Privacy

Controller

The controller responsible for the processing of personal data within the meaning of the General Data Protection Regulation (GDPR) is Jean-Noel Honisch, Ostenländerstr. 6, 33106 Paderborn, Deutschland, trading as “Second Exodus”. Full details are given in the Legal Notice.

For questions about data protection and to exercise your rights, you can reach us at jean-noel@mnemorya.com.

Mnemorya is a learning app. Much of it runs on your device: your content is first stored in a local database and is only transferred to our servers once you have an account. This policy describes which data goes where in the process, and why.

Account and sign-in

You create an account with an email address and password, or sign in with your Google or Apple account. When you sign in via Google or Apple, we receive your email address and an identifier from them; the respective provider learns that you are signing in to Mnemorya.

Sign-in is handled by our service provider Supabase (see “Service providers we use”). We store passwords only as a hash.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR).

Your content and synchronisation

With an account, your learning content is synchronised between your devices: topics, cards, sources, Memory Palaces, Bildpfade, Atlases, goals, habits, notes, chat histories with Mnemo, learning progress and the files you upload (images, audio, video, PDF). This data is stored on servers of our service provider Supabase in Frankfurt am Main.

Every synchronised row carries a device identifier so that we can tell which of your devices a change came from. It serves synchronisation only and is not used for advertising or profiling.

Without an account, everything stays on your device; no transfer takes place in that case.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR).

Profile, community and online status

Your profile consists of a display name, a handle (@name), optionally a description, profile picture, country of origin and the appearance of your Mnemo. Profile pictures are kept in a publicly accessible storage area; your name, handle, description and picture are also publicly visible on our website at mnemorya.com/@handle. You can set your profile to “private” in the settings and remove it from search.

What you publish — content, comments, likes, star ratings, posts in communities — is visible to other members. Your display name and handle are recorded at the time of posting so that a post remains attributable even if you change your name later.

Anyone who opens your profile while signed in also sees your interests, trophies, communities and your level. We show your personality assessments from your conversations with Mnemo to others only if you explicitly switch this on in the settings; it is off by default.

Online status: we store when you were last active. Whether you are currently online is visible only to people you follow and who follow you; you can switch this off in the settings. A “currently online” count in communities is anonymous.

Following and blocking: we store whom you follow and whom you block in order to filter lists, messages and search accordingly.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR); for voluntary details and toggles, your consent (Art. 6(1)(a) GDPR).

Messages

Direct messages, message requests, attachments and chat within communities are stored on our servers and delivered through them. They are not end-to-end encrypted; we do not read them, but technically we could, for instance to follow up on a report. Ephemeral messages (“Instants”) are deleted automatically once they expire.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR).

Duels and shared learning progress

In a duel, your opponent sees your name, your score, whether you were right on each question, and your response times. A duel remains visible to the opponent after it has ended.

Communities can evaluate their members' learning progress on an Atlas. What is reported is a rough figure — levels completed, chapters, last day of learning — never individual cards or times. In the “aggregated” setting, the people responsible for the community see only a distribution, and only from five participants upwards. An individual view with your name exists only if you explicitly consent to it when joining; you can withdraw that consent at any time — in communities that require it, withdrawal means leaving the community.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR); for the individual view, your consent (Art. 6(1)(a) GDPR).

AI features

Mnemo, the learning companion in the app, content analysis, card creation, image generation and the voice features use language models from OpenAI and — as a fallback — from Anthropic. What is transmitted in each case is the specific request: your input, the conversation history, the content and files you attach, and a short profile briefing (for example learning goals and prior knowledge) so that the answer fits you. Your entire library is not transmitted.

Requests go through our servers, which forward your request using our key. We ask OpenAI not to store the requests; under their terms, OpenAI and Anthropic do not use data from API requests to train their models. Both providers are based in the USA (see “Transfers to third countries”).

In voice mode, once the session has been set up, your microphone audio is streamed directly from your device to OpenAI and processed there in real time.

Your own key: in the settings you can store your own OpenAI key. Text, image and audio requests then go directly from your device to OpenAI, without our servers and without using our quota; voice mode continues to run through us. The app stores the key only in encrypted form on your device.

For the quota, we count per account the time, model, volume (tokens) and cost of each request — not its content.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR); for each feature, you decide yourself whether to trigger it.

Speech and transcription

Voice recordings for Mnemo, dictation and speech output are processed by OpenAI (Whisper or speech synthesis respectively), see above.

Recordings (for example of a lecture) are transcribed by your operating system's speech recognition. Depending on your device, language and installed language packs, Apple or Google may process the audio for this on their own servers; their privacy policies apply. The wake word “Mnemo” is recognised by the app exclusively on the device.

For transcripts of audio and video files, you can load a local Whisper model; it is downloaded once from Hugging Face (Hugging Face sees your IP address in the process), after which the transcription itself does not leave your device. Alternatively, OpenAI transcribes.

We fetch subtitles for YouTube videos through our own service, which uses the provider Supadata; only the video's identifier is transmitted.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR).

Search, maps and media

Web search for Mnemo: search queries go via our servers to Tavily; they may contain text from your conversation. Image and book search uses Pexels, Wikimedia Commons, Google Books and Open Library; the search term is transmitted.

Videos and podcasts: searching for YouTube videos runs via our servers to the YouTube Data API; the number of searches is counted per account. When playing an embedded YouTube video and when loading thumbnails, your device connects directly to Google (YouTube), even without a connected account. The podcast search queries the Apple Podcasts directory and loads the respective podcast feeds directly from their providers.

Maps and Street View for Memory Palaces: when you create or edit a Memory Palace, your device loads map material directly from Google Maps and in doing so transmits the displayed coordinates and your IP address. If you allow location access, your current location is queried once to centre the map; there is no background tracking. Place search uses Nominatim (OpenStreetMap). The coordinates of the places you add to a Memory Palace are stored and synchronised as part of your content — and are published along with it if you share the tour.

During reviews, the app shows reaction images from Giphy; your device loads them directly from there. For link previews, the app fetches the linked page to load its title and image; the operator of that page sees your IP address in the process.

The legal basis is the performance of the user agreement (Art. 6(1)(b) GDPR) or our legitimate interest in a functioning presentation (Art. 6(1)(f) GDPR); you grant location access as consent via the system prompt (Art. 6(1)(a) GDPR).

Connected accounts: YouTube and Spotify

If you connect your YouTube or Spotify account, we receive an access token with which the app can retrieve or control subscriptions, playlists and playback on your behalf. The app stores the tokens in encrypted form on your device; our servers exchange them only when connecting and renewing. You can disconnect at any time in the settings.

The legal basis is your consent (Art. 6(1)(a) GDPR).

Notifications

If you allow notifications, we store a push token for your device together with its platform and language. We send notifications via the Expo service to Apple or Google. A notification contains the sender's name and, for direct messages, a preview of the text. You choose in the app which events reach you; we delete the token when you switch notifications off or delete your account.

The legal basis is your consent via the system prompt (Art. 6(1)(a) GDPR).

Mnemorya Pro and payments

The Pro subscription is processed through the payment service provider Stripe. When you purchase, we create a customer there with your email address and an account identifier; you enter card or bank details exclusively on Stripe's payment page — we neither see nor store them. From Stripe we receive the status of your subscription (active, payment pending, cancelled) and the payment events, which we retain for accounting and evidence.

The legal basis is the performance of the contract (Art. 6(1)(b) GDPR) and our statutory retention obligations (Art. 6(1)(c) GDPR).

Waiting list

If you join the waiting list on the pricing page, we store your email address, your language and the page from which you signed up. Your IP address and browser identifier are deliberately not stored. The confirmation and notification emails are sent by our service provider Resend.

The purpose is exactly two messages: one at launch and one before pricing takes effect. You will not receive any further advertising through it.

The legal basis is your consent (Art. 6(1)(a) GDPR). You can withdraw it at any time — with one click on the unsubscribe link in every message, without an account and without signing in.

After withdrawal, we no longer use your address for any message. We retain it solely as a suppression record so that a later sign-up does not silently override your objection; the basis is our legitimate interest in enforcing exactly this objection (Art. 6(1)(f) GDPR). At the latest when the purpose of the list has been fulfilled, the suppression record is deleted too.

Website

This website does not set any cookies and does not use any analytics or advertising tools. We serve fonts ourselves, not from Google. Your choice between light and dark design is remembered only by your browser.

When you visit, our hosting provider processes, as technically necessary, your IP address, time of access, page requested and browser identifier in server logs; the basis is our legitimate interest in security and operation (Art. 6(1)(f) GDPR).

The website displays public profiles (see “Profile, community and online status”) and offers account deletion and the waiting list.

Service providers we use

Supabase (database, file storage, sign-in; server location Frankfurt am Main) · OpenAI and Anthropic (AI) · Stripe (payments) · Resend (email delivery) · Expo (push delivery to Apple and Google) · Supadata (YouTube subtitles) · Hostinger (website hosting).

With service providers that process data on our behalf, we conclude data processing agreements under Art. 28 GDPR.

Contacted directly from your device — and therefore under the respective provider's own responsibility — are Google (Maps, Street View, YouTube, sign-in), Apple (sign-in, Podcasts directory, speech recognition), Spotify, Giphy, Hugging Face, Nominatim/OpenStreetMap and the operators of linked pages and podcast feeds.

Transfers to third countries

OpenAI, Anthropic, Stripe, Expo, Giphy, Hugging Face as well as Google and Apple are based in the USA. Transfers rely on the EU-US Data Privacy Framework where the provider is certified under it, and otherwise on the standard contractual clauses of the EU Commission (Art. 46(2)(c) GDPR).

Your rights

You have the right of access (Art. 15), to rectification (Art. 16), to erasure (Art. 17), to restriction of processing (Art. 18), to data portability (Art. 20) and to object (Art. 21 GDPR).

You can withdraw any consent you have given at any time with effect for the future. The lawfulness of processing carried out until then remains unaffected.

For all matters, an email to jean-noel@mnemorya.com is sufficient. Independently of this, you have the right to lodge a complaint with a data protection supervisory authority (Art. 77 GDPR); the authority responsible for us is the State Commissioner for Data Protection and Freedom of Information of North Rhine-Westphalia (Landesbeauftragte für Datenschutz und Informationsfreiheit Nordrhein-Westfalen).

Retention period and deletion

We store account and content data for as long as your account exists.

You can request deletion of your account in the app under “Account”; a route via this website is also available in case you no longer have access to the app. You can withdraw the request in the app until processing begins. We then carry out the deletion.

With deletion, your account, profile, content, messages, uploaded files, push tokens and connected accounts are removed. Content you have already published and that other members have downloaded remains with them as their own copy; comments and duels in which others were involved are retained without attribution to you.

We retain payment receipts and accounting records in accordance with tax and commercial law: accounting records for eight years, business correspondence for six years (§ 147 AO, § 257 HGB, German law). After account deletion, they are no longer linked to your account.