its name, start and end time and activity type; the active calories the App estimated for it, when it has an estimate; and, for a strength workout you logged live, each completed set as a segment of that workout (on Apple Health with the exercise name, reps, weight and set type; on Health Connect with the kind of exercise and the reps)

5.3 Health Data Commitments

We make the following commitments regarding your Apple HealthKit and Google Health Connect data:

  • We run no health-record database of our own. We do not maintain a synchronized server copy of your tracking database (we retain only the operational records this policy describes), and we hold no health record of yours on our servers for us to look at — with one transient exception: if you use the Bodydex Assistant, a reply that pauses mid-answer leaves a working copy on our proxy on the criteria described in Section 6.2, and that copy can contain records you consented to send. What the App reads from a health platform either goes into its own database on your device or is not stored at all — Section 5.1 says which, per record type. When you use the Bodydex Assistant with your explicit consent, the records and summaries listed in Section 6.2 are sent to OpenAI through our proxy, and what our own server keeps while doing that is described there. Some of those values were imported from a health platform (see the callout above and Section 5.1) and are sent as the ordinary log rows they became, under the ordinary data-access categories you grant in Profile › Bodydex Assistant › What the assistant can see.
  • Dedicated vitals reads are a separate opt-in. Reading the vitals listed in Section 5.1 as readings — resting heart rate, heart rate, heart rate variability, blood pressure, blood glucose, blood oxygen, respiratory rate, sleep, body fat, lean body mass, and the daily steps and active energy read from your health app — and sending them to OpenAI requires the "Health vitals" category, which is off unless you tick it yourself and which no preset ever turns on. While it is off, the Assistant performs no dedicated read of your health app and no such reading is sent.
  • Imported values are not covered by that tick. A weight or water entry that came from Apple Health or Health Connect is a row in your own log from the moment it is imported, and the Assistant reads it under the matching ordinary category — body or nutrition — which the App's default data-access answer turns on. So with "Health vitals" off, the Assistant can still be sent a weigh-in that originated in your health app, because it reads it as your weight log and not as a health-platform reading. A workout recorded by another app is not a log row and is never sent as one; what it can reach the Assistant as is the per-muscle recovery percentage described in the callout, under the fitness category. If you do not want any of that, switch the relevant category off in Profile › Bodydex Assistant › What the assistant can see.
  • Readings and imported values alike are never sold, never used for advertising, and never shared with anyone beyond the recipients named for the feature you are using.
  • Health data is NOT used for advertising or marketing of any kind
  • We do not sell health data or disclose it to data brokers. The online features described in this policy disclose specified health data to the recipients identified for those features
  • Health data is NOT shared with third parties for purposes unrelated to providing core health functionality within the App
  • Health data is NOT used for any purpose other than providing health and fitness functionality directly to you within the App
  • What these commitments cover: they are about data read from Apple HealthKit or Google Health Connect. That data is never sent to an AI provider for model improvement: it can reach an AI provider only through the Bodydex Assistant, on the terms in Section 6.2. Other health-related content you submit yourself — a food photo, a voice note, a document you import — follows the AI-provider terms described in Sections 6 and 6.6, which differ by feature.
  • You can revoke health data permissions at any time through your device's Settings

6.AI Feature Data Processing

Certain features use AI services, and their requests pass through our servers (Cloudflare Workers). Our servers keep the operational records described throughout this policy. What our servers and each AI provider keep differs by feature: the sections below say what, and Section 9 gives the periods. AI requests are keyed to a pseudonymous account identifier (used for authentication and rate limiting); your profile name and email address are not attached automatically, but content you submit may itself identify you.

6.1 AI Food Photo Scanner

  • Data sent: Food photo (image data), scan mode type, and your pseudonymous account identifier
  • Data NOT sent: Your name, email address, or profile details
  • Your agreement (from app version 1.2.7): the Terms and Privacy Policy agreement you tick before Get Started or Sign In in app version 1.2.7 or later also allows this sending; users who ticked the earlier wording are not asked again (Section 6.6). A photo is sent only when you take or choose one for a scan; there is no separate question before the first scan.
  • Storage: We do not intentionally retain ordinary scan payloads after processing, with one exception: if you rate a scan result as inaccurate and explicitly choose "Send with photo," that single photo is stored (with no account identifier attached) so we can reproduce and fix the failure. Opt-in report photos are stored for troubleshooting; the 90-day automatic deletion control is not yet operationally verified. What our AI providers may retain is described in Section 6.6.

6.2 Bodydex Assistant

The Bodydex Assistant is an automated AI system. It is off until you turn it on, and it only runs after you give explicit consent in the App to the processing described here, including health data. The provider is OpenAI (OpenAI OpCo, LLC, United States), reached through our server proxy.

  • Data sent with each message, within the categories you allow: your message text (limited to 300 characters); your earlier messages in the current conversation, which the App sends to our proxy up to the 50 most recent, while our proxy forwards only the 5 most recent messages, counting your new one, to OpenAI as context; your profile and setup details (age, sex, height, weight, goal and goal pace, activity level, diet type, cooking frequency, focus areas, obstacles, motivations, whether you work with a human coach, any coach or motivation note you entered, and your preferred day window if you set one); a nutrition summary (today's totals, targets, per-meal calories and water, and 7-day averages); your recent weight readings (the last few you logged) and goal estimate; the names of your active supplements, medications, and peptides with the doses you recorded; and your saved Assistant notes and the details it remembered from earlier chats.
  • Data sent for some questions: at the Assistant's request during a chat, the App may read from your device and send: dated nutrition, water, and per-meal totals for up to the last 30 days; dated weight measurements for any period you ask about, up to your whole recorded history, with your goal estimate; your body measurements, and the recipes you have saved with their ingredients, macros, and how often you cook them; a protocol overview (each active medication, supplement, and peptide with the dose you recorded, its schedule summary, 7- and 30-day adherence, supply level, and the next dose and time slot recorded in your schedule); for every active protocol, its containers on hand (open and sealed) and days of supply; dose history for up to 30 days (counts of taken, skipped, and missed doses, injection sites used, and the last taken date, for one protocol or all); workouts (sessions with dates, names, duration, volume, calories, and streak); reminder settings (notification preferences, permission status, and how many protocols have reminders); your logging streak and badges (names, earned status and dates, progress); and the full record of one protocol, workout, or routine you ask about. Opening the Assistant from a Research Library entry or a protocol page also sends that entry's topic and a short summary of the selected protocol (name, recorded dose, schedule, 7-day adherence, supply level). With the separate "Health vitals" opt-in turned on, the App may also read and send readings from Apple Health or Health Connect — resting heart rate, heart rate, heart rate variability, blood pressure, blood glucose, blood oxygen, respiratory rate, sleep duration, body fat, lean body mass, and daily steps and active energy — for the dates your question is about, and only the readings that question needs; where a reading is missing, the App may also send the reason (for example, that the permission was not granted or the health app has no data for those dates) so the Assistant can tell you how to fix it. With the separate "Lab results" opt-in turned on, the App may also read and send the lab results you recorded — the values, their dates, the lab, and the reference range from your own report, whether you entered them yourself or imported them from that report, as a photo or as a PDF (Section 6.7) — for the question you asked and no other; apart from the working copy described below, none of it is stored on our servers, and Bodydex does not interpret them. Some of the other values above may also be derived from Apple Health or Health Connect if you connected them.
  • Data NOT added automatically: your profile name and email address are not attached to Assistant requests; a pseudonymous account identifier is used instead. Messages and notes you write may themselves contain identifying information — do not include details you wish to keep private.
  • Abuse-monitoring identifier sent to OpenAI: each Assistant request to OpenAI carries a keyed pseudonymous identifier derived from your account identifier, sent only so that OpenAI can detect abuse of its service under its usage policies. It is not your name or email address, and it is derived with a secret key, so it differs from the identifier our analytics use.
  • What the Assistant may read: seven categories — nutrition and calories, protocols and dose logs, workouts and fitness, body weight and measurements, health vitals, lab results, and memory — each of which you turn on or off in Profile › Bodydex Assistant › What the assistant can see. "Health vitals" and "Lab results" are each off unless you tick them yourself, and no preset turns either on. Your setup answers (age, sex, height, activity level, units and the goal you chose) ride with every message while at least one category is on; allow nothing there and the Assistant answers from your messages only.
  • Storage on our servers: your conversation history and Assistant memory are stored on your device, not on our servers. Two server-side records exist. First, when a reply pauses mid-answer to look something up in the App, our Cloudflare proxy holds a working copy of that reply — which can include AI text, the pending requests for your data, the data returned, and personal information. We attempt to replace the content of that working copy with a record that carries no message text as soon as the reply finishes — an opaque turn reference kept only so a repeated request can be told it has already been handled. We do that on every completed reply, normally within seconds of the reply landing; we cannot promise every attempt succeeds, and where one does not the content is not removed at completion but stays subject to the criteria below. If a paused reply is never finished and never resumed — you close the App, your phone locks, the network drops — we cannot give you a measured maximum, so here are the criteria instead, as data-protection law allows where a period cannot be given: the pause stops being resumable after 3 minutes, the record becomes eligible for deletion 10 minutes after that, and a deletion scheduled at the moment of the pause then attempts to remove it once it is eligible, rather than waiting for a later request to trigger a sweep. Those two figures describe when the record becomes eligible for deletion, not when deletion succeeds: the delivery of a scheduled deletion is retried by the platform, an attempt can still fail, and where one does the copy stays in place until a later attempt succeeds. So we cannot bound how long an abandoned working copy persists. Second, answers to standard Research Library questions, which are sent without any of your personal context, are cached for up to 7 days so they need not be regenerated.
  • Retention by OpenAI: OpenAI states that content sent through its API is not used to train its models unless the customer opts in, and that it keeps abuse-monitoring logs for up to 30 days unless a longer period is required by law or reasonably necessary to protect its services or any third party from harm. We use OpenAI's standard API terms and have not arranged zero-data-retention, so you should assume OpenAI may hold Assistant content for up to 30 days under its own terms, and longer where those exceptions apply.
  • Narrowing what the Assistant may read: if you switch a category off — including unticking "Health vitals" or "Lab results" — the Assistant starts a fresh conversation. The earlier conversation is never sent again, including anything the Assistant had remembered from it, so data covered by a permission you withdrew cannot be replayed to OpenAI by a later message. The turns the App removed from the live thread are kept on your device as a read-only transcript you can open and delete, and it cannot be sent: nothing in the App reads it back into a request. It is kept for that single reason — so that tightening a permission does not silently destroy your own history — and Delete transcript removes it, including while your consent is withdrawn. Only the 200 most recent messages are kept in a conversation; older ones are dropped.
  • Withdrawing consent: see Section 11.7. Withdrawing consent stops the sending: new Assistant requests stop at once, including further requests from a reply already in progress, and nothing from your conversations, memory or notes is sent to OpenAI afterwards. It does not erase them. They stay in the App's database on your device — your live thread, any archived or read-only transcript, your notes, and the facts the Assistant saved — and so, in one narrow case, does a conversation the App was unable to read back: rather than overwrite or destroy your history, the App sets those bytes aside untouched, and nothing parses, sends or replays them. We are not relying on the consent you withdrew to keep any of this, and we keep none of it for a purpose of our own. It is on your device, we have no access to it, and it is yours to delete: Delete transcript for an archived or read-only conversation, New Chat or Clear everything for the current one, per-note delete and Reset Assistant Memory for notes and remembered facts, Redo Assistant Setup to clear every one of them at once, or Delete All Data for the whole App. Withdrawal cannot recall information already received by our providers, which remains subject to OpenAI's retention described above.

6.3 Voice-Assisted Food Logging & Voice Corrections

  • Data sent: Your phone's own speech recognition (Apple or Google) hears what you say and produces a text transcript; it may process the audio online, under Apple's or Google's terms. The App also sends our servers a recording of up to about 25 seconds, which they transcribe using a speech-to-text AI service (currently Cloudflare Workers AI, with Groq as a fallback), with optional vocabulary hints taken from the names of items you track. A longer recording is not uploaded; the phone's transcript (limited to 2,000 characters) is sent instead. The transcript text is then sent to OpenAI to be parsed into food entries.
  • Data NOT sent: Your name, email address, or profile details
  • Storage: Our servers discard recordings and transcripts after processing. Cloudflare Workers AI, which transcribes first, states that it does not use customer content to train AI models or to improve its or third-party services without the customer's explicit consent. Groq, used only if that fails, states that by default it does not retain inputs or outputs, but may temporarily log them, for up to 30 days, to troubleshoot errors affecting its platform or to investigate suspected abuse. The transcript text then goes to OpenAI under the API terms described in Section 6.2.
  • Your agreement (from app version 1.2.7): the Terms and Privacy Policy agreement you tick before Get Started or Sign In in app version 1.2.7 or later also allows the sending for voice logging (food or doses) and spoken scan corrections; users who ticked the earlier wording are not asked again (Section 6.6). A recording is sent only when you record one; there is no separate question before the first recording.

6.4 Video Recipe Import

  • Data sent: The video link you share and the post's caption. Our server retrieves the video's captions and, when needed, its media from that link (via a content-retrieval service). Where the platform already provides a usable transcript or subtitles, those are used and no speech-to-text request is made. Otherwise the retrieved audio file is sent to Groq's speech-to-text service to transcribe the speech — or the whole video file, where a separate audio file is not available or the audio file is not accepted (Groq's retention terms are described in Section 6.3). The caption, captions and transcript are then sent to Google's Gemini API to extract the recipe.
  • The video itself: when the caption and transcript are not enough to produce a usable recipe, the video itself is also sent to Google's Gemini API, which watches up to its first 5 minutes to read the spoken words, the text shown on screen and what is being cooked. For a YouTube video, our server passes the link to Gemini and Google retrieves the video from YouTube; for other platforms, our server sends Google the video file it retrieved. This is done only for imports made with a Bodydex subscription, including during its free trial, and not for the free imports available without one. It uses the same Gemini tier selection as the rest of recipe import (Section 6.6): requests normally go to the unpaid tier, and go to the paid tier after a daily usage threshold we set or to retry a request that failed on the unpaid tier, unless our own daily limit on paid-tier requests has been reached, in which case they stay on the unpaid tier. Google's terms for the tier used, described in Section 6.6, apply to the video too, including, on the unpaid tier, Google's use of it to improve its products and review by human reviewers.
  • Data NOT sent: Your name, email address, or profile details
  • Storage: A status record for your import (including the shared link and the resulting recipe) is kept for up to 7 days so the result can be delivered to your device, then automatically deleted. A shared result cache, keyed only to a code made from the video link, holds the extracted recipe and technical notes on how it was extracted; it carries no account or device identifier and is kept for up to 30 days so the same public video is not processed again.
  • Your agreement (from app version 1.2.7): the Terms and Privacy Policy agreement you tick before Get Started or Sign In in app version 1.2.7 or later also allows the sending for recipe import, including the content the link leads to; users who ticked the earlier wording are not asked again (Section 6.6). A link is sent only when you share or paste one to import; there is no separate question before the first import.

For ordinary web recipe pages (non-video links), your device fetches the page directly: the destination website receives the URL request and ordinary network metadata (including your IP address) under its own policies, and the parsed recipe is stored only on your device.

6.5 AI Transparency

When you use the features described in this section, you are interacting with an automated artificial-intelligence system, not a human. The outputs these features produce — nutritional estimates, Assistant responses, and analysis results — are AI-generated content, and the App identifies these features as AI-powered at the point of use: the Assistant announces itself as an AI system when you set it up, and its chat screen carries a persistent notice that you are talking to an AI that can make mistakes. If you ask the Assistant whether it is a human or an AI, it should tell you that it is an AI. This disclosure is provided in line with applicable AI transparency laws, including Article 50 of the EU Artificial Intelligence Act. The Assistant is a tracking and general-information tool; it is not designed to provide companionship, therapy, counselling, or crisis support.

6.6 AI Training Data Usage

We do not collect, use, or sell your personal data for the purpose of training large language models, our own or anyone else's. We train no model. What follows is what the providers we send data to are permitted to do with it under their own terms, which differs between them.

Your agreement (from app version 1.2.7): the food, label and menu photo scans, scan corrections, voice recordings and recipe imports described in this section are sent only when you use one of those features. From app version 1.2.7, the Terms and Privacy Policy agreement you tick before Get Started or Sign In says, in words: "I agree to the Terms of Service and the Privacy Policy, and allow Bodydex to send what I give its AI features (photos, voice recordings, text, links and the content they lead to) to the outside AI services listed in the Privacy Policy, for analysis." Those services are the providers named in this section and elsewhere in Section 6, and ticking it is your permission for that sending. The App does not ask a separate question before these features first send, and there is no setting that switches them off: you prevent new requests by not using them (Section 11.7). The Bodydex Assistant keeps its own explicit consent (Section 6.2), and document import its own separate consent (Section 6.7); the Get Started agreement stands in for neither. If you ticked the agreement in app version 1.2.6 or earlier, its text did not include that sentence and the App does not ask you again; in every version, these features send this content only when you use them, as described here.

Bodydex Assistant (OpenAI): OpenAI states that content sent through its API is not used to train its models unless the customer opts in; we have not opted in. OpenAI may still retain content for service operation, abuse monitoring, or legal requirements under its terms (see Section 6.2).

Food scanning and recipe extraction (Google): food, nutrition-label and menu scans, scan corrections, the text and captions of recipe imports and, for some video imports, the video itself (Section 6.4) may be sent to Google's Gemini API on its unpaid tier. Under Google's terms for that tier, Google may use both what is sent and what the model returns to provide, improve and develop its products, and human reviewers may read, annotate and process them. A request goes to the paid tier instead — where the terms described below for document import apply — once a daily usage threshold we set below that tier's limit is reached, or to retry a request that failed on the unpaid tier. Scan corrections and recipe imports stay on the unpaid tier once our own daily limit on paid-tier requests has been reached; photo scans are not subject to that limit.

Voice transcription (Cloudflare Workers AI, then Groq): Cloudflare states that it does not use customer content sent to Workers AI to train AI models or to improve its or third-party services without the customer's explicit consent. Groq, used only when Cloudflare's transcription fails, states that by default it does not retain inputs or outputs, but may temporarily log them for up to 30 days to troubleshoot platform errors or investigate suspected abuse. Neither says it uses the audio for its own model training. The transcript text is then parsed by OpenAI under the API terms above. A lab report or product label you import is never sent on Google's unpaid tier: document import uses the paid tier only (below). We do not attach your name or email address to these submissions; an app-specific pseudonymous identifier may accompany the request for rate limiting, and submitted content can itself contain identifying information if you choose to include it (for example, in a photo).

Document import (Section 6.7): when you import a lab report — as one or more photos of it, or as its PDF — or photos of a product label, what you chose is sent to a third-party AI provider to be read. That provider processes it under its own terms, not ours, and those terms — not the promises we make about our own servers — govern what it may retain and what it may use the content for. We send imported documents on our provider’s paid tier. Under that provider’s published terms, content sent on the paid tier is not used to improve or train its models, and is processed under its data-processing addendum for products where it acts as a data processor. The provider does log prompts and responses for a limited period, solely to detect and prevent violations of its prohibited-use policy and to meet legal or regulatory obligations, and that data may be stored or cached in any country where the provider or its agents maintain facilities. We have not arranged zero data retention. As above, we attach no name or email address; a document you photograph or choose can itself carry identifying information, which is why the extractor is instructed never to return it and why our server rebuilds the result from a fixed list of fields (Section 6.7).

6.7 Document Import (Lab Reports & Product Labels)

The App can read a document you photograph, or choose from your photo library, so that what is printed on it can be filled in for you instead of typed: a lab report, or the label on a medication, supplement or peptide. A lab report can also be imported as a PDF — the file your lab sent you — which you open from your files rather than photograph; a product label is photographed only. This is off until you turn it on. The first time you use it, the App asks for your separate, explicit consent to send the file — it is never bundled into accepting the Terms, into onboarding, or into a subscription — and every import after that shows a notice before anything leaves your device. You can withdraw the consent at any time at Profile › Legal & Data › Document import, which stops both report reading and label reading (Section 11.7).

  • Data sent: Only what you chose — one or more photos of the document, or, for a lab report, its PDF — and nothing else from your records. The request goes through our server proxy to a third-party AI provider, for that one request, and is authenticated to your account with a session token.
  • Data NOT sent: We do not add profile identifiers or other logs to the request; the selected document itself may contain identifying information. Your photo library is not read — only the images you selected. Choosing a PDF is the same in scope: the system file picker hands the App that one file and nothing else on your device is read (Section 7.3).
  • What it does: It transcribes what is printed. For a report: the marker names, the values, the units, and the reference range your lab printed. For a label: the name, the strength, the amount, and the directions. Bodydex does not interpret results, states nothing about any value, gives no medical advice, and never suggests a dose or a schedule from a label.
  • The identifiers printed on the page: A report or a label usually carries your name, date of birth, address, a record or prescription number, or a prescriber's name. The extractor is instructed never to return any of them, and because an instruction is not a control, our server then rebuilds the result from a fixed list of fields and copies nothing else across. We restrict returned fields and instruct the model to omit identifiers; these measures cannot guarantee removal — an identifier placed inside a permitted field, such as the lab's name, can still come through, which is one more reason to check what is shown before you save it.
  • Storage on our servers: none. We store neither the file nor what is read from it. No part of the document is written to a log, to our key-value store, or to any other store of ours. This is a statement about our servers; what the AI provider may retain under its own terms is Section 6.6.
  • On your device: To send it, a working copy is made in the App's own storage and deleted as soon as the file has been prepared. The two paths differ in who makes that copy: for a photo the App makes it; for a PDF the system file picker copies the file into the App's cache when you choose it, and the App deletes that copy once it has read the file — before the request is sent, because what is sent is the data it read and not the file on disk. Your own photo or PDF is left where it is; the App neither moves it nor deletes it.
  • Nothing is saved until you save it: What was read is shown to you to check first. You are responsible for checking every value against your own report before you save it, and nothing reaches your records until you do. From that point it is a record on your device like any other, and you can correct or delete it.

A lab report you import — photographed, or the PDF your lab sent you — is health data, and, where the GDPR or UK GDPR applies, special-category data under Article 9. That is why the consent is separate and explicit, why it is recorded with the version of the disclosure you actually read, and why changing that disclosure asks you again rather than relying on a tick from months ago (Sections 11.6 and 11.7). For residents of Washington, Nevada and Connecticut it is also consumer health data — see our Consumer Health Data Privacy Policy.

7.Progress Photos, Camera & Microphone

7.1 Progress Photos

The App allows you to take and store body progress photos for personal tracking. These photos are:

  • Stored on your device, in the App's local files, referenced by its database
  • Not uploaded to our servers, and not shared by the App with any third party
  • Included in your device's own backup if you use one (on iPhone, an iCloud or computer backup); deleting a photo in the App does not delete a backup copy
  • Deleted from the App when you use the "Delete All Data" feature

7.2 Camera Access

The App requests camera access to enable AI food scanning, barcode scanning, nutrition label scanning, photographing a lab report or a product label to import it (Section 6.7), and progress photos. The camera is only activated when you explicitly open a scanning or photo feature. Food photos captured for AI analysis are transmitted for processing and then discarded from our servers (except the optional scan-accuracy reports described in Section 6.1). On your device, a copy of each scan photo is kept temporarily for your scan history and is automatically cleaned up on a schedule of roughly 30 to 90 days depending on subscription tier and scan status.

Photos you take of a lab report or a product label to import — one or more per import — are handled differently from a scan photo: they are sent for that one request, we keep neither them nor what is read from them, and the App does not keep a copy of them alongside the record it creates — only the working copy it makes in order to send them, which it deletes as soon as they have been prepared (Section 6.7). Importing a lab report as a PDF instead uses no camera access at all (Section 7.3).

7.3 Photo Library and File Access

The App may request access to your photo library to allow you to select an existing food photo for AI analysis, or an existing photo of a lab report or product label to import (Section 6.7). Only the images you select are read; selected photos are processed identically to camera captures, and are left in your library where they are.

Importing a lab report as a PDF uses your device's own file picker instead, and asks for no photo-library or storage permission: choosing the file is what grants access to it, the App reads only that one file, and nothing else in your files is read or listed. The file you chose stays where it is (Section 6.7).

7.4 Microphone Access

The App may request microphone access for voice-assisted food logging and voice corrections. Voice features use your phone's speech recognition and also send a recording or its transcript through our servers (Section 6.3). The phone's speech recognition may process audio online. Our servers discard recordings and transcripts after processing. Audio is recorded only while you are actively using a voice feature.

8.Third-Party Services

The App integrates with the following third-party services. Each service has its own privacy policy governing their handling of data:

ServicePurposeData Shared
FatSecret Platform APIFood search & nutrition dataSearch queries, barcode numbers
OpenAI (OpenAI OpCo, LLC, United States)Bodydex Assistant chat; parsing voice-log transcripts into food entriesAssistant messages and the personalization context listed in Section 6.2 (including health data, with your explicit consent); voice-log transcript text; keyed to a pseudonymous account identifier
Other AI Service Providers (currently Google, Groq, and Cloudflare Workers AI)Food photo, label and menu analysis; reading a lab report or product label you import; voice transcription; recipe extractionFood photos, the lab report you chose to import — as a photo or as its PDF — and the photo of a product label you chose to import (Section 6.7), voice audio clips, and short text prompts, keyed to a pseudonymous account identifier (see Section 6); for recipe videos you share (Section 6.4), the audio or video file, sent to Groq to transcribe the speech, and the link, caption, captions and transcript, sent to Google's Gemini API, together with, for some imports made with a subscription, the video itself
ApifyVideo content retrieval for recipe importThe video link you share, and nothing else about you or your account. Apify uses the link to retrieve the post's public caption, its subtitles or transcript where available, and a link to its video or audio file for us; for TikTok, Apify keeps a copy of the video in its own storage for our server to download, under Apify's retention terms (see Section 6.4)
RevenueCatSubscription managementApp-scoped user ID, purchase transaction data, and the answer you chose to "How did you find us?" during setup; on iOS, Apple Ads attribution (campaign, ad group, keyword) obtained from Apple's AdServices token — no advertising identifier (from app version 1.2.6, not for an App Store account in the countries listed in Section 11.6). RevenueCat shows the country of a purchase from the store transaction or, when there is none, from the country of the IP address it last saw; RevenueCat states that it does not keep the IP address itself. From app version 1.2.6: on a device that identifies itself as an automated Google Play test device, and only there, a marker saying so is added to that installation's RevenueCat record so test devices can be left out of our counts; and RevenueCat's software in the App reads your store account's country or region on your device for the decisions described in Section 3.4 (the App sends it to us only with a food search while the Bodydex Food Core beta is on, Section 8). RevenueCat is not part of ad measurement: the App gives it no advertising identifier, and no RevenueCat integration sends anything to an advertising or measurement company for us (Section 8.1)
CloudflareSecure API proxy & infrastructure; storage of the setup screen counts, the genuine-app check records and the install message records; receiving email sent to our bodydex.app addressesAll AI requests are routed through Cloudflare Workers; rate-limiting counters stored in Cloudflare KV (auto-expiring, 48 hours to 365 days by tier — see Section 9); the install message records of app versions before 1.2.6 (Section 8.2), stored in Cloudflare KV for about 90 days; from app version 1.2.6, the setup screen counts and the genuine-app check (Section 3.2), received by Cloudflare Workers at funnel.bodydex.app and attest.bodydex.app and stored in Cloudflare's D1 database under the retention in Section 9, with the IP address of each request processed to deliver it and apply a rate limit but not stored with them; from app version 1.2.6, a message from the "Trouble signing in?" form (Section 3.3) passes through Cloudflare Workers on its way to Supabase, the form's network-limit markers are stored in Cloudflare KV for up to 25 hours, and its submission-limit admission records (a random reference, a time, a status and a keyed value derived from the form code; no network-derived value, email address or message text) are held in Cloudflare Durable Objects, normally removed within about 27.5 hours; email you send to one of our bodydex.app addresses, with its attachments, is received by Cloudflare Email Routing and passes through Cloudflare Workers, which hold a copy in Cloudflare R2 storage until our internal support system collects it, normally within minutes; a copy left there is scheduled for deletion after 14 days (Section 3.3); support messages, bug reports and food-data reports, including attachments, pass through Cloudflare Workers to Supabase; from app version 1.2.7, there is no direct-database fallback
Apple / GoogleAuthentication, push notifications, health data sync; your phone's speech recognition for voice features; from app version 1.2.6, the genuine-app check (Apple App Attest, Google Play Integrity), when it is switched on; Google (Gmail) hosts our mailboxSign-in credentials (managed by Apple/Google), push notification tokens; with your permission, the nutrition, body-weight, water and workout values listed in Section 5.2, written to Apple Health or Health Connect on your device; what you say to a voice feature, which the phone's speech recognition may process online (Section 6.3); for the genuine-app check, a request from your device to confirm that the App is genuine, which Apple or Google handles under its own terms (Section 3.2); email you send to one of our addresses, with its attachments, delivered to our mailbox at Google (Section 3.3)
Expo (650 Industries, Inc.)Relaying support-reply push notifications to Apple or GoogleYour push notification token and the notification itself (who replied, the opening of the reply up to 120 characters, and a ticket reference), only if you granted notification permission (see Section 3.3)
Supabase (Supabase, Inc., US region)Backend for feedback, bug reports and support conversations and, from app version 1.2.6, for messages from the "Trouble signing in?" formThe submissions described in Section 3.3: what you typed and your replies in a support conversation, any screenshots or label photos you attached, your email address if you typed one, app version, platform and OS version, subscription status and store, the pseudonymized device hash, a keyed hash of your account identifier if you are signed in, and your push notification token if you granted notification permission. From app version 1.2.6, for a message from the "Trouble signing in?" form (Section 3.3): the email address and the message you typed, app version, platform and OS version, the time, a random code made for the message, and a keyed one-way value made from the form code
Telegram (Telegram Bot API)Internal support notifications to the developerFor a new report, its type and title and any screenshots attached to it; for a reply in a support conversation, the opening of your reply (up to 300 characters) with the ticket's type and status and whether the account is Premium; for a crash report, its title, app version, platform and a shortened pseudonymous account hash; for a report of incorrect food data, the product name, brand, barcode and the change you proposed (label photos are not sent); and reminders naming reports that are still waiting; for email you send to our addresses, alerts about email that is waiting too long, is close to a reply deadline or did not reach our mailbox, naming the address it was sent to, its category, its reference number, its age, its status and any deadline, but never the sender, the subject or the text (Section 3.3)
Meta (Meta Platforms, Inc.)Measuring our own ads: ad measurement from app version 1.2.6, if you agree (Section 8.1); the install message of app versions before 1.2.6 (Section 8.2)From app version 1.2.6, only if you agree during setup: from Meta's software in the App, your install and first open, a trial started or plan bought during setup with its price and currency, your advertising ID (on iPhone only if you allow tracking), Meta's own installation code, your IP address and phone and app details (Section 8.1). From app versions before 1.2.6: one install message per installation, sent by our server the first time the App is opened: that the App was installed, the time our server received the message, a random code made for that message, the platform, and the App's name and version; on Android also the Google Play install referrer when it shows that the installation came from one of our Meta ads (Section 8.2). Neither includes your health measurements, logged records, name or email address; both reveal to Meta that Bodydex was installed
Tenjin (Tenjin, Inc., United States)Our ad-measurement processor: attributes installs, trials and purchases to the ad that brought them, from app version 1.2.6, if you agree (Section 8.1)From Tenjin's software in the App: your install and first open, a trial started during setup (no price) or a plan bought during setup (product, price and currency; on iPhone also the App Store's details of that purchase so Tenjin can check that it is real), on iPhone, the IDFA and the app-vendor ID, received by Tenjin's SDK only when tracking permission and the other eligibility conditions allow the SDK to run; on Android, the advertising ID, after agreement and eligibility checks; Tenjin's own installation and session codes, how long the App is used during setup, your IP address, phone and app details, on Android the Google Play install referrer and on iPhone Apple's ad-attribution token; on iPhone also a copy of Apple's attribution reports (Section 8.3)
Google (Google Analytics for Firebase; iPhone only)Measuring our own Google ads, from app version 1.2.6, if you agree (Section 8.1)From Google's software in the iPhone App: your install and first open, a trial started or plan bought during setup with its price and currency (for a plan bought, also the App Store's details of that purchase, such as its product and transaction number), Firebase's own installation code, your IP address, phone and app details and app-use time during setup and up to about a minute after it. Google Analytics reads neither your advertising ID nor the app-vendor ID. On Android, Google receives nothing for ad measurement
TikTok and AppLovinMeasuring our own ads on their networks, from app version 1.2.6, if you agree (Section 8.1)Nothing from the App directly. Through Tenjin: TikTok hears of each installation Tenjin measures, so it can check whether one of its ads led to it, and may hear of a trial or purchase only when one of its ads brought the installation; AppLovin may receive the install, trial or purchase only of an installation one of its ads brought. Each comes with the identifiers and details Tenjin received (such as your advertising ID, IP address and phone details)
USDA FoodData CentralNutrition reference dataSearch queries and barcode numbers, sent directly from your device — the service also receives ordinary request metadata (including your IP address) under its own policies
Open Food FactsBarcode product dataSearch queries and barcode numbers, sent directly from your device (plus ordinary request metadata, including your IP address); public data used under ODbL license

We encourage you to review the privacy policies of these third-party services. Providers have their own terms; our responsibilities under applicable law remain unchanged.

Bodydex Food Core (our own service, not a third party): from app version 1.2.6, if you are signed in and have turned on the Bodydex Food Core beta, your food searches go to our server with two more details: the country of your App Store or Google Play account (or US when it cannot be read) and whether your phone is set to French. While the beta is being tested, a search reaches Bodydex Food Core, our own food database hosted on Cloudflare, only if "Help improve food search" is also on, and only so that we can measure how well Food Core answers; its results are not shown to you. Food Core receives the search words, that country and that language setting; our server does not pass it your account identifier, your name or your IP address. Food Core keeps performance records (timings, counts and status codes), not a record of what you searched, and, to answer repeated searches faster, a cache of search words and their results that expires after 24 hours and holds no account identifier. Food Core does not send these searches to any other company. Searches through FatSecret and the public food databases listed in the table above work as described there.

8.1 Ad Measurement (optional; app version 1.2.6 and later)

In short. From app version 1.2.6, Bodydex contains ad-measurement tools from Meta, from our measurement partner Tenjin and, on iPhone, Google Analytics for Firebase. They help us see which of our own ads bring new installs, trials and purchases. They are optional and off unless you agree to them on the "Ad measurement" screen during setup (on iPhone, also by allowing tracking when your iPhone asks). If you agree, they run on your phone during setup and briefly afterward, then stop for good on that installation. Your food, health and other logs are never sent to them. The App shows no ads, and saying no changes nothing in the App.

When it is offered. The tools are switched off in the App and nothing of them starts unless all of the following are true. We have switched ad measurement on for new installs. That switch does not follow our ad campaigns: it can be on whether or not we are running an ad campaign, and while it is on, the conditions below decide whether the tools can start. The installation is new: not one we detect as updated from an earlier version of the App. The country or region of your App Store or Google Play account, the region set on your phone, and the network location our server sees (Section 3.2) are all in the United States. We do not start the tools when we detect you are in Washington, Nevada or New Jersey, and once we have detected that, ad measurement stays off on that phone for good. Ad measurement is not meant for people who live in Washington, Nevada or New Jersey, but the App can only detect where your phone is, not where you live; if you live there, please choose "Don't allow" (or "Ask App Not to Track"), also when you set up the App somewhere else. Your birthday in setup shows that you are 18 or older; if your age is not known, the tools stay off. Local-only mode is off (Section 12). On Android, the phone is not an automated Google Play test device. On Android, if any of these is not true, or the App cannot read our settings when the screen would appear, the screen is not shown and the tools stay off for that installation; nothing is asked later. On iPhone, setup shows the screen on every new installation whose birthday shows 18 or older while local-only mode is off, wherever you are. After you tap "Continue", your iPhone may ask whether Bodydex may track you; the conditions above then decide only whether the tools start: if any of them is not true, nothing starts, whatever you answer, and nothing is asked later.

What is sent, if you agree. Depending on the tool:

  • Events: that the App was installed and opened for the first time; and a trial you start or a plan you buy on the subscription screen during setup, with its price and currency. Tenjin receives no price for a trial. On iPhone, when you buy a plan during setup, Tenjin also receives the App Store's details of that purchase (such as its transaction number, product, price, currency, storefront and date) so it can check that the purchase is real, and Google Analytics receives the App Store's details of that purchase (such as its product, price, currency and transaction number). On Android, Tenjin receives the product, price and currency only. Purchases made with an App Store or Google Play test account, in TestFlight or App Review, Premium given to you by us, and Family Sharing send no event.
  • Identifiers: your phone's advertising ID (on iPhone, Apple's IDFA, only if you allow tracking; on Android, the Android advertising ID) and whether ad tracking is limited on your phone; the iPhone's app-vendor ID (IDFV), which Tenjin receives only when tracking permission and the other eligibility conditions allow its SDK to run; and the installation and session codes the tools create themselves. Google Analytics reads neither your advertising ID nor the app-vendor ID.
  • Network and device details: your IP address, which shows roughly where you are; phone details such as model, system and app version, language, time zone, carrier, screen size and, for Meta on Android, storage and processor details. On iPhone, Meta's software also sends the App's link schemes and the type name of the screen that is open. If Meta's own software fails while it runs, it may send Meta a report of that failure, which carries none of your logs.
  • App use during setup: Tenjin's session counts and lengths, and Google Analytics' engagement time during setup and, while it finishes sending, up to about a minute after it.
  • Where the install came from: on Android, the install referrer Google Play keeps for the installation (to Tenjin); on iPhone, Apple's ad-attribution token (to Tenjin).

These are sent from your phone directly to each company. Our server receives no ad-measurement event; it handles your phone's identifiers only to pass on a deletion request (below) and does not store them. If a tool fails to start, the App may send our server a count of that failure, with no identifier (Section 3.2). Meta's automatic logging of app events and purchases is turned off, so Meta's software sends only the events listed above; on iPhone, it also sends two technical notes with them: whether you allowed tracking, and that Meta's automatic event setup is turned off. The App gives RevenueCat, our subscription processor, no advertising identifier, and no RevenueCat integration sends anything to an advertising or measurement company for us. The App sets no account ID, name or email address in any of these tools.

Who receives it, and why.

RecipientRole and usePlatformsPrivacy policy
TenjinOur measurement partner, processing for us: it works out which ad brought an installation. It tells TikTok about each installation it measures, so TikTok can check whether one of its ads led to it, and may pass a trial or purchase on to TikTok, and the install, trial or purchase on to AppLovin, only when one of their ads brought you. It passes nothing to Meta or Google for us.iPhone, Androidtenjin.com/privacy
Meta (Facebook, Instagram)Independent company. Receives the events from its own software in the App.iPhone, Androidfacebook.com/privacy/policy
Google (Google Analytics for Firebase, Google Ads)Independent company. Receives the events from its own software in the App, which does not read your advertising ID.iPhone onlypolicies.google.com/privacy
TikTokIndependent company. Receives nothing from the App directly. Through Tenjin it hears of each installation Tenjin measures, so it can check whether one of its ads led to it, and may hear of a trial or purchase only when one of its ads brought the installation.iPhone, Androidtiktok.com/legal
AppLovinIndependent company. Receives nothing from the App directly. Through Tenjin it may receive the install, trial or purchase only of an installation one of its ads brought.iPhone, Androidlegal.applovin.com/privacy

This table is the complete list of companies that receive ad-measurement data from the App. If we add another ad network, we add it to this table and to the "Ad measurement" screen in the App before it receives anything.

Meta, Google, TikTok and AppLovin use what they receive to measure and improve our ads, under their own terms, which we do not control. Meta, TikTok and AppLovin may also combine it with other information they hold about you and use it to personalize ads. For Google, we have turned off Google signals and personalized advertising in our Google Analytics and Google Ads settings. Through these tools, these companies may collect information about your activity over time and across other apps and websites, but only after you agree. We build no audience, lookalike or remarketing list from our users, and we use no health-themed audience.

When it runs, and when it stops. Only during setup. If you agree, the tools start at once and send the install. A trial or purchase on the setup subscription screen is sent when it happens. When setup ends, no new install, trial or purchase is logged, but the tools may finish uploading shortly after setup: each tool stops once its last events have had time to upload, Meta after about 5 seconds, Tenjin after about 10 seconds and, on iPhone, Google Analytics after up to about a minute while the App is open and online. On iPhone, Google Analytics may also count up to one minute of use after setup. Once these tools stop, they send nothing more from that installation, and they never start again on it, also not after Reset & Start Over, Delete All Data, Delete Account or a new sign-in. Renewals, purchases made after setup, cancellations, refunds and anything else you do in the App after setup are not sent to any of these companies. If the App is closed during setup before setup ends, the next start of setup sends, once, only what was owed and not yet sent. Removing and reinstalling the App is a new installation.

What is never sent to them. Your food logs, weight, body measurements, medications, supplements, peptides, goals, lab results, photos, voice, Assistant messages, notes, setup answers, Apple Health or Health Connect readings, or anything else you log; your name or email address. The events carry generic names (install, trial, purchase), and our product names contain no health word. We do not give the contents of your logs (food, body, medications and the rest) or your Apple Health or Health Connect readings to advertising or measurement companies. These companies do learn that a phone installed Bodydex, a nutrition, body and medication tracking app, and whether it started a trial or a purchase during setup, and some laws may treat that fact itself as health-related information. Giving these identifiers and events to advertising companies may count as a "sale", "sharing" or "targeted advertising" under some US state laws; Section 11.5 explains your rights.

How to stop it. It ends by itself shortly after setup. During setup you can stop it on iPhone in Settings > Privacy & Security > Tracking, and on Android by deleting or resetting your advertising ID in your phone's settings (Privacy > Ads); the App checks both each time it starts or comes back to the foreground and then stops every tool at once, and it never uses a new advertising ID. Right after setup, while the tools finish sending, Profile > Ad measurement in the App has a switch that stops them at once. After that the row shows "Finished": nothing is left to stop, because nothing more is sent. Stopping does not recall what was already sent; see Retention and Requests and deletion, below. See also Your Privacy Choices.

Retention. Tenjin's terms with us have it delete or anonymize attribution data after up to 24 months, unless the law requires longer. In our Google Analytics settings, event-level and user-level data retention is set to 2 months; this does not cover Google Analytics' aggregated reports or conversions imported into Google Ads. Meta, TikTok and AppLovin keep what they receive under their own policies (Meta's business terms allow up to two years, and AppLovin says typically up to two years). On your phone, the App keeps the record of your answer described above and a note of which setup events were sent, never sent anywhere; on Android, Tenjin's software keeps a copy of the advertising ID in its own storage once it has started. Our server keeps no advertising identifier. For ad measurement it keeps only the deletion-request record described below; to stop deletion requests being abused, a count of the deletion requests sent each month from each internet address and from each block of addresses it belongs to, filed under a keyed hash of the address or block (never the address itself), which expires one day after the end of that month (UTC) and is then removed by our daily cleanup (later if a cleanup run fails); and the error and request counts of Section 3.2, with no identifier, for about three months. RevenueCat keeps no advertising identifier.

Requests and deletion. Tenjin keys its records to your phone's identifiers, not to your name or account. If ad measurement ran on your phone, Delete Account or Profile > Ad measurement > Delete ad-measurement data queues a deletion request for Tenjin, sent when possible, with the identifiers Tenjin received from that phone; our server does not store them. The App queues the same request by itself if your app store later reports an age under 13. It waits on this phone, and the app retries when a retry is due while it is installed and open. Our server sends only a limited number of these requests, so one can wait for a later month. A request still not sent after 60 days ends unsent (from version 1.2.7, 180 days if our server held it back because of one of its limits), and the app then shows how to contact us. Our server keeps a pseudonymised record of the deletion requests it processes, though not of every refused or repeated attempt: a keyed hash of the device identifier (never the identifier itself), the times, and our email service's acceptance when it gave one. Records are deleted after 3 years. Only a record with that acceptance shows that we sent the request; none shows that the data was deleted.Once the App is removed from the phone it can't be sent, and Tenjin's schedule under Retention applies. You can also write to at any time (Section 11.9); we answer within 45 days. You can also ask Meta, Google, TikTok and AppLovin directly through the privacy routes linked above. Your state-law rights, including opting out and how to appeal, are in Section 11.5.

Back to contents

8.2 Install Message to Meta (app versions before 1.2.6)

App versions before 1.2.6 do not include Meta's software. They send one install message through our server to Meta (the company behind Facebook and Instagram) the first time the App is opened after it is installed, so that we can tell whether our own ads work. No question is asked first. From app version 1.2.6 the App sends no install message on either platform; ad measurement in Section 8.1 replaces it, and only if you agree.

What it holds. The App sends our server: a random code made for this installation's install message (not your account identifier and not your advertising identifier), the platform, the App version and, on Android, the install referrer that Google Play keeps for the installation, when there is one, a short text that says where the installation came from. Our server sends Meta: that the App was installed, the time our server handled the message, the random code, the platform, and the App's name and version. Our server adds the Google Play install referrer only when it shows that the installation came from one of our Meta ads; an install referrer from anywhere else stays with our server and is not sent to Meta. No name, email address, phone number, IP address, advertising identifier, health measurements or logged records are included. The message does tell Meta that Bodydex was installed. With the install referrer, Meta can connect the installation to the ad that was tapped, and may be able to connect it to the person who tapped it. Meta uses what it receives to measure our ad and for its advertising services under Meta's own terms, which we do not control. We do not use the message to build a profile of you or to target ads at you. The message is never sent in local-only mode, and an update of an existing installation does not normally send one.

Sending, retries and our records. There is one install message per installation. If our server cannot be reached, the App tries again with the same code, on that launch or a later one, so that a repeat can be recognised. Our server keeps a record of each message: when it was received, the platform, the random code, the App version, whether Meta acknowledged it and the Google Play install referrer if there was one. That record expires about 90 days after it was written or last updated (it is updated when Meta acknowledges the message). A separate operational record notes the outcome of each message with the platform and a shortened one-way hash of the code; for a request that was refused, rate-limited or arrived while passing-on was switched off, the hash is made from the IP address instead. Operational records are kept for about three months. Server logs, which our provider keeps for a shorter time, note outcomes too; for an installation that came from one of our Meta ads, the identifying numbers of the ad campaign and its campaign group, taken from the install referrer; and, if a message cannot be passed on because of our own configuration, its random code. Your IP address reaches Cloudflare with the message, as with any internet request; it is used for a rate limit against abuse, is not passed to Meta, and is not written into the install message record. None of these records is stored against your name, email address or account.

Sending cannot be undone; you can still ask for deletion. Once Meta has received the message, we cannot take it back. To request deletion, write to us as described in Section 11.9. We will delete what we can identify in our systems and handle requests about recipients as the law that applies to you requires. We will tell you what we need to find the record, and any limits on what we can do.

8.3 Apple's Own Ad Measurement (iPhone and iPad)

Apple offers ad measurement that does not identify you, through two of its own frameworks. Neither is the advertising identifier, and neither needs a tracking permission.

SKAdNetwork / AdAttributionKit. The App tells this framework, on your device, how far an installation has got: app versions before 1.2.6 do so once, at the first open, with a fixed value of zero. From app version 1.2.6, the App does so only for a new install whose setup showed the "Ad measurement" screen of Section 8.1, whatever you answered there, and only if, when you answered (in the first iPhone builds of version 1.2.6: when the screen appeared), the conditions of Section 8.1 for the tools to start held apart from your answer (so not when you are outside the United States, when we detect you in Washington, Nevada or New Jersey, or while we have ad measurement switched off), and not if your iPhone restricts tracking, for example through Screen Time; and only during setup: a value for the first open, and a higher one if you start a trial or buy a plan on the setup subscription screen. If the installation followed an ad, Apple — not the App — may later send the ad network that showed it a delayed notice that an install followed one of its ads, and from app version 1.2.6 also a copy to Tenjin; Apple's notice carries no identifier for you or your device, and Apple decides what it contains and when.

Apple Ads attribution. RevenueCat's software in the App collects an attribution token from Apple's AdServices framework and sends it to our subscription processor RevenueCat, which exchanges it with Apple; RevenueCat's software decides when to send it and may retry. Apple may return attribution information, such as the campaign, ad group and keyword of an Apple Ads (App Store search) ad. An answer with no attribution does not prove that no ad was involved. From app version 1.2.6, if you agree to ad measurement, Tenjin's software also reads this token (Section 8.1).

From app version 1.2.6, the App starts neither when the country or region of your App Store account is one listed in Section 11.6, and it waits while it cannot read that country. It starts neither in local-only mode or once the birthday question has shown an age under 13 (Section 12). If the token collection had already started earlier in the same session, RevenueCat's software may still send the token until the App is closed.

Back to contents

9.Data Retention

  • Local data: Stored on your device indefinitely until you choose to delete it (via "Delete All Data" or by uninstalling the App). Because this data exists only on your device, we cannot recover it if the App is uninstalled or your device is lost, reset, or damaged. Profile → Export Data produces a portable copy of your records (JSON/CSV — a data copy, not a restorable in-app backup; image files and certain settings are not included).
  • Rate-limiting counters: Pseudonymous counters stored in our server infrastructure automatically expire and are deleted on a tiered schedule: daily usage counters within 48 hours, monthly usage counters within approximately 40 days, and lifetime free-tier anti-abuse counters (video recipe imports) within 365 days.
  • Trial records: While you are on a free trial we also keep the start and end dates of your trial period, stored against your account ID, together with the usage tallies described next.
  • Usage-limit tallies: Per-account usage tallies for Premium weekly and daily limits stop counting about 40 days after their last use. Usage tallies for a free trial's allowances stop counting no later than 401 days after the trial started, or 40 days after their last use if that is later. Tallies that have stopped counting are removed during routine maintenance of your account's usage records, not on a fixed schedule.
  • Family plan records: If you buy the Family plan, or join one with a seat code, our server keeps a Family record so the seats keep working: the account identifier of the purchaser and of each member who redeemed a code, the seat codes and the state of each seat (not yet claimed, active, revoked or expired, with its dates), the plan's product, renewal status and current period end, and the last 20 changes made to the seats. To give a member Premium, our server asks RevenueCat to grant it to that member's account until about the end of the plan's current period, to extend it when the plan renews, and to take it away when the seat ends. These records have no fixed expiry: we keep them while they are needed to run the seats and to stop a seat code from being used again.
  • Anti-abuse device identifier: A minimal, app-scoped device identifier is retained for fraud and abuse prevention and intentionally survives "Delete All Data" (see Section 11.2). It is pseudonymous and not intended to directly identify you.
  • Setup screen counts (from app version 1.2.6): kept on our servers only as hourly totals, deleted from our live database about 45 days after the hour they cover (recovery copies of that database expire within a further 30 days); the daily totals made from them are kept indefinitely. On your device they exist only in memory until they are sent (Section 3.2).
  • Genuine-app check (from app version 1.2.6): when the check is switched on, our servers keep the records listed in Section 3.2: a daily counter for each iPhone installation that fetches app passes, normally removed within about two days (a failed cleanup can delay this; recovery copies of that database expire within a further 30 days); a one-way record of each one-time code used, for up to about a day; a one-way record of each used app pass, for about two days; and totals with no code. On your device, app passes stay until they are used or expire, and on iPhone the key identifier stays, unless Delete All Data, an under-13 answer, local-only mode or uninstalling the App clears them earlier.
  • Install message records: our server's record of each install message expires about 90 days after it was written or last updated. Operational records of install messages are kept for about three months, and server logs for the shorter time our provider keeps them (Section 8.2). On your device, the random code and a note that the message was sent stay until the App is removed. What Meta keeps is governed by Meta's terms.
  • Ad measurement (from app version 1.2.6, only if you agreed): our servers keep nothing for it that can be linked to you apart from the pseudonymised deletion-request records of Section 8.1 (kept 3 years) and the monthly counts of deletion requests per internet address and per block of addresses, under a keyed hash (they expire one day after the end of that month and are then removed by our daily cleanup, Section 8.1), and counts with no identifier (Section 3.2). On your device, the App keeps the record of your answer on the Ad measurement screen and a note of which setup events were sent, until the App is removed; on Android, Tenjin's software keeps a copy of the advertising ID once it has started. What Tenjin, Meta, Google, TikTok and AppLovin keep is set out in Section 8.1.
  • AI-processed data: Food photos, a lab report you import as a photo or as a PDF, and the photo of a product label you import (Section 6.7), text messages, voice audio, and transcripts submitted for AI processing are NOT stored on our servers — they are processed transiently and discarded — with these exceptions, none of which is a document import: the result cache for a video link you import (up to 30 days; Section 6.4), the cache of answers to standard Research Library questions, which holds no personal context (up to 7 days; Section 6.2), opt-in scan-accuracy report photos (stored for troubleshooting; the 90-day automatic deletion control is not yet operationally verified — see Section 6.1), video-import status records (deleted within 7 days; see Section 6.4), and the working copy of an Assistant reply that paused mid-answer, whose content we attempt to replace with a contentless record as soon as the reply finishes and which — then, or if that attempt does not succeed — we attempt to remove on the criteria stated in Section 6.2, where a failed attempt can leave it in place longer — read that section for the criteria and its limits.
  • Retention by AI providers: Our providers may retain submitted content under their own terms after we have discarded it. For the Bodydex Assistant, OpenAI states a default abuse-monitoring retention of up to 30 days (longer where required by law); we have not arranged zero-data-retention (see Section 6.2). Other providers' retention is governed by their published terms (see Section 8).
  • Feedback submissions: Feedback rows are scheduled for deletion after 180 days; automatic deletion of every attachment is not yet operationally verified. If we cannot safely verify ownership of an older attachment, including where there are too many attachment references for automatic verification, we retain the report, our replies and its attachments beyond 180 days until the issue is reviewed and deletion can safely be completed. Copies delivered to our internal support tooling are retained as long as needed to investigate and resolve the issue (see Section 3.3).
  • "Trouble signing in?" messages (from app version 1.2.6): a daily cleanup on our server removes each message from our backend once it is 28 days old, so it is normally gone from our support system within 30 days (a failed cleanup can delay this); our internal support tooling keeps no copy. If our database provider keeps recovery copies of that database, a deleted message can remain in them until they expire. The form's network-limit markers are gone within 25 hours; its admission records (Section 3.3) are normally removed within about 27.5 hours, and failures can delay removal. A reply by email stays in our support mailbox, and anything you write back is kept as the next point describes. On your device, the App drops the form's current random code the next time it checks after more than 31 days have passed since the last send, and an older code it only remembers more than 31 days after the time recorded with that code. A code the App needs for a deletion request does not expire this way: it normally stays until our server has confirmed the deletion. The App keeps at most five older codes and drops the oldest when there are more, and removing the App removes them all (Section 3.3).
  • Email you send us: the copy in our internal support system is kept, in this order of priority, for 7 years (our legal address, and receipts and billing records from our vendors), 3 years (privacy, data and account-deletion requests, and our privacy address), 90 days (newsletters and suspected spam) or 3 years (all other email), each from when we marked it handled, else its last change of status, else its arrival; email about a request that is still open or waiting is not deleted because of its age. Cloudflare's copy on the way to our system is normally removed within minutes and is scheduled for deletion after 14 days (Section 3.3).
  • Subscription data: Managed by RevenueCat per their data retention policy. Cached subscription status on our server expires within 5 minutes.
  • Delete All Data and these periods (from app version 1.2.6): if you are signed in when you use Delete All Data, the App asks our servers to delete your feedback submissions with our replies and their attached files, and your video-import status records, without waiting for the periods above (Section 11.2). It does not shorten the periods for the rate-limiting counters, the trial records, the usage-limit tallies, the cached subscription status or RevenueCat's customer record. Those stay, to keep purchases working and to prevent misuse of free trials and usage limits. Copies of feedback submissions in our internal support tooling, email you sent us, and the notifications about feedback already delivered to us, are not removed by Delete All Data; email us to ask (Section 11.2).

10.Data Security

We take reasonable measures to protect your information:

  • Local encryption: Data on your device is stored in a local database protected by your operating system's built-in encryption at rest (iOS Data Protection / Android File-Based Encryption)
  • Transport encryption: All data transmitted between the App and our servers uses HTTPS/TLS encryption in transit
  • Secure proxy: AI service requests are routed through our secure server infrastructure — API keys are never exposed on your device
  • Server-side verification: Subscription status and rate limits are verified server-side to prevent tampering
  • Credential separation: The App includes only the public client credentials needed to access certain services; credentials capable of privileged access are kept server-side

While we implement commercially reasonable security measures, no method of electronic transmission or storage is 100% secure. We cannot guarantee absolute security. If a security breach affects your personal information, we will provide notification as required by applicable law.

11.Your Rights

You can manage your tracking records in the App. Requests about records on our servers follow the process described below.

11.1 Right to Access

Your tracking records are accessible to you directly within the App at any time; Profile → Export Data produces a copy of them, which excludes images and some settings (Section 9). Contact us about server-side records. For the install message in Section 8.2 and the genuine-app check in Section 3.2, see Section 11.9. For ad measurement, the only thing on our servers that can be linked to your phone is the pseudonymised record of a deletion request (a keyed hash of the device identifier, kept 3 years); to see it, write to us as below with the device identifier the request was made for. The monthly counts of deletion requests per internet address and per block of addresses (Section 8.1) are kept under a keyed hash, not under a phone, and expire one day after the end of that month. Section 8.1 explains how to ask for Tenjin's data for your phone to be deleted, by the App or by email, and how to ask the other companies. The setup screen counts carry no installation or account identifier, so there is nothing in them that we can find for one person.

11.2 Right to Delete

You can permanently delete the data stored on your device through the App (Profile → Delete Account; its confirmation is titled "Delete All Data", the name this policy uses), apart from a few technical items; the main ones are named below. This action is irreversible and removes all personal data from the local database and signs you out. Subscription and billing records are managed by Apple, Google, and RevenueCat under their own policies. The minimal device identifier kept for abuse prevention is preserved on your device (Section 9); it is pseudonymous — not intended to directly identify you — and may still be personal data. A few other technical items also stay on the device, among them what the App needs to finish deletion requests it has started (Sections 3.2 and 3.3), the yes/no setting that the age check was not passed, the yes/no setting that local-only mode is on and the copy of your last confirmed subscription status (Section 12), the install message's code and sent note (Section 9), and the ad-measurement answer and the note of which setup events were sent (Section 8.1). Deleting local data does not delete exported files, device backups, records you wrote to Apple Health or Health Connect, or content already processed by our AI providers under their retention terms (Section 9). Delete All Data does not delete our record of an install message sent by an app version before 1.2.6 (Section 8.2); to ask about it, see Section 11.9. If ad measurement ran on your phone, it also queues a deletion request for Tenjin, sent when possible (Section 8.1). What Delete All Data asks our servers to delete, and what stays there, is set out in the paragraphs that follow. It does remove from your device any support message still waiting to be sent (Section 3.3) and the app passes of the genuine-app check (Section 3.2). Uninstalling or reinstalling the App should not be relied on as a way to erase every copy.

What Delete All Data asks our servers to delete (from app version 1.2.6). If you are signed in when you use it, the App also asks our servers to delete what you sent us under your account:

  • the feedback, bug reports, food-data reports and support conversations stored under your account, with our replies and the files attached to them (Section 3.3); that is normally what you sent while signed in, with the exceptions below;
  • the status records of your video recipe imports, which hold the links you submitted (Section 6.4);
  • our records of which support message produced which reply (identifiers and a time, no text);
  • the working copy of a paused Assistant reply, if one is still held (Section 6.2).

The App erases the device first and does not wait for our servers. If the request cannot be delivered, or our servers cannot finish every step, the App keeps the request and sends it again later, normally until our servers have confirmed the deletion. Two things end it early. If you sign in to the same account again on that device before the deletion was confirmed, the request is cancelled, so that it cannot reach an account that is in use again; a request already on its way can still be carried out. When someone signs in on that device and the App cannot tell which account the waiting request belongs to, it cancels the request too. If you remove the App before the deletion was confirmed, the request is lost, and it also ends if the copy of your sign-in token that the App keeps for it (below) is lost from the phone's secure storage. In these cases, and to ask without the App, email us (Section 11.9). If you are not signed in, the App sends no such request. If our servers refuse the request on three different days, or it is still unconfirmed after 60 days, the App ends it and tells you how to ask by email. While a request is waiting, the App keeps a copy of your sign-in token in the phone's secure storage to prove the request is yours. The App normally removes the saved sign-in token when deletion is confirmed or ended. From app version 1.2.7, if the store's age answer is still pending, it can retain that token for up to three days after sending the request; if an under-13 deletion becomes required for that account, it retains the token while that deletion remains pending. If you delete a second account on the same phone before the first deletion was confirmed, the first request is cancelled and the App tells you.

What stays on our servers, and why. Delete All Data does not delete your trial records, your usage counters and usage-limit tallies, the cached copy of your subscription status, your customer record at RevenueCat (your subscription, or Premium you were given) or Family plan records. We keep them to keep purchases working and to prevent misuse of free trials and usage limits: deleting your data does not start a new free trial, does not reset a usage limit and does not take away what you paid for or were given. These records follow the periods in Section 9. Delete All Data does not cancel a subscription; you cancel it with Apple or Google.

What that request does not reach. Copies of your support conversations in our internal support tooling, email you sent us, and the notifications about your submissions (with any attached screenshots) that our operator notification service has already delivered to us (Section 3.3), are not removed by Delete All Data; to ask us to delete them, email us. Something that is still on its way to our servers or still being processed when the deletion runs, such as a message the App was sending at that moment or a video recipe import in progress, can be stored after the deletion; email us to have it deleted. The request also does not reach feedback, reports or support messages that are not stored under your account, with our replies and the files attached to them: those you sent while you were not signed in, and those you sent while signed in that the App could not send under your account. That happens when the App could not confirm your sign-in at that moment, or, in app versions before 1.2.7, when it sent the message by its backup route because our main server could not be reached, did not answer properly or could not take an attached screenshot that was too large. The App does not show you when this happened; email us to have such messages deleted. From app version 1.2.7 there is no backup route: a message that hits a temporary delivery failure is normally saved to the outbox and retried, within the limits in Section 3.3. Feedback rows are scheduled for deletion after 180 days; automatic deletion of every attachment is not yet operationally verified (Section 3.3). If we cannot safely verify ownership of an older attachment, including where there are too many attachment references for automatic verification, we retain the report, our replies and its attachments beyond 180 days until the issue is reviewed and deletion can safely be completed. Nor does the request reach a scan photo you chose to send with a rating (Section 6.1), the records that Apple, Google and RevenueCat hold under their own policies, or the recovery copies of our database, in which a deleted record can remain until they expire (Section 11.9).

"Trouble signing in?" messages. From app version 1.2.6, Delete All Data in the App also asks our server to delete any message sent from the sign-in screen's "Trouble signing in?" form by this installation of the App (Section 3.3). The App does not wait for the answer. If the request cannot be delivered, for example because the phone is offline, the App asks again later, including when it is next opened or brought back to the front. It keeps asking while it still holds the code for that message, normally until our server has confirmed the deletion; it holds at most five older codes and drops the oldest when there are more, and removing the App removes them (Section 9). The App can ask only for a message this installation sent, because the random code that names the message is kept on this installation. To have such a message deleted without the App, or one sent before the App was reinstalled or from another phone, email us from, or naming, the address you gave.

11.3 Right to Export (Data Portability)

You can export your data at any time through the App (Profile → Export Data). Data is exported in portable formats (JSON and CSV) that you can save, transfer, or use as you see fit.

11.4 Right to Revoke Permissions

You can revoke health data, camera, microphone, notification, and photo library permissions at any time through your device's Settings. Revoking permissions may disable certain features but will not affect your existing data. On Android, the App may also ask for the "Alarms & reminders" permission, which it asks for only so the rest timer's alert rings on time; no data is collected through it, and the timer still works without it. From app version 1.2.6, on iPhone, turning tracking off for Bodydex in Settings > Privacy & Security > Tracking, and on Android, deleting or resetting your advertising ID, stops ad measurement on that phone (Section 8.1).

11.5 US State Privacy Rights (California, Connecticut and Other States)

Wherever you live, and whether or not California's privacy law applies to a developer of our size, you can ask us what personal information we hold about you, ask us to correct it, and ask us to delete it. Section 11.9 says how, and you can write to . We do not sell your personal information for money. Delete All Data erases what is on your device; what it asks our servers to delete, and what stays there, is described in Section 11.2.

Ad measurement under state privacy laws. Giving identifiers and events to advertising companies, as ad measurement does from app version 1.2.6 if you agree (Section 8.1) and as the install message of app versions before 1.2.6 does (Section 8.2), may count as a "sale" or "sharing" of personal information, or as "targeted advertising", under some US state laws, including California's. We do not give the contents of your logs (food, body, medications and the rest) to advertising companies. When ad measurement runs, these categories go to the recipients named in Section 8.1: identifiers (advertising ID, the iPhone's app-vendor ID, the tools' own installation codes, IP address); commercial information (a trial started or a plan bought during setup, with its price); internet or other electronic network activity (the install and first open, and how long the App was used during setup); and the approximate location that an IP address shows. The install message of app versions before 1.2.6 sends Meta a random code, the platform, the App version and, on Android, Google Play's record of the ad. We offer no financial incentive for any of this, and saying no changes nothing in the App.

Opting out. Ad measurement runs only if you agree, and only during setup. To opt out, choose "Don't allow" (or "Ask App Not to Track" on iPhone); to stop it during setup, use the phone settings in Section 8.1; right after setup, use Profile > Ad measurement. Once the tools stop, shortly after setup (on iPhone, Google Analytics may count up to one minute of use after it), nothing more is sent from this installation, so nothing is left to opt out of in the App; to ask Tenjin to delete what it received, use Delete ad-measurement data, or write to us (Section 8.1). You can also opt out by email of any sale, sharing or targeted advertising, whether by ad measurement or by the install message of app versions before 1.2.6, as the next paragraph describes. A Global Privacy Control signal from your browser does not reach the App, and we cannot link a browser to an installation of the App, so the choices above and email are how you opt out of what the App sends; the website itself does not sell or share personal information, so the signal has nothing to switch off there (Section 14). Our Your Privacy Choices page puts these routes in one place.

Your rights and how to use them. Depending on your state, you may confirm whether we process your personal data and access it, correct it, delete it, obtain a portable copy of it, opt out of targeted advertising and of the sale or sharing of your personal data, and not be treated differently for using any of these rights. Write to with "privacy request" in the subject, or use Contact support in the App. We act on a request to opt out of the sale or sharing of your personal data or of targeted advertising as soon as we can, and within 15 business days; we answer other requests within 45 days, which the law lets us extend once by 45 days when we tell you why. If we refuse, you can appeal by replying with "APPEAL" in the subject ("health data appeal" for a consumer health data request); we answer an appeal in writing, with reasons, and if we deny it we tell you how to contact your state Attorney General. Section 11.8 and our Consumer Health Data Privacy Policy give the deadlines that apply in Connecticut, Washington and Nevada.

AI providers. We send your data to a provider only to perform the feature you asked for. What each provider may then do with it is set by its own terms and differs between them: Section 6.6 says which may use submitted content to improve their models. For the Bodydex Assistant, OpenAI's Data Processing Addendum records that OpenAI gives us nothing of value for your data, will not "sell" or "share" it, will not process it outside our direct business relationship, and will not combine it with data it holds about you from anywhere else.

The Meta install message. The install message of app versions before 1.2.6, described in Section 8.2, goes to Meta, which may use it under its own terms. The message itself holds no health measurements, logged records, name, email or advertising identifier, but it does tell Meta that Bodydex was installed. It is one message, sent when the App is first opened, and there is no setting in those versions to turn it off. From app version 1.2.6 the App sends no install message. Section 11.9 says what you can ask us for.

Setup counts and the genuine-app check. The setup screen counts in Section 3.2 carry no account or installation identifier and are kept only as totals. The genuine-app check in Section 3.2 uses, on iPhone, a key identifier and a daily one-way code, for security only. We do not sell either or provide them to advertisers; Cloudflare processes them for us (Section 8).

11.6 European Users (GDPR)

If you are located in the European Economic Area (EEA), you have additional rights under the General Data Protection Regulation (GDPR), including the rights of access, rectification, erasure, restriction, data portability, and objection. Since your data is stored locally on your device, you can exercise most of these rights directly. For any requests you cannot fulfill through the App — including requests concerning the limited server-side records described in this policy — contact us at (or ) and we will respond within the timeframe required by applicable law — ordinarily within one month of receipt for GDPR and UK GDPR rights requests, extendable where the law allows. You also have the right to lodge a complaint with your local supervisory authority. UK users: we acknowledge data-protection complaints within 30 days, investigate them, and tell you the outcome.

Legal bases. Where the GDPR or UK GDPR applies, we process: Bodydex Assistant messages and context, including health data, on the basis of your explicit consent (Articles 6(1)(a) and 9(2)(a)); the lab report you choose to import, as a photo or as a PDF, and the photo of a product label you choose to import, and what is read from either, also on the basis of your explicit consent (Articles 6(1)(a) and 9(2)(a) — a separate consent, asked before the first import and recorded against the version of the disclosure you read, Section 6.7); the food, label and menu photo scans and scan corrections, voice recordings and recipe imports the App sends to outside AI services when you use those features, as far as that is needed to give you the result you asked for, on the basis of our contract with you (Article 6(1)(b)) — the Terms and Privacy Policy agreement you tick before Get Started or Sign In in app version 1.2.7 or later also records your permission for this sending, but we do not rely on that tick as consent under Article 6(1)(a); what a provider does with content under its own terms, such as Google's use on its unpaid tier to improve its products (Section 6.6), is the provider's own processing under those terms (Sections 6.6 and 11.7); Apple Health / Health Connect data on the basis of the permission you grant on your device; subscription verification and delivery of the features you use on the basis of our contract with you (Article 6(1)(b)); rate limiting, abuse prevention, service monitoring, the setup screen counts and the genuine-app check in Section 3.2 on the basis of our legitimate interest in operating a secure, reliable service (Article 6(1)(f)); the copy we keep of email you send us, and the record of how we answered it, on the basis of our legitimate interest in answering and tracking requests, including privacy requests (Article 6(1)(f)); and feedback you choose to submit on the basis of your consent. Apple's ad measurement (iPhone) in Section 8.3 is a separate case: from app version 1.2.6 it does not run for the store accounts listed below, and where it runs the App does not ask for consent first. The install message of app versions before 1.2.6 (Section 8.2) is sent without a question. Ad measurement in Section 8.1 runs only for a store account in the United States, and never in the countries listed below; wherever a law requires a legal basis for it, its basis is your consent (Article 6(1)(a)), which you can withdraw as Section 8.1 describes.

Store accounts for which the App starts no Apple ad measurement and sends no setup counts and no genuine-app check. As an engineering choice, from app version 1.2.6 the App does not start Apple's ad measurement in Section 8.3, and does not send the setup screen counts, the genuine-app check or the ad-measurement settings request in Section 3.2, when the country or region of your App Store or Google Play account is a member state of the European Union, Iceland, Liechtenstein, Norway, the United Kingdom or Switzerland, or one of these territories: Åland, French Guiana, Guadeloupe, Martinique, Mayotte, Réunion, Saint-Martin, Saint-Barthélemy, Saint Pierre and Miquelon, Svalbard and Jan Mayen, Bouvet Island, the Caribbean Netherlands (Bonaire, Sint Eustatius and Saba), Gibraltar, Jersey, Guernsey, the Isle of Man, the Faroe Islands, Greenland, Andorra, Monaco, San Marino and Vatican City. While the App cannot read that country, it sends no setup counts, no genuine-app check and no ad-measurement settings request, and Apple's ad measurement waits until the country can be read (Section 8.3). This list is a product decision based on the store account's country, which the App checks when it starts. It is not a statement about which laws apply to you, and it does not depend on where you are or where you live: if your store account is registered in another country, these can be sent while you are in Europe. This choice does not change the rights you have under the law that applies to you.

Your records that stay on your device. The App's purpose is to hold your records for you, and it holds them in a database on your device: we have no access to that database, it is not synced to us, and nothing in it is sent anywhere except as this policy describes. Where the GDPR or UK GDPR applies, our basis for providing that storage is the performance of our contract with you (Article 6(1)(b)) — it is the service you chose — and what goes into it is what you enter, import or ask the App to keep. When you withdraw a consent, that withdrawal ends the basis for the processing it covered: we stop sending. We do not treat a withdrawn consent as permission to keep anything, and we do not rely on it as a basis for the records that remain on your device. Those remain because deleting your own history without your asking would be a loss you did not choose — they are readable and deletable by you, and Section 11.2 and Section 6.2 name the controls. If you would rather they were gone, use them; Delete All Data removes the records the App holds on your device, subject to the retained technical records and deletion credentials described in Sections 9, 11.2 and 12, and does not reach copies outside the App (Section 11.2).

International transfers. Data processed by our online features is transmitted to infrastructure operated by our service providers primarily in the United States and, where our edge network operates, in other countries. Bodydex is operated by Daniel Bobunov, a sole trader established in New Jersey, United States, who is the controller for the processing described in this policy and the customer under each of the provider agreements below. Where the GDPR or UK GDPR applies, these are the safeguards we rely on for transfers to the providers named in this list, and we do not rely on your consent as the transfer mechanism for any of them:

  • OpenAI (Bodydex Assistant; parsing voice-log transcripts). We use OpenAI's published, non-negotiated terms. Its Services Agreement provides at Section 5.3 that if the customer uses the services to process personal data, OpenAI and the customer will comply with OpenAI's Data Processing Addendum, "which is incorporated by this reference into the Agreement" — so the Addendum governs our account without a separate signature. Under that Addendum, EEA and Swiss data is processed by OpenAI Ireland Limited, and onward transfers out of the EEA or Switzerland are made on the basis of agreements containing the EU Standard Contractual Clauses or an adequacy decision under Article 45 GDPR; UK data is processed by OpenAI OpCo, LLC under the EU Standard Contractual Clauses as amended by the UK International Data Transfer Addendum, which the Addendum deems entered into between us as data exporter and OpenAI OpCo, LLC as data importer, with the Information Commissioner's Office as the competent supervisory authority. Read them at openai.com/policies/services-agreement and openai.com/policies/data-processing-addendum.
  • Cloudflare (the server proxy every AI request passes through, the rate-limiting counters, and the speech-to-text service used for voice features). Cloudflare's Self-Serve Subscription Agreement states that where customer content includes personal data of European data subjects, Cloudflare acts as processor or sub-processor and will handle it in compliance with Cloudflare's Data Processing Addendum, "which is hereby incorporated by reference into this Agreement." That Addendum applies the EU Standard Contractual Clauses to restricted transfers to Cloudflare, Inc. (United States) and deems the UK International Data Transfer Addendum executed between us and Cloudflare for UK transfers. Read them at cloudflare.com/terms and cloudflare.com/cloudflare-customer-dpa.
  • Apify (retrieving a video you share for recipe import). Apify's Data Processing Addendum is incorporated into its agreement so that "no separate signature is required," and incorporates the EU Standard Contractual Clauses and, for UK transfers, the UK Addendum. Read it at docs.apify.com/legal/data-processing-addendum.

The other providers named in Section 8 receive the content described for each feature in Section 6, and Section 6.6 states how each of them may use it under its own terms. Section 8 is the full list of recipients. If you want a copy of the agreement or transfer clauses we rely on for a particular provider, ask us at and we will send you the version that applies to our account. Your consent is a lawful basis for the processing, not a transfer tool. The Bodydex Assistant sends health data only after your explicit consent, and that consent is given in the knowledge that the data will be processed in the United States; it does not replace the safeguards above, and withdrawing it stops further sending (Section 11.7).

11.7 Right to Withdraw Consent

You can withdraw any consent you gave in the App as easily as you gave it, from the same device: Bodydex Assistant — Profile → Bodydex Assistant → Withdraw consent stops new Assistant requests at once, including further requests from a reply already in progress, and the Assistant stays off until you consent again. Withdrawing does not delete anything: your setup, conversations, memory, and notes stay on your device. Redo Assistant Setup is a separate, destructive option that also clears those. Document import — Profile › Legal & Data › Document import turns off both report reading and label reading; no document is sent after that, and the lab results you already reviewed and saved stay on your device as your own records, yours to keep, correct or delete. AI features (photo scans and scan corrections, voice logging, recipe import) — these send content only when you use them, so you prevent new requests by not using them; there is no separate switch. The permission in the Terms and Privacy Policy agreement applies when you tick it in app version 1.2.7 or later; users who ticked the earlier wording are not asked again (Section 6.6). A request already under way when you stop — for example a recipe import our server is already processing, or an automatic retry of a scan in progress — may still complete and reach the provider. To delete your tracking records from this device, use Delete All Data; some identifiers and technical records remain, as described in Section 11.2. Health platforms — revoke access in your device's Settings; Everything stored on your device — Profile → Delete All Data. Withdrawal does not affect the lawfulness of processing before withdrawal and cannot recall information already received by a provider, which remains subject to that provider's retention (Section 9). Ad measurement (Section 8.1) is based on your consent; how to withdraw it, and what withdrawing cannot recall, is set out there. The install message of app versions before 1.2.6 (Section 8.2), the setup screen counts, the genuine-app check and the ad-measurement settings request in Section 3.2 are not based on a consent, so there is no consent to withdraw for them, and there is no setting in the app to turn them off; Section 11.9 says what you can ask us for. You can also contact us at .

11.8 Connecticut Residents

Since July 1, 2026, the Connecticut Data Privacy Act applies to any person that controls or processes consumers' sensitive data, without a volume threshold (Conn. Gen. Stat. § 42-516(2), as amended by P.A. 25-113, § 6). We process exactly that, so the Act applies to us in full and we claim no exemption. Our Consumer Health Data Privacy Policy sets out the applicability reasoning, the categories, the recipients, the 45-day response deadline and the 60-day appeal deadline.

As a Connecticut resident you may confirm whether we process your personal data and access it, correct inaccuracies in it, delete it, obtain a portable copy of it, and opt out of targeted advertising, the sale of your personal data, and profiling in furtherance of automated decisions that produce a legal or similarly significant effect (Conn. Gen. Stat. § 42-518(a)). Access, export and deletion are direct in the App (Sections 11.1–11.3), and you correct a record by editing it where you entered it. For anything else, or to appeal a refusal, email — put "health data appeal" in the subject for an appeal; we answer an appeal in writing, with reasons, within 60 days, and if we deny it we tell you how to complain to the Connecticut Attorney General.

We do not sell personal data for money. We show no advertising in the App. Ad measurement (Section 8.1), which runs only if you agree to it during setup, and the install message of app versions before 1.2.6 (Section 8.2) give identifiers and setup events to advertising companies, which may count as targeted advertising, a sale or sharing under the Act. For ad measurement we ask for your consent first, and it is off for anyone whose age shows them to be under 18 or whose age is not known. We do not profile you for automated decisions that produce a legal or similarly significant effect. We do not collect, use or sell personal data to train large language models (Section 6.6). If any of that changes, this section changes with it and we will tell you before it takes effect (Section 15).

App versions before 1.2.6 send the install message in Section 8.2: one message per installation, sent through our server to Meta so that we can measure our own ads. It carries a random code and, on Android, Google Play's record of the ad when the installation came from one of our Meta ads; it tells Meta that Bodydex was installed. We do not use it to choose advertising for you or for profiling; what Meta does with it is governed by Meta's terms. Those versions ask no question about it, and there is no setting in them to turn it off. From app version 1.2.6 the App sends no install message, and ad measurement (Section 8.1) runs only with your consent; you can withdraw it as Section 8.1 describes, and it stops on your phone at once. Our Consumer Health Data Privacy Policy describes it, and Section 11.9 says what you can ask us for.

11.9 Install Messages, Ad Measurement, Setup Counts and the Genuine-App Check: Your Requests

The install message. For the install message of app versions before 1.2.6 (Section 8.2), write to us as below to ask what we hold or to request deletion. We will delete what we can identify in our systems and handle requests about recipients as the law that applies to you requires. Our records of install messages are stored against a random code only, not against your name, email address or account, and the App does not show that code. We will tell you what we need to find the record, and any limits on what we can do.

Ad measurement. Our servers and RevenueCat hold nothing from the ad-measurement tools of Section 8.1 that can be linked to you (RevenueCat's record of Apple Ads attribution is described in Section 8.3). The exceptions are the pseudonymised record our server keeps of the deletion requests it processes, deleted after 3 years (Section 8.1): it holds a keyed hash of the device identifier, never the identifier itself, so we can find it only when you give us that identifier, and we can then tell you what it shows; and the monthly counts of deletion requests per internet address and per block of addresses, kept under a keyed hash until they expire one day after the end of that month (Section 8.1). The quickest way to ask Tenjin to delete your phone's data is Delete ad-measurement data in the App (Section 8.1). You can also write to us as described above; we answer within 45 days and tell you any limits. You can also ask the other companies directly (Section 8.1).

Setup screen counts. The setup screen counts carry no installation or account identifier, so there is nothing in them that we can find for one person.

The genuine-app check. The records our servers keep for the check (Section 3.2) are held under one-way codes that change every day or belong to a single request, not under your name, email address or account, so a request made by name or account does not reach them. They follow the retention schedule in Section 9. An app pass carries nothing that identifies your phone, and we keep no list of which phone received which app pass.

Recovery copies. Our database provider keeps recovery copies from which a database could be restored to an earlier point for up to 30 days; deleted records can remain in those copies until they expire.

Requests by email. Write to with "privacy request" in the subject. We will answer and act on any request we can match to records we hold. This section does not limit any request about the other data described in this policy; Washington, Nevada and Connecticut residents also have the routes and deadlines in our Consumer Health Data Privacy Policy.

12.Children's Privacy

Bodydex is a general-audience health and fitness app. It is not directed to children and is not intended for anyone under 13 years of age. Setup works the same way for everyone until the App learns an age, so information from the setup steps named in the next paragraph can reach us before then. When the App learns that the person using it is under 13, it stops sending information from that installation, apart from the few requests listed in this section, and asks our servers to delete what they hold under that account, as described here.

How the App learns an age. The App can learn that the person using it is under 13 in two ways: from the age group that the App Store or Google Play reports for the account on the device, where the law requires the store to provide it (Section 3.4), and from the birthday question in setup. On iPhone, if you allowed Bodydex to read your date of birth from Apple Health earlier in setup, that date is filled in on the birthday question for you to keep or change; the App acts on it only when you continue past that question. Before then, setup works the same way for everyone: sign-in, the subscription service, in app versions before 1.2.6 the install message (Section 8.2) and, from app version 1.2.6, the setup screen counts run as described in Sections 3.2, 3.4 and 8. Ad measurement (Section 8.1) is never offered before the birthday question, and it is off for anyone whose birthday shows an age under 18 or whose age is not known. From app version 1.2.6 the "Trouble signing in?" form on the sign-in screen (Section 3.3) also comes before the birthday question, so a message can be sent from it before the App knows any age.

The birthday question asks you to confirm an age under 13. From app version 1.2.6, if the birthday entered shows an age under 13, the App first shows that birthday back, says the age it makes, and asks whether it is right. Like every setup answer, the date is saved on the device with your unfinished setup as you enter it. Until you confirm it, nothing is blocked, deleted or sent because of it, and choosing to change it goes back to the date. A birthday that shows an age of 13 or over is not shown back.

If the birthday is confirmed and shows an age under 13. Setup stops and Bodydex cannot be used on that device. On the device, the App deletes the birthday that was entered and the saved setup progress, and saves no setup progress afterwards. It also deletes any support message still waiting to be sent, support form drafts and screenshot copies, the code used for support-reply notifications, and any app passes and key identifier from the genuine-app check (Section 3.2). It discards setup counts not yet sent, starts no ad-measurement tool, and runs no genuine-app check. Of your answer, the App keeps only a yes/no setting that the age check was not passed, so that the answer cannot simply be changed. It also keeps a few technical records that hold no birthday and no age: whether its deletion requests were confirmed, and a note that the push-notification registration was deleted. From app version 1.2.6, if a message was or may have been sent from the "Trouble signing in?" form on that device, the App also keeps that form's random code until our server has confirmed the deletion described in the next paragraph; anything typed into the form and not sent is dropped, and the link to the form is no longer shown. Files kept by Google's and RevenueCat's software inside the App (a switched-off registration file, the old installation and customer codes) stay until the App is removed.

What the App still sends after that. The App sends our server one request to delete what the earlier setup screens created: any records our server holds under that sign-in's account identifier, and the customer record that our subscription processor RevenueCat created for it, which includes the country RevenueCat recorded and, on iPhone, Apple's ad-attribution token if one was passed on (Sections 3.4, 8 and 8.1). This is a fuller deletion than the one Delete All Data asks for (Section 11.2): it also covers the usage, trial and Family records that are kept for an adult. The App repeats its request until our server has confirmed the deletion, except as described next, and then signs out. From app version 1.2.7, an account-linked under-13 deletion is sent only with credentials belonging to that account. An older pending request is matched using its saved deletion credential, never another account signed in after updating. If the account or its credentials cannot be established, the App gives an email-request notice and waits for at least 60 days from the under-13 answer before ending the unresolved request. A request created without a signed-in account may still ask for deletion of the installation's anonymous RevenueCat customer record; if there is neither a sign-in nor such a record, the App sends no request. From app version 1.2.6, if a message was or may have been sent from the "Trouble signing in?" form on that device, the App also sends our server a request to delete it, and repeats that request while it still holds that form's code, normally until our server has confirmed the deletion; our server deletes every message stored under that form's code. On Android, the App also asks Google once to delete the installation's push-notification registration; if Google cannot be reached, it repeats that request each time the App is opened, until Google has confirmed it, and then never again. Two more things can happen. A request that was already under way when the birthday was confirmed can finish. And each time you leave the App and return to it without closing it, the subscription software inside the App (RevenueCat) can contact its servers again, with the customer code it already had. After the App has been closed, a blocked installation does not start that software again. Apart from these, for as long as that setting is on the device, the App makes no online requests from it.

What our server does with that request. When the request arrives, our server normally deletes the records it holds under that account: the feedback, reports and support conversations sent while signed in, with our replies and the files attached to them; video-import status records; usage counters, usage-limit tallies and trial records; the cached subscription status; and Family plan records. It also asks RevenueCat to delete the customer record. It does not ask this if, just before the deletion, RevenueCat shows that record linked to another account, because that would delete the other account's purchases. Our team is then alerted, and that record is not deleted while it stays linked to the other account. If a step cannot be finished, our server answers that the deletion is not done, and the App sends the request again. About a day later, our server normally checks once, without deleting anything, whether RevenueCat holds a customer record for it again (the subscription software can create one again shortly after a deletion), whether our server holds its cached subscription status or trial record again and, for an account, whether it holds a Family plan seat again. If RevenueCat is briefly unreachable and nothing else is found, our server checks one more time at a later daily run. If something is still there, or the check cannot be finished, a member of our team is alerted, checks it and, within 30 days of that alert, deletes by hand what is still held for that request, except a RevenueCat record linked to another account (see above). If the deletion of an account has already started but did not finish, our server keeps the request and normally, at its next daily run, alerts our team in the same way, even if the App is no longer installed; our team then finishes the deletion by hand within 30 days of that alert, unless the App's own repeated request has already finished it. Each daily run starts at most two such hand-overs, and any others wait for a later day. While that one more check is waiting, and when our server cannot record a hand-over, our server normally keeps the identifier, in encrypted form, under a separate hand-over key and tries again at its next daily run. That key normally expires 7 days after it was first kept, even if the hand-over is recorded sooner, the later check finds nothing left or the account signs in again; while a hand-over still cannot be recorded, the key is renewed for another 7 days before it would expire. Our server does not run a deletion again by itself. This check and this hand-over normally do not happen if the account signs in again before them, and an unfinished deletion whose hand-over has not started yet waits while we have switched this deletion off on our server (for example, to fix a fault) until it is switched back on. Our server keeps one marker for that account: a keyed one-way name, with the account identifier in encrypted form until that check is done and any hand-over is recorded, or until the marker expires if that comes first. While the identifier is also kept under a hand-over key, it stays in the marker until the marker expires. It is needed for that check, and while it exists the account's paid online features normally stay closed, so that a deletion cannot be used to reset usage limits. The marker normally expires about 48 hours after the request, and later while an unfinished deletion is still waiting to be handed to our team. If a deletion cannot be completed or recorded, we may keep a separate record containing the deletion marker's keyed identifier, encrypted account or subscription identifier, failure reason and timing information so we can investigate and resolve the failure; when a deletion is handed to our team, such a record holds what they need to finish it. Each record expires within 30 days of its creation, including if the account is deleted meanwhile; our server never deletes anything because of it. While the failure remains unresolved, a new record may be created on each day it is observed, and each one expires within 30 days of its own creation. A request whose sign-in token expired more than 425 days before it arrives cannot be matched to an account, so nothing is deleted because of it. If a deletion cannot be finished this way, a parent can have the records deleted by email, as described under Parents below. Copies of support conversations in our internal support tooling (Section 3.3) are not removed by this request; to have them deleted, write to us as described under Parents below.

Records that are not kept under the account. Setup counts that were already sent are part of totals that carry no account or installation identifier, so nothing in them belongs to one person. In app versions before 1.2.6, the install message in Section 8.2 is sent when the App is first opened, before the App knows any age; our server's record of it is held under a random code, not under the account, cannot be found from the account, and expires about 90 days after it was written or last updated. If the birthday first entered showed an age of 13 or over, the genuine-app check may already have run before the answer was changed: the daily counter our server keeps for it is not kept under a name or account and follows its own schedule (normally removed within about two days; recovery copies expire within a further 30 days). From app version 1.2.6, the network-limit markers of the "Trouble signing in?" form (Section 3.3) are kept under a keyed value made from the network and the day, not under the form's code, so the App cannot ask for them; they hold no address and nothing of the message, and are gone within 25 hours. The form's admission records (Section 3.3) hold no network-derived value, address or message text, are not removed by a deletion request, and are normally removed within about 27.5 hours; failures can delay removal. If the App was removed and installed again before the birthday was answered, the form's code is gone and the App cannot ask for the message; the daily cleanup then removes it (normally within 30 days), or we delete it sooner on a request by email. A reply we had already sent by email stays in our support mailbox like any other email, and email sent to us is kept as Section 3.3 describes; no request from the App can reach a mailbox or that copy, so write to us as described under Parents below.

If the App Store or Google Play reports an under-13 account (local-only mode). The App switches that installation to local-only mode. Local-only mode blocks new online-feature requests: the App starts no new requests to our servers, our AI providers, our subscription processor or any other online service, and sends no setup counts, no install message, no ad-measurement settings request and no genuine-app check, and starts no ad-measurement tool; it deletes the app passes and key identifier it holds for that check. Deletion requests are the exception. The App may still send our server a request to delete the records our server and RevenueCat hold under that account, which is handled as described above; from app version 1.2.6 it may also send our server a request to delete a message sent from the "Trouble signing in?" form on that device; from app version 1.2.6, if ad measurement ran on that phone, it may also send, through our server, a request for Tenjin to delete that phone's ad-measurement data, with the identifiers Tenjin received; our server does not store them (it keeps only the pseudonymised record of the request described in Section 8.1), and until the request is sent they wait on the phone for as long as Section 8.1 says; and on Android it asks Google to delete the installation's push-notification registration, repeating that request at each launch until Google has confirmed it. The App starts deleting the registration when the store reports an under-13 account, and tries again at later launches if the deletion is not confirmed. If the store later corrects itself, the App registers again by itself when reply notifications were on; a support reply sent in between is shown in the App once the mode has ended, but may not have been shown as a notification. The code used for support-reply notifications stays on the device in this mode, because it is what lets the App register again. One more thing can still happen: if the mode starts while the App is already open, the subscription service started earlier in that session (RevenueCat) may continue to check your subscription with the store until the App is closed; it is not started again after that. A request already under way when the mode starts can finish. If the genuine-app check had already run on that installation, the daily counter our server keeps for it follows its own schedule, as described above. The App keeps working on the device and your tracking records remain available on it: everything already recorded stays where it is, while the features that need the internet — scanning, the Bodydex Assistant, online food search, imports, backup, sign-in, purchases and support messages — are off, and the App says so where you would use them. From app version 1.2.6 that includes the "Trouble signing in?" form: in this mode it shows only a notice and sends nothing. Links you tap to our website or to the store still open in your browser. Of the store's answer, the App remembers on the device only that local-only mode is on — a yes/no setting, never the age or the age group — so that the mode holds from the next launch, and it turns the mode off by itself when the store later reports an older age group; the App then cancels a deletion request it had not yet seen confirmed by our server, including one for a "Trouble signing in?" message (an account deletion our server has already started may still be finished, by our server or by our team, as described above). So that paid features you already have stay available without contacting the store, the App keeps on the device a copy of the last subscription status the store confirmed (every installation keeps this copy, and the App never sends it anywhere) and uses that copy while the mode is on.

The first start after an update. On its first start after an update to app version 1.2.6, an installation that was already blocked by the birthday question or already in local-only mode, and that had support-reply notifications turned on, may send its push registration to Expo's push service once more. From the second start on it does not.

Parents. If you are a parent or guardian and believe your child under 13 has used Bodydex, email us at . Tell us what you can — for example the Apple or Google sign-in that was used, or the installation's date and platform. We will look for the records we hold under that account, delete what we can find, ask RevenueCat to delete its customer record, and tell you what we did and any limits. If your child used the "Trouble signing in?" form (from app version 1.2.6), tell us the email address that was typed into it, so that we can find and delete that message too. Setup counts and other records that carry no account or installation identifier cannot be traced to a child. If a child under 13 writes to us, we answer once if an answer is needed and then delete the message and the address it came from.

Users between the ages of 13 and 18 may use the App only with the consent and supervision of a parent or legal guardian.

13.Consumer Health Data Privacy Policy

Our Consumer Health Data Privacy Policy explains the consumer health data we process, its sources and purposes, the recipients it is shared with, and how to exercise your applicable rights, for the purposes of the Washington My Health My Data Act, Nevada SB 370, and the Connecticut Data Privacy Act. It is published at its own address and linked from every page of this site. In short: consumer health data is stored on your device for tracking; selected records and summaries are processed through the online features described in Section 6, with your explicit consent for the Bodydex Assistant and for document import; we do not sell the health information in your logs, do not share it with data brokers, do not give your health logs to advertising or measurement companies, and do not use geofencing. Ad measurement (Section 8.1) and the install message of app versions before 1.2.6 (Section 8.2) hold no health measurements, logged records or health-related setup answers, but they do tell the companies that receive them that Bodydex, a health app, was installed (and, for ad measurement, whether a trial or plan was started during setup). Ad measurement does not start for people we detect are in Washington, Nevada or New Jersey, and stays off for good on a phone once we detect it there; that policy describes both. It also describes the setup screen counts in Section 3.2, which carry no installation or account identifier, and the genuine-app check. Neither includes your health measurements or logged records.

14.Website Privacy

The Bodydex marketing website (bodydex.app) uses Cloudflare Web Analytics to measure page traffic and understand which pages visitors reach. It is privacy-first by design: it uses no cookies, no tracking pixels, and no fingerprinting, and it does not track you across other websites. It records only aggregate, non-identifying information such as page views, referrer, country, and device type.

Cloudflare processes website request metadata; we count store clicks and QR redemptions using same-origin beacons without adding cookies or user identifiers. The site's fonts are served from our own domain. We do not sell any data collected through the website. The website is a static informational site hosted on Cloudflare Pages.

14.1 Website and Free Tools

The free tools on bodydex.app, such as the Peptide Reconstitution Calculator, run entirely in your browser. What you type into them is not sent to us or to any other service: we do not receive, collect or store it, and it is never added to the page's web address. It leaves your browser only if you copy or share it yourself. The tools set no cookies and contain no advertising trackers.

A tool may offer an optional "Remember my numbers on this device" box. It is off unless you tick it. While it is ticked, your last figures are kept in your own browser's local storage on that device so the page can show them on your next visit; they are not sent to us. "Clear" removes the saved figures, unticking the box deletes them and the setting, and clearing this site's data in your browser removes both.

Like every page of the website, the tools are delivered by Cloudflare, our host, which processes standard technical data such as your IP address and request metadata to deliver and secure the site. The Cloudflare Web Analytics measurement described above also runs on these pages. It receives technical measurements such as the page address, browser and operating-system version, and page-load and responsiveness timings, which can name the part of the page you used (such as a field), but not what you type into a tool.

Do Not Track and Global Privacy Control. The website does not track you across other websites, so it behaves the same whether or not your browser sends a Do Not Track signal. The website also does not sell or share personal information and uses none for targeted advertising: it sets no cookies and runs no advertising or tracking pixels, and Cloudflare, our host, runs the Cloudflare Web Analytics described above for us as our processor, under terms that do not let it sell or share that data. A Global Privacy Control signal therefore has nothing to switch off on the website, and the site works the same with or without it. The signal does not reach the App, and we cannot link a browser to an installation of the App: ad measurement in the App is chosen in the App and on your phone (Section 8.1), and you can opt out of it, or of the install message of app versions before 1.2.6, by email (Section 11.5).

15.Changes to This Policy

We may update this Privacy Policy from time to time. When we make changes, we will update the "Last Updated" date at the top of this page. It is your responsibility to review this Privacy Policy periodically for changes.

We will notify you of material changes and obtain fresh consent before changed processing where the law requires it. Continued use alone does not provide that consent.

If we make material changes to the way we handle your data, we will make reasonable efforts to provide notice through the App or our website.

16.Contact Information

For questions, concerns, or requests regarding your privacy or this policy:

Privacy inquiries:

General legal inquiries:

Developer: Daniel Bobunov