English translation. The contract language is German. Read the German original.
Privacy policy
Last updated: 21 November 2026
1. Controller
Mario Falkenberg, Graphitweg 9, 12349 Berlin, Germany, email: support@metaflex-rhythmus.de, contact form: /kontakt. No data protection officer has been appointed; please contact the address above directly with questions about data protection.
2. What this is about
Metaflex is a nutrition, training and weight diary with coaching features that you use through your own AI (such as Claude, ChatGPT or Grok). This website is used to sign in with an email address and password (with an email code as the recovery method), to take out and manage your subscription, and to manage your OAuth app sign-in and connection key.
3. Data we process
3.1 Diary data (nutrition, training, weight, body measurements, check-ins)
- Content: What you or your AI send to Metaflex, such as meals with nutritional values, workouts, weight, waist circumference, steps, body measurements, wellbeing (scale 1–5), profile information (goal, height, birth year, sex, activity) and — only with your explicit consent — health information (such as medical conditions or medication including GLP-1).
- Encryption and processing: Diary contents are stored encrypted using a key derived from your connection key (AES-256-GCM). With the older key-based method via /mcp, your AI supplies the connection key for each call. For sign-in with an email address and app password via /app/mcp, we store the connection key in encrypted form and use encrypted OAuth grants (see 3.14). The server decrypts the content needed for a call in memory, processes it and returns the result to your AI, where your AI provider processes it (3.7). This is not end-to-end encryption against the operator, nor a promise that the operator is technically unable to read content during processing. Technical information such as entry type and date, account/billing data, consent records (3.11) and separately authorised plausibility reports (3.10) are not encrypted with your diary key.
- Purpose and legal basis: Providing the service (Article 6(1)(b) GDPR). We treat body weight, body measurements and health information as health data; we process them only with your explicit consent (Article 9(2)(a) GDPR), given once in the consent step at the start (consent text version ‘health-v3-2026-11’). Without it, Metaflex does not store weight, calculate weight trends or calculate personal calorie targets. This consent does not cover plausibility reports under 3.10 (which require separate consent). You can withdraw it at any time in the chat; we then delete your entire weight history, health information, body measurements, body fat and waist values (including in your profile), photo descriptions, health-related notes and your plausibility reports (3.10). Processing carried out before withdrawal remains lawful.
- Retention: Until you delete entries or your account; after paid access ends, we automatically delete diary and coaching data, including weekly plans and related reminder settings, after three calendar months unless new access begins beforehand. Your account and necessary contractual records are retained. About two weeks before deletion, we send a reminder email stating the date and explaining that a new subscription before deletion preserves the data. If the email is sent later, deletion takes place no earlier than 14 days after confirmed sending; it is suspended if the contract association or sending status is unclear. You can export your data in your account beforehand. Short-lived chart links expire after about 15 minutes.
Without valid health consent, the AI tools and account export do not disclose values classified as health data, including residual older entries. Mixed diary and profile records are filtered; food, training without heart-rate values and general preferences remain available. Short-lived chart links contain the account reference only in encrypted form and recheck health consent on every retrieval. They stop working after withdrawal; older links without a verifiable account reference are also rejected.
When context is retrieved, we check the core inputs for personal energy and macro targets, their field-specific confirmation dates and the age of weight entries. A confirmation date is set only for fields actually provided or confirmed; other profile edits do not renew it. Unknown older field dates are labelled unknown. Missing inputs do not block meal planning, recipes or the diary; personal target numbers may be omitted or labelled as rough estimates or provisional calculations. In the existing encrypted settings, we store which missing fields have already been asked about and the account-local calendar day when an update question was issued. This is intended to limit your AI to one update question per day. We respect an explicit refusal for 30 days; body fat stays optional and, after a stored refusal, is not requested again until you explicitly reopen the topic. These rules cannot guarantee that the AI provider follows them. The technical data and refusals are included in your encrypted data export and deleted with the associated data. When health consent is withdrawn, we remove question history; explicit negative disclosure preferences remain respected so the refusal is not asked about again.
3.2 Account, sign-in and subscription
- Content: Email address, times of account creation, sign-in and key issuance, subscription status, plan, trial and term end, Stripe customer and subscription identifiers, and time of agreement to the Terms and immediate service commencement. If you redeem an invitation code, we also store the code, the type of free access (trial or friends access), redemption time, trial access end and the Terms version applicable at redemption.
- Email verification: For new accounts, we store the verification time and a random verification token only as a hash, its binding to the password generation, validity (24 hours) and sending time. Recovery before verification blocks the previous password access until the mailbox owner sets it up again. Verification links contain no password or connection key.
- Recovery sign-in codes: For code verification, we store the code and email address only as cryptographic verification values (HMAC). We store your web account email separately in plaintext so we can associate sign-in, contract and mandatory notices. Codes are valid for 10 minutes and are deleted no later than one day after expiry. We send the code to your email address through Resend (3.5).
- Session cookie: After sign-in, we set a technically necessary session cookie (
__Host-mf_session, HttpOnly, Secure, 14 days). Legal basis: section 25(2) no. 2 TDDDG, Article 6(1)(b) GDPR. Optional statistics cookies are set only after your separate consent (see 3.22). If you decline, we store only that choice without a visitor identifier. We use Cloudflare Turnstile on the public cancellation and withdrawal forms to prevent abuse (see 3.3); Cloudflare may process technical browser and connection data there. - Purpose and legal basis: Entering into and performing a contract (Article 6(1)(b) GDPR).
- Retention: Account data until you delete your account. Diary and coaching data are deleted after the contract ends as described in 3.1; automatic data deletion does not delete the account, sign-in data, consent records or necessary contractual and billing records. For deletion scheduling, advance warning and evidence, we store the internal account/user identifier, associated contract reference, contract end, planned deletion date, sending attempts, sending confirmation and deletion status (Article 6(1)(b) and (f) GDPR: implementing the agreed retention rule and preventing erroneous deletion). The data export includes these times and status, but no technical execution tokens. Tax-relevant records are subject to statutory retention requirements (see 3.4).
Founding round. Before your account is created, we store your email, a verification value for the confirmation link, the request and expiry times, and the applicable terms version. The link is valid for 24 hours. We remove the pending link after confirmation or expiry. We store your confirmed place, the date access ends and the associated account data. To prevent repeat participation, we generate a verification value using your email and a secret key. We delete this value when the round has closed and there are no valid signup links outstanding. Your eligibility for a founding subscription remains. We use Cloudflare Turnstile to check signups (3.6) and send the confirmation link via the email provider described in 3.5. This is for running the founding round and preventing repeat signups. After free access ends, you can still export or delete your account and entries. Stripe receives data only if you later start a paid signup.
Founding round statistics. With your website statistics consent (3.22), we count form starts and confirmed registrations. Only if you make an additional voluntary choice do we also record whether and when you connected your AI, made a first entry or used the service again in week two. Statistics do not include the contents of your entries. You can withdraw your choice in the statistics settings.
3.3 Cancellations and withdrawals
If you submit a declaration through ‘Cancel contracts here’ or ‘Withdraw from contract’, we store your name, email address, optional contract number, type and any reason, and receipt time as evidence (Article 6(1)(c) GDPR in conjunction with sections 312k and 356a BGB; Article 6(1)(f) GDPR for defending claims). The declaration is stored without sign-in or an additional email approval click. For automatic processing, we compare the name, email and optional contract details with contact details held by Stripe. We send confirmation only to the verified matching Stripe contract address, not to a different address entered in the form. We do not change the sign-in address. The public receipt page neutrally confirms only the declaration you entered; cases that cannot be clearly matched are handled manually. To prevent abuse, we use Cloudflare Turnstile (provider in 3.6) and request limiting under a transient cryptographic verification value. We send the challenge proof to server-side Turnstile verification, not your name, email or health data. Missing or failed proof does not prevent the declaration from being stored, but leads to manual processing. After verified cancellation, the confirmation includes a signed, single-use cancellation-retraction link valid until the recorded period end. Opening the page alone changes nothing; only the confirmation button requests retraction. We store technical random and verification values, times and execution status for this purpose (Article 6(1)(b), (c) and (f) GDPR). Retention: three years from the end of the year of receipt (standard limitation period); entries are then automatically deleted.
Optional satisfaction rating: In the Metaflex cancellation form you may indicate your satisfaction without free text (five levels from ‘very dissatisfied’ to ‘very satisfied’). Answering is optional; cancellation remains possible without an answer or if a technical error occurs. We store the rating with the cancellation case, separately from the legal reason for extraordinary cancellation. It is linked to your contact details through the case and is not anonymous. The purpose and legitimate interest are to improve Metaflex (Article 6(1)(f) GDPR). Only the operator has access; the rating is included in your data access response. You may object to its use or request earlier deletion. We automatically delete the rating 90 days after receipt, independently of the longer retention of cancellation evidence. It may remain in backups for up to 30 further days (3.6).
3.4 Payments through Stripe (Managed Payments)
Stripe handles payments through Stripe Managed Payments. Through its Link service, Stripe acts as reseller (Merchant of Record) and handles payment processing, invoicing, VAT calculation and remittance, fraud prevention and payment support. According to Stripe, the contracting party for customers in the EEA is Stripe Payments Europe, Limited, One Wilton Park, Wilton Place, Dublin 2, D02 FX04, Ireland, where applicable together with affiliated Stripe companies (such as Stripe Technology Europe, Limited) (‘Stripe’).
- What Stripe processes: The information you enter on the payment page (such as email, name, billing address or country, payment details), and transaction and device data. You enter these directly with Stripe; we do not receive your full payment details. Stripe processes them as an independent controller (payments, invoices, taxes, fraud prevention, support and its own statutory obligations). Stripe or Link sends receipts, invoices and subscription-related emails (such as trial confirmation and a reminder before it ends) directly to you.
- What we receive from Stripe: Customer and subscription identifiers, plan, subscription and payment status, term and trial end, and the email address used for payment.
- Legal basis: Article 6(1)(b) GDPR (entering into and performing the subscription) and (c) GDPR (tax and commercial law obligations).
- Third countries: Stripe bases transfers to the USA on the EU–U.S. Data Privacy Framework or standard contractual clauses.
- Retention: We retain billing records (such as Stripe payout and fee reports) for up to 10 years in accordance with tax law requirements (in particular section 147 AO). Records issued to you by Stripe are subject to Stripe’s retention periods.
- Stripe privacy notice: https://stripe.com/de/privacy and Link’s: https://link.com/privacy
3.5 Sending email
We send sign-in codes, confirmations and mandatory notices through Resend (Plus Five Five, Inc., USA) as processor. The email address and content of the respective email are transmitted, as is the reply address for contact enquiries. We do not send marketing emails or diary contents through this service. We do not promise exclusively European processing by Resend: its data processing agreement describes processing in the USA as well and standard contractual clauses for such transfers. Information: https://resend.com/legal/dpa and https://resend.com/legal/privacy-policy.
3.6 Hosting, database, backups and logs (Cloudflare)
We run the website, interface and database with Cloudflare, Inc., 101 Townsend St., San Francisco, CA 94107, USA (Cloudflare Workers, D1, R2) as processor. The D1 database is set to Cloudflare’s EU jurisdiction; its stored data are located in the EU. This setting does not automatically restrict where all Worker calls, logs or other services execute. The R2 backup store has a Western Europe (WEUR) location hint but no separately configured EU jurisdiction; we therefore do not claim technically enforced exclusive EU storage for it. Cloudflare necessarily processes your IP address when you visit; we use it only transiently to limit requests (abuse prevention) and do not store it in our database. Error and operational logs contain no contents, keys, email addresses or IP addresses and are deleted after no more than seven days. To ensure no cancellation, withdrawal, deletion or withdrawal of consent is lost after a database restore, we additionally log these events there using internal identifiers (case, account or pseudonymous user identifier, Stripe subscription ID), without names or email addresses (Article 6(1)(c) and (f) GDPR); these entries are also deleted after no more than seven days. We store daily database backups in a non-public Cloudflare R2 store and automatically delete them after 30 days; diary contents remain encrypted with your key in these backups. Deleted diary and regular account data are therefore also removed from backups within 30 days at the latest. The limited founder-price allocation records under 3.19 are exempt from this. Legal basis: Article 6(1)(b) and (f) GDPR (secure, stable operation). Transfers to the USA: EU–U.S. Data Privacy Framework or standard contractual clauses.
3.7 Your AI providers
When you use Metaflex in Claude (Anthropic), ChatGPT (OpenAI) or Grok (xAI), the respective provider processes your chat contents under its own privacy policy and as an independent controller. Your AI calls Metaflex; Metaflex’s responses — including decrypted diary contents where needed for the request — go to your AI and thus its provider, which processes them under its own terms. We cannot influence this or promise confidentiality towards your AI provider. Please review their privacy notices and settings (such as using chats for training).
3.8 Usage measurement
For cost and capacity planning, we count calls and database operations per month under a pseudonymous identifier (derived from your key). After your account is deleted, this identifier can no longer be associated with a person. Legal basis: Article 6(1)(f) GDPR.
3.9 Contact form and email contact
If you write to us through the contact form (/kontakt) or by email, we process your name, email address and message to respond to your enquiry. The contact form does not store the message in our database; it forwards it by email to our mailbox through our email provider Resend (see 3.5), with your email set as the reply address. To prevent abuse, we limit requests per IP address; the IP address is used only transiently and is not stored. Legal basis: Article 6(1)(b) GDPR (questions about contracts or subscriptions), otherwise Article 6(1)(f) GDPR (responding to enquiries). Our support@metaflex-rhythmus.de mailbox is hosted by Google (Gmail); emails to that address and contact form messages are forwarded there and processed there. Incoming emails are first received by Cloudflare, Inc. (see 3.6) as processor through Email Routing and forwarded to this mailbox without storage; the information in 3.6 applies to transfers to the USA. The recipient in this context is Google Ireland Limited, Gordon House, Barrow Street, Dublin 4, Ireland, as processor. Transfer to Google LLC in the USA cannot be ruled out; the basis is the EU–U.S. Data Privacy Framework, under which Google LLC is certified, supplemented by standard contractual clauses. The legal basis for processing in this mailbox is the same as for your enquiry (Article 6(1)(b) or (f) GDPR; our legitimate interest is reliable, secure email handling). Google privacy notice: https://policies.google.com/privacy
We delete messages once your enquiry has been resolved, unless statutory retention duties apply. Please do not send health data through the form and never send your connection key.
3.10 Plausibility reports from your AI
If your AI finds a number calculated by Metaflex implausible (such as a daily target that does not match your entries and weight history), it can report this to us — only if you have separately consented. This is a separate item to be selected independently in the initial consent step (consent text version ‘diag-v2-2026-11’); health consent under 3.1 does not cover plausibility reports. Without your consent the server accepts no reports. We store only: time, affected area (such as ‘daily target’), numeric values from a fixed list of Metaflex metrics (such as calorie target, average logged calories, weight trend in kg per week — no body weight, body measurements or health information), one to three reason codes and program version — no free text. Reports are not encrypted with your key, so we can find and fix calculation errors. They carry your pseudonymous identifier (see 3.8), not your name or email; however, we can associate them with your account through this identifier, so they are not anonymous. Only the Metaflex team can view them on an internal admin page; we receive a notification without contents when new reports arrive. Purpose: ensuring your values are calculated correctly. Legal basis: your consent (Article 6(1)(a) GDPR and, insofar as the numbers allow conclusions about your health, Article 9(2)(a) GDPR). Consent is voluntary and not required to use the service. You may withdraw it at any time in the chat; we then delete your reports immediately. Retention: no more than 90 days; reports are deleted immediately on withdrawal or deletion of your account or all data.
3.11 Consents and consent records
At the start, your AI asks once in a single message for your consents — each as a separate yes/no item: basic consent (required for Metaflex), health data (3.1), error reports (3.10) and coach memory (3.12). We do not ask again afterwards; only a material change to a consent text causes that item to be presented again. For each grant and withdrawal, we store the type of consent, text version, time and source (such as ‘consent step’, ‘withdrawal’) under your pseudonymous identifier as evidence (Article 7(1) GDPR; Article 6(1)(c) GDPR). These records are stored separately from the diary and cannot be removed by deleting individual entries or notes; they are part of your data export. You can withdraw consent at any time in the chat; functions requiring withdrawn consent are then blocked until you consent again. Without basic consent, you can only export or delete your data and cancel your subscription. Retention: until account deletion.
3.12 Coach memory (AI-derived notes)
With your consent, your AI may record its own observations about you (such as ‘hunger is strongest in the evening’). It tells you each time: ‘I’ll remember: …’. Such notes are marked as AI-derived, encrypted like your diary (3.1), viewable in the chat at any time and individually deletable; confirmation of each individual note is not required. They never authorise a plan change without you. Legal basis: your consent (Article 6(1)(a) GDPR, and additionally Article 9(2)(a) GDPR for health-related notes). On withdrawal we delete all AI-derived notes.
Metaflex does not read your AI’s chat history, memory or conversation summaries. Coach memory contains only notes that your AI explicitly sends to Metaflex with your consent; you see each note and can delete it.
Long-term coaching memory also includes the following, likewise encrypted like your diary, readable only with your key, part of your data export and deleted with your account: the result of each completed or adopted plan change (lever, period, targets before/after, weight trend, adherence, assessment ‘helped / did not help / unclear’ — as an observation, not proof of causation), a record of the safety check for each change proposal (accepted, reduced or rejected with reason), your daily targets for each day and small observational experiments that never change targets. Because results contain weight trends, we delete them on withdrawal of health consent (3.11). Legal basis as in 3.1/3.11.
For traceable evaluation, we also store the days, measurements and targets valid at the time as an encrypted evidence snapshot. A later diary correction does not retrospectively change this historical evidence. These records, AI proposals not yet applied and permanently adopted target adjustments can also be exported and individually deleted; on withdrawal of health consent we remove the associated health data and evidence copies. A proposal alone does not change your targets. The standard coach extract contains no health notes; your AI retrieves additional details specifically. To switch AI applications, you can export notes and past coaching results and import them into another Metaflex account; consents and ongoing plans are not transferred.
3.13 Usage counters, sending logs, checkout transactions
For cost control and abuse prevention, we count how often your AI calls Metaflex per month (number of calls and database rows, no contents) under your pseudonymous identifier (Article 6(1)(b) and (f) GDPR); deleted after the current and 12 previous months. To prevent abuse of our confirmation emails, we log a cryptographic verification value of each recipient address (not the address itself) and time; deleted after 30 days. We delete details of a started but unfinished checkout (Stripe session) 30 days after its expiry; the checkout log (plan, status, times, Stripe session ID) 400 days after completion. For a cancellation or withdrawal, we also store which contract it concerns (Stripe subscription ID, contract start, plan, current billing period end) and execution status at Stripe (Article 6(1)(b) and (c) GDPR). Refunds are reviewed and processed only manually in the Stripe Dashboard; a public declaration does not trigger an automatic refund. Open review tasks remain until processed. We retain technical execution jobs for 400 days after receipt unless an open review task or current contract binding requires longer storage. Retraction verification values are removed on expiry or account deletion. Stripe manages booking records subject to statutory retention; older completed refund evidence remains under existing retention rules. All these data form part of your data export.
To diagnose storage and retrieval failures, we keep a technical invocation log: timestamp, pseudonymous connection-account ID, tool name, counts of confirmed, rejected and unconfirmed entries, and a fixed error code. It contains no foods, training or health values, entry identifiers, free text, email addresses, passwords or connection keys. The reference is pseudonymous, not anonymous; only the operator can access it for troubleshooting and reliable operation (Article 6(1)(f) GDPR). This necessary operational log is separate from optional plausibility reports (3.10), statistics and advertising. Automated deletion runs remove it after 30 days; it is also deleted when the account is deleted. Retained records are included in your data access export. Backups may contain them for up to a further 30 days (3.6). Inside encrypted diary entries, we also store a technical fingerprint of the write request to distinguish identical retries from conflicting entries; it is deleted with the corresponding entry.
3.14 Sign-in for AI apps (app password, OAuth)
Instead of providing your key in the chat, you can connect Metaflex in Claude, ChatGPT or Grok through /app/mcp with sign-in. At registration you set a Metaflex password that also works on the website. Existing app passwords remain valid. We do not store the password: we store a verification value derived from it (PBKDF2) and your key, encrypted with a second key also derived from the password (AES-256-GCM), along with setup time and the number of failed sign-in attempts (blocked for 15 minutes after 10 failures). When you allow an AI app access, we store the app registration (name, redirect address), grant (account identifier, time, ‘metaflex’ permission) and access tokens only as verification values (hash); the grant contains your key only in encrypted form, decryptable only with the respective token. Access tokens last one hour; refresh tokens last at most one year and expire after 60 days without use. Purpose and legal basis: providing sign-in (Article 6(1)(b) GDPR). Website sessions are bound to the password generation used. Changing your password through password settings revokes previous website sessions and AI grants; the data key is encrypted with the new password without deleting diary data. An explicitly confirmed fresh start replaces the key and deletes the old data area. Account deletion also removes stored password credentials, verification data and sessions. On refresh we additionally store an encrypted technical retry record: an identical, still authenticated retry within 60 seconds can receive the same token pair; later reuse is rejected. These records expire with the associated grant. Expired entries are automatically removed.
3.15 Overview: data from Metaflex tools
Your AI uses Metaflex through tools, for example to save entries, retrieve the daily overview, plan or coach memory, export and delete.
- What: Only what your AI passes to the tool (3.1, 3.10, 3.11, 3.12) — no full chat histories, contact details or location data. Metaflex does not forward your diary contents to any other services.
- What for: Storing your diary, calculating targets and evaluations, coaching guidance. No advertising, sale or AI model training with your data.
- Retention: Diary, profile and coach notes until you delete them or your account; retention periods for other data are in 3.1–3.14; backups no more than 30 days (3.6).
- Viewing and deleting: ‘What the coach knows about me’ in your account or in the chat (tool
coach_memory_overview); delete all coach notes in your account or in the chat (coach_memory_delete_all); individual entries or all data in the chat (delete_entries); export in your account or in the chat (export_data); delete the account in your account settings. - Contact: support@metaflex-rhythmus.de or /support.
3.16 Minimum age
Metaflex is for people aged 18 and over. We do not knowingly collect minors’ data. If your birth year indicates an age below 18, Metaflex calculates no weight-loss targets. If you have reason to believe an account belongs to a minor, write to support@metaflex-rhythmus.de; we will delete it.
3.17 Older routines with storage permissions
For older routines managed through set_settings with explicitly agreed storage permissions, Metaflex records the selected times, AI app and your storage decision in encrypted form. Before setup, the AI names the topics (for example meals, training or steps) and asks whether it may save them directly or must ask first. Creating such an agreed routine constitutes ongoing consent to save exactly these topics in the selected mode. Health data and AI-derived notes additionally require the respective separate consents under 3.1/3.12. A routine cannot replace these consents or change targets without you. New reminder requests under 3.20 grant no such storage permission.
Deleting or disabling the routine withdraws its storage consent for the future; you can delete previously stored entries separately. The routine itself is in your AI app. Metaflex cannot technically detect deletion there without notification. Therefore ask your AI to remove the routine from Metaflex too; if it cannot do so, you can withdraw storage consent in the chat. Metaflex does not schedule its own background calls. Without suitable scheduling and connector functionality in the AI app, no automatic routine is created. A refusal to set up routines is not asked about again until you request routines yourself.
3.18 Food databases
Food searches by name use our local copy of the German Nutrient Database. Source: Max Rubner-Institut (2025): Bundeslebensmittelschlüssel (BLS), version 4.0 — German Nutrient Database. Karlsruhe. DOI: 10.25826/Data20251217-134202-0, licence CC BY 4.0. The data are prepared for searching; original values including source information and references are retained.
For barcode searches, we send only the barcode and fixed technical retrieval fields to Open Food Facts, with a generic Metaflex identifier in the User-Agent. No account, chat, health or diary data are transmitted. Purpose: the product lookup you request (Article 6(1)(b) GDPR). As with any server request, the provider receives our server’s technical IP address. Public product data are cached separately from BLS and diary data; regular lookups are refreshed after seven days, missing products after one hour. Public cache data contain no user association. At higher user numbers, product data are provided from a separately held Germany export.
Product data © Open Food Facts contributors, Open Database License (ODbL). Nutritional values carry a source label (BLS, OFF, user or AI estimate); when saving, Metaflex adopts the source supplied by your AI without independently certifying its accuracy. Older entries without a source remain recognisable as unknown origin. Database information can be incomplete or wrong. If no matching result exists, the AI may estimate, but must clearly label those values as estimates.
3.19 Limited founder-price allocation
When we offer a founder price, we count the first 500 allocated places on the server. Started checkouts and trials initially reserve a place; a verified active founder subscription or one demonstrably paid for counts permanently, even after cancellation. We store place number, reservation/allocation status, times and technical account, checkout, Stripe customer, subscription and price identifiers for this purpose. An unclear payment transaction does not automatically release a place. The website publishes only aggregate place counts, no customer list.
To prevent the same founder price being allocated again after cancellation and account recreation, we store an HMAC verification value of the normalised email address generated with a server-side secret, together with allocation/exclusion status. The value is pseudonymous, not anonymous; it contains no diary or health data and no additional plaintext email address. A permanently allocated place and this limited allocation evidence are retained after account deletion for as long as the limit and exclusion from repeat participation must be enforced. The purpose is to implement the agreed pricing conditions and protect limited allocation against repeated use (Article 6(1)(b) and (f) GDPR). Account-related allocation information is part of the data export. You may send questions or an objection to the contact address in section 1.
3.20 Optional reminders and Web Push
If you want reminders, we store the selected neutral labels, times, weekdays, time zone (Europe/Berlin by default), channel, active status and the technical task identifier for ChatGPT/Grok tasks. These settings are operational data in D1 on servers in the EU; they are not encrypted with your personal diary key. A ‘no’ is saved and reminders are not offered again without a new request from you. Reminders grant no additional permission to store diary or health data. The separate previously agreed storage permissions for older routines under 3.17 remain governed separately.
For Claude you can explicitly allow Web Push on the reminders page. Your browser uses its push service for this (such as Google, Apple, Mozilla or Microsoft; processing outside the EU is possible). Metaflex sends only an empty push message with no diary, target, training or health contents. Your device generates the neutral notification itself; tapping it opens claude.ai. The push endpoint is stored encrypted on the server; Metaflex can decrypt it for sending. A reminder link lasts no more than 15 minutes and permits only reminder management, not account or diary access. The legal basis for optional push is your consent (Article 6(1)(a) GDPR; additionally section 25(1) TDDDG for device access).
For the routine check, we store the time of the last joint review, technical reasons for changes and ignored reminders explicitly reported by you. Missing diary entries do not count as ignored reminders. Preferences and review status remain until account/data deletion; you can switch reminders off at any time. Technical delivery records contain date, reminder identifier and acceptance/error status, no contents, and are cleared after 30 days. Inactive push devices are cleared after 30 days; expired access links at the next daily cleanup. The export contains settings and delivery metadata, but no usable access links or push endpoints. Acceptance by a push service does not prove receipt or that a notification was read. Email reminders are off by default; Telegram is not currently configured. Your AI manages ChatGPT/Grok reminders in the respective app; Metaflex cannot automatically detect a task deleted there without notification. Switching reminders off only in Metaflex does not delete external tasks.
For the silent routine check, your AI receives the existing routine, target and plan context when it calls Metaflex. It should not mention this in conversation without a specific reason. If you reject a particular suggestion, we store only a verification value of its neutral category or affected routine and times, so the same suggestion is not offered again for at least seven days. Rephrasing does not reset this block. These technical suggestion records are cleared after 30 days, included in the export and deleted with user data. They contain no additional health or conversation contents.
3.21 Optional ad measurement for ChatGPT ads
If you arrive at /start through a ChatGPT ad, you may voluntarily consent during registration, independently of the contract, to us reporting your successful account creation or your Stripe-confirmed trial start to OpenAI. You can give these consents separately. We only use the ad identifier and send an event for the action you have agreed to share. We use neither a browser pixel nor advertising cookies for this. The identifier from the ad click initially stays only in the displayed form; only with your consent do we associate it with your newly created account. The legal basis is Article 6(1)(a) GDPR.
We send OpenAI the unchanged ad identifier (oppref), a random event ID, the time, the event type registration_completed for account creation or trial_started for trial start, and the fixed page address https://metaflex-rhythmus.de/start. We do not send email addresses, visitors’ IP addresses, passwords, Stripe IDs, form inputs, diary, goal or health data. This measurement is pseudonymous, not anonymous. It serves to attribute and evaluate our advertising. Under its Conversion Terms, OpenAI also processes the data to provide and improve its advertising services; we apply the available opt-out from future user-based personalisation (opt_out). This does not replace your consent.
The recipient for EEA data is OpenAI Ireland Limited, 1st Floor, The Liffey Trust Centre, 117–126 Sheriff Street Upper, Dublin 1, D01 YC43, Ireland. OpenAI and Metaflex generally act as independent controllers for this purpose. Further processing outside the EEA is possible; in the Ad Tools DPA, OpenAI commits to a valid transfer mechanism for such onward transfers. We do not promise exclusively European processing. Information: https://openai.com/policies/conversion-terms/ and https://openai.com/policies/ad-tools-dpa/ sowie https://openai.com/policies/privacy-policy/.
Ad attribution lasts no more than 30 days. We remove the ad identifier for the respective event after successful delivery, withdrawal of consent or expiry. Technical consent and delivery records contain the time, text version, event ID and delivery status; they are deleted within 90 days. They may remain in backups for up to a further 30 days (3.6). The records are included in your account data export and are deleted with your account.
You can withdraw your OpenAI ad measurement consents at any time under Werbemessung verwalten after signing in. This stops further transmissions; a transmission already in progress cannot reliably be recalled. Withdrawing consent does not automatically delete data already sent to OpenAI. Processing carried out before withdrawal remains lawful. Your rights under section 5 and your use of Metaflex are unaffected.
3.22 Optional website and funnel statistics
On the public pages /, /start, /setup and /support, we offer optional statistics. Only after consent do we load Google Analytics 4 and, on suitable public pages, Cloudflare Web Analytics. Without consent, we do not load a Google statistics tag or send statistics events to Google. Account, sign-in, cancellation, withdrawal and diary pages do not receive a Google statistics tag. This choice is independent of OpenAI ad measurement under 3.21. The legal basis is your consent (Article 6(1)(a) GDPR and, for device access, section 25(1) TDDDG).
We store a random, signed visitor identifier in the cookie __Host-mf_statistics (HttpOnly, Secure, SameSite=Lax, up to 30 days). Google may set statistics cookies (_ga and _ga_…) with a configured lifetime of 30 days. If you decline, our cookie contains only “off”, no visitor identifier, and lasts up to one year. You can change your choice at any time under Statistics settings . Withdrawing consent removes the Google identifier from our visitor attribution and stops future transmissions. Transmissions already in progress cannot reliably be recalled. Withdrawing consent does not automatically delete data already sent to Google; for access or deletion enquiries, use the contact in section 1. Earlier processing remains lawful.
Statistics record the fixed public page address without query parameters, a random event ID, event and session times, a restricted selection of source channels and technically restricted campaign identifiers. The operator can choose campaign names freely; they are normalised to no more than 80 letters, digits, underscores and hyphens. We do not collect URL, search or referrer text, or identifiers containing email addresses or URL/query components. Events include page views, CTA clicks, form starts, product video use and reaching the FAQ. No form content, email addresses, passwords, payment identifiers, chat, diary or health content is transmitted; there are no session recordings. The Google browser tag may process technically required browser/device information and your IP address when establishing the connection. We do not promise anonymous or exclusively European processing.
If you create a new account after consenting, we link the visitor identifier internally to that account to count registration, creation of a Stripe checkout, a trial confirmed by the current live Stripe contract, the first payment confirmed by a positive live Stripe payment record, and the first successful authenticated MCP contact. Payment checks cover monthly subscriptions and annual passes; free invoices and test-mode payments do not count. We do not send payment identifiers or amounts to Google. The extended statistics use consent text statistics-2026-11-11-r2; first payments are not recorded retrospectively on the basis of earlier statistics consent. This link is not sent to Google as an account, customer or user identifier. We send Google only the random statistics identifier, neutral events and restricted source information. MCP contact does not establish any particular quality of subject-specific answers. Existing accounts and manually granted test access are not retrospectively counted as new trial subscriptions.
The recipient is Google Ireland Limited, Gordon House, Barrow Street, Dublin 4, Ireland. We use Google Analytics to evaluate our website’s reach, sources and use, and the steps leading to a trial and first payment. Google Signals, ad personalisation and optional sharing of Analytics data with other Google products remain disabled. Information: https://policies.google.com/privacy and https://business.safety.google/adsprocessorterms/ . Data may also be processed outside the EEA; the agreed data protection and transfer terms apply.
We also use Cloudflare Web Analytics for loading times, page views and technical errors on the same public pages after statistics consent. The provider and international processing are described in 3.6. The beacon is embedded manually, not automatically on account pages. Because the Cloudflare script determines the browser address itself, we load it only for visits without query parameters or fragments, and without those components in the referrer. This beacon sets no cookies or persistent user identifiers of its own. Website statistics available through Cloudflare are independent of our own conversion count. Information: https://developers.cloudflare.com/web-analytics/faq/ .
Our visitor attribution lasts no more than 30 days. Our technical event and consent records are deleted within 90 days of consent, or with the account when it is deleted. Account-linked records are included in the account export. Failed transmissions are retried for no more than 71 hours. GA4 retention for user- and event-level data is limited to two months; aggregated Google reports may remain longer. Server-side reporting counts only consented visitor identifiers and is neither a complete count of all visitors nor a reliable count of distinct people. Backups are governed by section 3.6.
3.23 Optional Meta ad measurement (Facebook and Instagram)
On our public German and English pages, we offer a separate “Marketing (Meta)” category. Only after your voluntary consent do we load the Meta Pixel and send events through the Conversions API. Metaflex remains fully usable without consent. This choice is independent of Google statistics (3.22) and OpenAI ad measurement (3.21). The legal basis is your consent under Article 6(1)(a) GDPR and, for device access, section 25(1) TDDDG. The current notice version is marketing-meta-2026-11-11-r1.
The purpose is to attribute visits, successful account creation, active trials confirmed by Stripe and the first actual payment after a trial to our Facebook/Instagram ads and improve ad delivery. An opened checkout and a free invoice do not count as a trial start or payment. We use PageView, Lead, StartTrial and Purchase events. Browser and server use the same random event ID for the same conversion so that Meta can deduplicate it. Retries are limited to 47 hours; browser mirroring of conversion events takes place only when you return to a suitable public page within 24 hours.
We transmit the event name, time, random event ID, a fixed public page address without query parameters, and browser/ad identifiers (_fbp and, where available, _fbc from an actual ad click). For the first actual payment, we transmit the actual amount paid and the currency EUR. After successful account creation, this consent may also be used to send your normalised email address to Meta as a SHA-256 hash for matching. Hashing is not anonymisation. The Conversions API receives no email address in cleartext, visitor IP address, account or Stripe IDs, passwords, form values, chat, diary, goal or health content. When connecting, the browser Pixel may collect your IP address and technical browser information; Meta may link this information to an existing Meta account. Account, OAuth, password and token pages do not receive a Pixel. We also skip it if URL parameters are not allowlisted or the referrer is sensitive.
We store your consent with a random signed identifier in the __Host-mf_marketing cookie (HttpOnly, Secure, SameSite=Lax, up to 30 days). After consent, we initially set the Meta identifiers _fbp and _fbc for 30 days. Under Meta’s Cookie Policy, the subsequently loaded Meta Pixel may set or renew _fbp, in particular, with a 90-day lifetime. Our server-side ad attribution and consent remain limited to 30 days; on withdrawal or your next public page visit after expiry, we remove the Meta cookies that we can access. If you decline, our cookie contains only “off” and lasts for up to one year. When you create a new account, we internally link the consent record to that account so that confirmed conversions can be attributed and consent withdrawn. We do not send this internal account ID to Meta. Before consent, we do not store Meta identifiers; an existing ad click identifier may be carried in links during public navigation.
Empfänger ist Meta Platforms Ireland Limited, Block J, Serpentine Avenue, Dublin 4, Irland. Eine Weiterverarbeitung auch außerhalb des EWR ist möglich; eine ausschließlich europäische Verarbeitung sagen wir nicht zu. Meta verarbeitet die Daten nach seinen Business-Tools-Bedingungen und Datenschutzhinweisen, auch zur Personalisierung und Verbesserung seiner Produkte und Werbung. Für die Erhebung und Übermittlung von Ereignisdaten sind Metaflex und Meta Ireland gemeinsam verantwortlich; die anschließende Verarbeitung durch Meta gehört nicht zu dieser gemeinsamen Verarbeitung. Die Vereinbarung über gemeinsame Verantwortlichkeit weist uns insbesondere die Nutzerinformation und korrekte Einbindung zu; Meta ist für die Erfüllung der Rechte nach Art. 15–20 DSGVO hinsichtlich der nach der Übermittlung bei Meta gespeicherten Daten zuständig. Du kannst deine Rechte auch direkt gegenüber Meta ausüben. Hinweise zu Rechtsgrundlagen, Zuständigkeiten, Datenübermittlungen und deinen Rechten: https://www.facebook.com/legal/terms/businesstools , https://www.facebook.com/legal/controller_addendum and https://www.facebook.com/privacy/policy/ .
You can withdraw your consent at any time under Marketing settings to stop future transmissions. We discard pending jobs and remove the email hash, browser/ad identifiers and browser information from our attribution record. The same happens after 30 days. Transmissions already in progress cannot reliably be recalled; withdrawal does not automatically delete data already sent to Meta. The lawfulness of processing before withdrawal remains unaffected. We delete technical consent and event records no later than 90 days after consent; records linked to an account are deleted together with that account. They are included in your account export. Backups are subject to section 3.6. For access, deletion and exercising your rights, see section 5 and our contact in section 1.
3.24 Personal coaching wishes and food-photo questions
You can store, edit and delete up to five personal wishes, each with at most 280 characters. These may include whether your AI should ask about materially unclear food photos or log a clearly labelled estimate directly. A photo choice occupies one of the five slots. We store your wishes, their technical identifiers, the photo setting, a revision and the time when the one-time choice question was requested or a choice was saved in the existing encrypted settings. No images are sent to or stored by Metaflex for this purpose. Your AI receives the wishes before the general coach memory when retrieving context; we cannot guarantee that the AI provider follows them. Wishes do not grant consent or change safety rules. The legal basis is Article 6(1)(b) GDPR for personalisation you request. Wishes explicitly marked as health-related require the health consent in 3.1 and are deleted when it is withdrawn. Your explicitly stated wishes are not AI inferences and remain when the optional consent for AI-inferred coach notes is withdrawn. In your account, you can view, edit or delete them after confirming your password; they are included in your encrypted data export and are deleted with the associated diary/coaching data under the periods in 3.1.
3.25 Coaching brief and configurable prompts
For relevant tool calls, Metaflex adds a short coaching brief based on permitted stored data. It can identify missing core inputs, a weight entry at least seven days old, recorded nutrition patterns, differences between measurements and upcoming reviews. Missing inputs do not block meal planning, recipes or the diary; unknown values are not invented. These prompts never proactively ask for body fat. Your AI decides which one or two relevant points to mention; actual compliance and recognition of real conversation boundaries are not guaranteed.
You can enable, pause for a chosen number of days or disable all prompts or individual topics in your account or through your AI. A recorded refusal of profile questions is respected for 30 days. A declined offer to create a real scheduled reminder is not offered again until you explicitly re-enable it. Settings, question history and delivery markers are stored separately, encrypted with your existing data key. Your five personal wishes, consents and external reminder tasks remain independent.
We also record which prompt type was delivered and when, without content, using a verification identifier derived from your already pseudonymous account ID. The log contains no meals, measurements, free text, email addresses or keys and is deleted after 30 days by existing housekeeping. Delivery to the AI does not prove that it displayed the prompt. Encrypted settings follow the diary and coaching data retention rules; negative preferences remain respected after consent withdrawal so a refusal does not trigger another question. Export and account deletion include these data. No new transmission channels or automatic plan changes are introduced.
4. No automated decisions
Metaflex calculates suggestions (such as daily targets). It makes no decisions with legal effects within the meaning of Article 22 GDPR; you initiate plan changes yourself. Your AI’s suggestions are also started only after your consent; Metaflex merely checks them against fixed safety limits and may reduce or reject them.
5. Your rights
You have rights of access (Article 15), rectification (Article 16), erasure (Article 17), restriction (Article 18), data portability (Article 20) and objection to processing based on legitimate interests (Article 21 GDPR), and the right to withdraw consent at any time (Article 7(3) GDPR).
- Export: In your account under ‘Export my data’ (complete with your key, account data without it), or through your AI in the chat.
- Deletion: In your account under ‘Delete account’ (immediately, including ending the subscription), or in the chat. You can delete individual entries in the chat.
- For anything else, write to support@metaflex-rhythmus.de.
You may complain to a data protection supervisory authority, for example the authority responsible for us: Berlin Commissioner for Data Protection and Freedom of Information, Alt-Moabit 59–61, 10555 Berlin.
6. Security
Transmission is encrypted only (HTTPS/HSTS). Contents are encrypted with your key; sign-in codes, sessions and verification values are stored only as hashes. System access is restricted to the operator and protected by strong sign-in methods.
7. Changes
We update this notice when the service or legal situation changes. The version published here applies.
For guided setup we store your chosen usage level, setup status and, separately, your routine preference, question delivery, reported setup and checked status. Without a clear yes or no, the routine question is delivered to the AI at most every seven days on a request. Yes and no remain until you change them. Saying yes does not prove that a task exists or is active. Missing training days temporarily use zero, without saving it as a confirmed profile fact. Restart clears profile values only after your explicit choice; your diary and weight history remain. Technical claims contain the account reference, technical lock identifiers, hint type and time, are purged after 30 days and included in your export. This creates no new external data transfer channel.