SpiritArc Privacy Policy
Effective: 2026-11-09 · Last updated: 2026-11-09
Operator: Baskar Ellangudi, an individual operating SpiritArc, 264, Periyar Street, Chennai - 600016, India. Privacy and deletion contact: privacy@spiritarc.fit.
1. Scope
This Policy covers the SpiritArc Android beta, guest use, and the website at spiritarc.fit, including its waitlist and beta feedback. Website signup and mailing-list records are separate from an Android account. Disabled features do not collect information merely because supporting code exists. The first internal beta is for adults aged 18 or older in India; AI meals, meal replacement, camera form correction, purchases and referrals are disabled.
2. Information we process
Accounts and profiles: Email, name, optional avatar, account and login-provider identifiers, authentication sessions and the profile information you supply. Profile fields can include age or birth date, gender, height, weight, goals, fitness level, equipment, schedule, dietary preferences, cuisines and disliked ingredients. Onboarding derives an approximate birth date from your entered age; profile editing can correct it.
Activity and content: Selected or generated workouts and plans; exercise targets, sets, repetitions, times and weights; completion, ratings and notes; estimated calories; progress, achievements and rewards; food-log entries, portions and nutrition estimates; saved meal records; eligible coach messages and replies; feedback text and optional images. An avatar is served from a public bucket: avoid an image you do not want accessible to someone who obtains its URL. Free text can contain sensitive information you choose to enter. Do not submit medical records, credentials or another person's private information.
Guest and device data: Guest activity is stored on the device, and supported records may be imported on sign-in. Sessions use secure device storage; guest and offline records use encrypted local storage. Network requests can expose ordinary connection information such as IP address to the service or provider handling the request. Guest exercise-media verification uses Cloudflare. App updates, media delivery, food search, notifications and device diagnostics involve the providers needed for those features.
Website submissions: The waitlist collects email, signup source and campaign attribution values from the URL. Feedback additionally collects the goal, challenge, feature interest and willingness-to-pay answers you submit. Supabase stores these records. MailerLite receives email, configured group assignments and a waitlist-position field for waitlist signup; website feedback answers are not sent as MailerLite subscriber fields by our integration. These forms do not call an AI model or create an Android account.
3. Purposes and providers
We process information to provide and secure accounts, personalize and save your chosen activity, display progress, deliver media and reminders, answer eligible coach requests, respond to support/privacy requests, investigate abuse and diagnose failures. Website information is used to record beta interest, evaluate feedback and send the invitations or product updates requested through signup.
Providers include Supabase for authentication, database, uploaded files and backend functions; Groq for enabled AI requests; Google for Sign-In, Android push transport and applicable AdMob/Play services; Expo for app updates and push delivery; Cloudflare for guest verification; PostHog for optional app product analytics; Sentry for diagnostics; Open Food Facts for requested food searches; and hosts serving external media selected in the app. Vercel hosts the website and supplies its Web Analytics; MailerLite handles website mailing subscriptions. RevenueCat can hold historical subscription records from enabled purchases or earlier tests. A disabled purchase flow does not erase historical provider records.
We share information needed for the provider's function, respond to lawful requirements and protect users and service security where appropriate under applicable law. We do not offer personal information for sale. These services can process information outside India. The app's Supabase project is in Tokyo; that does not establish every provider's processing or backup location.
4. AI processing
Workout-program and eligible coach requests use Groq's server-side API and the openai/gpt-oss-20b model. Program context can include age, goals, level, equipment, schedule, available exercises and summarized previous activity/feedback. Coach context includes your message, relevant chat history and goals/level. The handler does not intentionally add your name, email or account ID; text you enter can still identify you. This processing is not anonymous.
Plans, replies and operational AI records are saved in Supabase. Provider retention controls do not delete that saved history. We do not use Groq batch or fine-tuning features for these requests, and there is no fallback to unpaid Gemini processing. AI meals are disabled and send no meal/allergy prompts in this beta.
Groq documents usage metadata separately from prompt/output content. Its ordinary inference content is not retained by default, with possible reliability/abuse logs for up to 30 days unless a legal requirement applies; Zero Data Retention disables that reliability/abuse content storage. The operator reports enabling ZDR. This does not mean every provider record or our own records have zero retention. See Groq data controls for its handling and controls.
5. Analytics, diagnostics, advertising and choices
New PostHog SDK installations default to opted out. Optional product analytics can be controlled in Help or Account Settings on that installation. Enabled app events can include registration, login, onboarding, generated-workout counts, workout completion duration/counts and food logging, associated with an account ID when signed in. Automatic screen, touch and lifecycle capture are disabled in our configuration. Opt-out does not itself erase earlier events or disable diagnostics, advertising or website analytics.
Sentry receives diagnostic events, selected device/SDK metadata, account ID and navigation/performance information. Current filters remove raw error text, arbitrary request/context fields, frame locals and breadcrumb text/data from the filtered JavaScript events; stack locations and selected metadata remain. Filtering is not a guarantee that every native or provider-added field is anonymous. Earlier releases may have sent account emails. Contact us about historical account-linked records.
The first internal beta uses test AdMob ads. Advertising SDKs can process identifiers, connection information, interactions and diagnostics even for non-personalized requests. No promise that a non-personalized or test request has zero data processing is made. Device and Google settings provide additional choices. Notifications can be disabled through the app or Android settings.
The website includes Vercel Web Analytics, separately from Android choices. Vercel describes aggregate page/visit measurement using request-derived identifiers rather than third-party analytics cookies; hosting/security logs are a separate service. URLs, referrers, browser/device and approximate location can contribute to website measurements. Campaign parameters are also saved with form submissions. MailerLite may process email delivery and engagement information under its settings and terms. Android settings do not control website or email processing. See Vercel Analytics privacy information.
6. Retention
Retention differs by category. The following periods apply to the stated operational records, not every copy held by an independent provider. A saved plan or chat is different from a duplicate operational log. Deletion and expiry tasks are scheduled processes rather than a promise that every copy disappears at the precise instant a period ends.
| Category | Current handling |
|---|---|
| Account, profile and saved workout, food, coach and progress history | Kept while the account remains open to provide its history, then removed through verified account deletion, subject to the exceptions below. Subscription expiry alone does not delete history. Uploaded account files are removed during the account-deletion process. |
| Duplicate AI operational content and detailed metadata | Eligible duplicate content is cleared after 7 days; detailed operational metadata is reduced after 90 days. Saved account plans/chat, quota integrity and necessary references are separate. |
| Unowned operational billing-event audits | Eligible minimized records expire 365 days after their later detachment/activity clock. Account identifiers in eligible payloads are minimized during deletion. Necessary transaction references and account hashes can remain until applicable expiry. |
| Eligible referral-attempt and rate-limit logs | Expire after 180 days. This does not delete active rewards or canonical financial/entitlement records. |
| Canonical purchase, entitlement, reward, security and dispute records | These do not have a blanket 365-day purge. Necessary records can remain to handle existing entitlements, reconciliation, fraud/abuse, disputes or applicable legal obligations. Hashes are pseudonymous, not anonymous. Retention is assessed for the category and purpose when handling a verified request; we do not promise a fixed expiry for every such ledger. |
| Website waitlist, feedback and mailing records | No automatic time-based expiry is currently applied to website submissions. They remain for beta-interest/contact and feedback purposes until removed through the website request process. MailerLite records are separate; unsubscribe stops the relevant mailing but is not a promise of erasure of every provider record. |
| Analytics, diagnostics, hosting, email and other provider records | Subject to the relevant service's retention and necessary security, delivery or legal handling. Our app-log periods do not set an expiry for these services. We assess applicable account-linked records and provider deletion options when you make a verified request; no universal 30/90-day provider deletion guarantee is made. |
| Device and cached records | Offline query data has a maximum read-cache age of 14 days and expired snapshots are removed/pruned when read. There is no closed-app wall-clock erasure timer. Guest history and other local state have separate lifetimes. Account deletion clears account-scoped local state through the app; native/media caches and OS copies may remain. Clear app storage or uninstall to remove remaining app-controlled local data, remembering that this does not cancel a subscription. |
| Backups and recovery copies | Separate from live records and not necessarily erased immediately by account deletion. Availability and expiry depend on actual provider plans and any recovery copies. We do not promise a verified universal 30-day backup expiry. Deletion records must be reapplied before restored data is returned to service. Contact us for the applicable status of a request. |
7. Delete an account or website data
Use Account Settings → Delete Account, or follow the web deletion instructions without reinstalling the app. We may verify ownership before acting. The app deletion process removes the Auth account, associated account data and uploaded avatar/feedback files; retained operational, financial, provider, backup and cached exceptions are described above. Failed or partial processing can require retry or follow-up. Deletion does not cancel Google Play renewal.
Email privacy@spiritarc.fit from the address used for the account or website submission. Say whether the request concerns the Android account, website waitlist/feedback, mailing records or all of them. The operator handles the request and applicable provider follow-up. Website records and MailerLite records are separate from the app; deletion of one does not establish deletion of the others. The signed unsubscribe integration can remove matching website rows when a valid provider event is delivered, but you can use the contact route if that has not happened.
Our service targets are acknowledgement within 2 working days and routine completion within 30 days after ownership and scope are verified. We will explain verification needs, delays and retained exceptions. These are service targets, not a restriction on any shorter deadline or other right that applicable law provides. Do not send passwords, OTPs, payment-card details or identity documents with an initial request.
8. Access, correction, children and contact
The app lets you view and edit supported profile/activity fields and manage account access. You can request access, correction, deletion, withdrawal of optional choices or other privacy assistance at privacy@spiritarc.fit. A general in-app export is not currently offered. We verify ownership proportionately before disclosing or changing records and assess requests under applicable law. Raise a privacy concern or grievance with Baskar Ellangudi through that address or the operator's postal address above. You retain any right to approach a competent authority.
The beta is not offered to anyone under 18, including guests. We use a self-declared adult check and supplied age/birth date; we do not independently verify identity documents. If someone under 18 supplied personal information, contact us so we can address it. India is the initial intended market; this statement is not a claim of India-only processing or certification of compliance with every applicable privacy law.
We update this Policy as the service changes, show the updated date and provide additional notice where required. Contact privacy@spiritarc.fit for privacy/deletion matters or support@spiritarc.fit for service and billing assistance.