Privacy policy

Error monitoring (Sentry)

So that we can find and fix errors in the website and the app, we use Sentry (Functional Software, Inc., USA). If an error occurs in your browser or on our server, an error report is sent to Sentry. It contains technical information: the error message and where in the code it happened, the affected page or endpoint, the time, browser, operating system and device type, and the last steps before the error (such as pages visited, without parameters). Reports from your browser are routed through our own server. Your videos, audio tracks and project content are not part of these reports. Personal details such as email addresses and IP addresses are removed or shortened where possible before a report is stored. Sentry sets no cookies and does not record sessions.

The legal basis is Art. 6 (1) (f) GDPR; our legitimate interest is running the service securely and without errors. Error reports are deleted automatically when Sentry’s retention period ends. We have a data processing agreement with Sentry under Art. 28 GDPR; where data is processed in the USA, transfers are based on the EU-US Data Privacy Framework, where the provider is certified, or the EU Standard Contractual Clauses. Privacy notice: sentry.io/privacy/.

Usage statistics (PostHog)

To see which pages and features are used and where Lumenwerk still falls short, we use PostHog (PostHog Inc., USA) with storage in the EU (Frankfurt). We record page views and a few events: joining the waitlist, finishing sign-up, creating a project, starting a tool (which tool and whether it uses AI) and finishing an export (which format). Technical data is added, such as browser, operating system, device type, language, time zone and the website you came from (its address only, without the path). Page addresses are sent without parameters and identifiers, page titles not at all. Your videos, transcripts, project content, inputs, your name and your email address are never part of the data, and we do not link the statistics to your account.

We run PostHog without cookies and without storing anything in your browser. There are no session recordings, no heatmaps and no automatic click tracking. So that visits on the same day are counted together, PostHog forms a hash of IP address and browser identifier with a key that changes daily; the IP address itself is not stored. The data goes to PostHog through our own server. If your browser sends “Do Not Track” or “Global Privacy Control”, we record nothing.

The legal basis is Art. 6 (1) (f) GDPR; our legitimate interest is improving the service based on how it is actually used. Since nothing beyond delivering the page is stored on or read from your device, we use no cookie banner for this. You can object at any time, for example by switching on “Do Not Track” or “Global Privacy Control” in your browser or by writing to us. We have a data processing agreement with PostHog under Art. 28 GDPR; where data is processed in the USA, transfers are based on the EU-US Data Privacy Framework, where the provider is certified, or the EU Standard Contractual Clauses. Privacy notice: posthog.com/privacy.

Sign-in cookies

We only set cookies that signing in needs. All of them are cryptographically signed and cannot be read by scripts (httpOnly):

• “lw_session” keeps you signed in. It contains your email address, your display name, whether sign-up is complete, and the expiry time. It expires after seven days or is deleted when you sign out.
• “lw_signup” remembers, for at most 15 minutes during sign-in, the email address the code was sent to and your newsletter and language choices.
• “lw_oauth” is only used when you sign in with Google or Apple. For at most 10 minutes it secures the return from that provider (PKCE).

Storing and reading them is permitted without consent under § 25 (2) No. 2 TDDDG because they are strictly necessary for the service you explicitly requested. The subsequent processing is based on Art. 6 (1) (b) GDPR.

Local storage for projects, transcripts and settings

We store your projects (timeline, settings, chats), transcription results and intermediate results of the tools (such as waveforms, image analyses and audio processed with Studio Sound) in the browser database (IndexedDB) on your device, so the same video does not have to be processed again. In local storage we keep interface settings and the fact that you have already accepted the terms and read the privacy policy in this browser. A copy of your projects and transcripts is also kept with your account (see section 8). Local data stays in the browser after you sign out; you can delete it at any time in your browser settings. The legal basis is § 25 (2) No. 2 TDDDG and Art. 6 (1) (b) GDPR.

Sign-in

You sign in without a password: we email you a one-time code. Sign-in and sending the code run through Supabase Auth (Supabase, Inc., see section 8); the email itself is delivered by our email provider Resend (see section 9). If you sign in with Google or Apple, that provider gives us your email address; their privacy notices apply as well. The legal basis is Art. 6 (1) (b) GDPR.

To prevent abuse such as guessing codes, we count sign-in attempts per email address and per IP address. For this we only store a one-way hash of the address with a counter for a short time window. In the same way we count calls to the AI features and requests for your account data (projects, transcripts, styles, credits) per account and per IP address, so a script or a stuck sync cannot overload the service; here too the IP address is only stored as a hash, and the counters only for their time window (at most one hour). The legal basis is Art. 6 (1) (f) GDPR (security of the service).

Beta waitlist

If you join the waitlist on the home page, we store your email address, the language of the page, the time and, if you tell us, whether you make your own videos or edit for others. We only use the address to send you a short confirmation of your entry and later your invitation to the beta with your beta key, never for a newsletter. The confirmation goes to an address at most once every 30 days. To prevent abuse we count sign-ups per IP and email address as we do for sign-in (as a hash only). The legal basis is Art. 6 (1) (b) GDPR (steps taken at your request before entering into a contract).

The entry is deleted if we decline your request, if you delete your account and at the latest when the beta ends. Write to us if you want to be removed earlier (section 2).

Beta forms before and after testing

As a beta tester you fill in one form before testing (lumenwerk.com/forms/beta) and one with your feedback afterwards (through a personal link). We store your beta key, name, email address, language, what you tell us about how and with what you edit, which tools you want to test, your computer and browser and, if you add them, your social media accounts. After testing, your ratings, answers and wishes are added. The feedback form shows you how many credits your key had and how many your account has used since. We use this to evaluate the beta and improve Lumenwerk. The legal basis is Art. 6 (1) (b) GDPR (taking part in the beta).

Only if you tick the box do we mention you and quote your feedback with your name and a link to your social media account, for example on the website or in posts, and contact you for a short call (Art. 6 (1) (a) GDPR). You can withdraw these consents at any time by email (section 2). To prevent abuse we count submissions per IP address (as a hash only). The answers are deleted when you delete your account and at the latest when the beta ends; quotes you approved stay until you withdraw.

Closed beta: beta key, request and feedback

During the closed beta an account can only be created with a beta key. When you redeem a key, we store which account redeemed it and when. If you have no key, you can request one after confirming your email address; we then store your email address, your optional message, the language, the time and our decision.

With the feedback button in the app you can send us bugs, ideas, reviews and anything else. We store:

• the type of feedback, your message and, for a review, the stars (1 to 5)
• the path of the app page you sent the feedback from (without parameters) and the time
• only if the box “Include page and browser info” is ticked: browser and operating system (user agent), window and screen size, language and time zone
• your account, so we can match the feedback to it

Your video and your projects are never part of the feedback. The legal basis is Art. 6 (1) (b) GDPR (access to the beta) and Art. 6 (1) (f) GDPR (our legitimate interest in fixing bugs and improving the product). We delete requests and feedback with your account, and at the latest once we no longer need them for the beta.

AI analysis of feedback

To summarise what all testers tell us, we occasionally have an AI model analyse it (usually Claude by Anthropic, see section 7). The provider of that model receives: the texts and ratings from the feedback button with page path and time, the answers from the beta forms without name, email address and social media accounts, the reasons and notes of refund requests, and totals of which features were used how often. Each item only carries a short label (such as “F3”), no account data. So please do not put anything into free text that identifies you and is not needed. Only we see the result (themes, suggestions, reported bugs). The legal basis is Art. 6 (1) (f) GDPR (our legitimate interest in improving the product based on the beta); you can object under Art. 21 GDPR (section 11).

Profile

When you sign up, we create a profile containing:

• your email address and how you sign in (email code, Google or Apple)
• your username (required) and your first name (optional)
• your language and the version and time of the privacy policy shown to you when you last agreed
• the version of the Terms of Service you accepted, and when
• whether and when you consented to the newsletter
• your answers to the sign-up questions (what content you make, what costs you the most time when editing, how you found us); answering is optional
• whether you have seen the introduction, and your editor settings (layout, shortcuts, level of detail)

The legal basis is Art. 6 (1) (b) GDPR (account and service). We use the optional answers to improve the product (Art. 6 (1) (f) GDPR). You can withdraw your newsletter consent (Art. 6 (1) (a) GDPR) at any time with effect for the future, in your account settings, via the unsubscribe link in every email or by writing to support@lumenwerk.com.

Minimum age

Lumenwerk can be used from the age of 13. At sign-up you confirm with a checkbox that you are at least 13 years old. We do not ask for or store your age or date of birth; sign-up cannot be completed without the checkbox. Minors can only take out a paid plan with their parents' consent (see the Terms of Service).

The newsletter is only available from the age of 16, because in Germany you can only give consent yourself from that age (Art. 8 GDPR). By ticking the newsletter box you therefore also confirm that you are at least 16 years old.

Local processing of your video

Videos you open in the editor are read, played and analysed only locally in your browser. Face detection and the local tools (such as Studio Sound, smooth cuts and blurring) also run locally. The video file is not uploaded to our servers or to third parties. Only the data named in this section leaves your browser, and only when you use the feature in question.

Transcription of the audio track (ElevenLabs)

For features such as Autocut, subtitles, Auto Zoom, AutoScale, Titles & Keywords or Lumi, your browser extracts and compresses the audio track of the video. Only when you start one of these features is this audio track sent to our server and passed on to our provider ElevenLabs Inc. (USA), which turns it into text with timestamps (“Scribe” model). The result is returned to your browser.

The audio track contains spoken language and may therefore contain personal data about you and other people who can be heard in the video. Please only process recordings for which you have the necessary rights and, where required, the consent of the people recorded. We only use the audio to turn it into text, not to identify people or to clone voices.

The legal basis is Art. 6 (1) (b) GDPR (providing the feature you requested). We do not store the audio track on our servers; it is only processed for the duration of the transfer. We have a data processing agreement with ElevenLabs under Art. 28 GDPR. Transfers to the USA are based on the EU-US Data Privacy Framework or the EU Standard Contractual Clauses. How long ElevenLabs retains data and what else it may use it for is governed by the data processing agreement and ElevenLabs’ privacy notice: elevenlabs.io/privacy-policy.

AI features (Anthropic, OpenAI, Google, xAI)

Some features use AI models: the AI review of Autocut cuts, AI zoom planning, AI keyword selection for subtitles, Titles & Keywords, AutoScale with AI, Lumi (chat, reading the video, planning and building), the editor’s AI chat, Lumi Artist (chat, planning, writing and reviewing the animation code), Lumi Stylist and the SFX chat. You choose which model a feature uses in the model menu of each app. On “Auto” we choose the model; currently this is usually Claude by Anthropic. Depending on the model, the provider is:

• Anthropic, PBC (USA) for the Claude models. Privacy notice: www.anthropic.com/legal/privacy
• OpenAI (USA) for the GPT models. Privacy notice: openai.com/policies/privacy-policy/
• Google (Gemini API, USA) for the Gemini models. Privacy notice: policies.google.com/privacy
• xAI (USA) for the Grok models. Privacy notice: x.ai/legal/privacy-policy

Only when you start one of these features does our server send the following data to the provider of the model in use, and only to that provider:

• the transcript text (with timestamps)
• your chat messages and descriptions, for example what a motion graphic should show
• project details as text, such as the timeline layout and file names
• for AutoScale with AI and when Lumi first reads your video: individual still frames from your inserts at low resolution (640 × 360 pixels), up to 40 for AutoScale
• for Lumi Stylist: the reference images you add to the chat (one by one, or combined into one moodboard from four images on), for reference videos on your computer a contact sheet and a short strip of downscaled frames, plus values your browser measures from them (colour palette, cuts per minute); up to six images per message
• for Lumi Stylist, when you paste a YouTube link and confirm “Watch”: the link, the chosen clip and your message, always to Google (Gemini API), whichever model you picked. Google fetches the public video from YouTube itself and analyses a clip of up to three minutes. We neither download nor store the video; what comes back is a description of the style as text
• for the Lumi Artist review: low-resolution frames of the rendered animation, which also show the images you provided for the graphic

The video file itself and the audio track are never sent to these AI providers. The model’s answer (for example cut or zoom suggestions) is returned to your browser. Transcript, chat and images may contain personal data about you and other people who can be heard or seen in the video.

The legal basis is Art. 6 (1) (b) GDPR (providing the feature you requested). We do not store this data on our servers unless it is part of your project (section 8). We use the providers’ application programming interfaces (APIs) for businesses, not their consumer apps. Under the terms of Anthropic, OpenAI and Google for (paid) API use, content sent through the API is not used to train their models; for xAI this is governed by its terms. The providers keep inputs and outputs only for a limited time, among other things to detect abuse, and then delete them. Transfers to the USA are based on the EU-US Data Privacy Framework where the provider in question is certified, and otherwise on the EU Standard Contractual Clauses (Art. 46 (2) (c) GDPR).

Lumi Artist: files, rendering and voice-overs

Lumi Artist turns your description into an animation. To render the finished video we use further service providers:

• Cloudflare R2 (Cloudflare, Inc., USA) stores the images you provide for a motion graphic (for example a logo or screenshots), the spoken lines of the voice-over and the finished render. Your browser uploads the images directly to it and downloads the finished video from it. Everything is deleted automatically after 7 days. Privacy notice: www.cloudflare.com/privacypolicy/
• Trigger.dev runs rendering and the review as background jobs. For this it receives the animation code, the images, the voice-over text and your email address to assign the job. For the review, the job sends frames of the animation to the provider of the chosen AI model (see above). The data of a job is only used to run it; how long Trigger.dev retains job data and logs is governed by the provider’s terms. Privacy notice: trigger.dev/legal/privacy
• ElevenLabs (ElevenLabs Inc., USA) turns the voice-over text into spoken audio (text-to-speech) when your motion graphic has a voice-over. Only this text is sent. Privacy notice: elevenlabs.io/privacy-policy

The legal basis is Art. 6 (1) (b) GDPR (providing the feature you requested). Please only use images and texts for which you have the necessary rights. Where data is processed outside the EU, in particular in the USA, transfers are based on Standard Contractual Clauses or the EU-US Data Privacy Framework, where the provider is certified.

Export

Project files (XML, FCPXML), subtitles (SRT), rendered videos and cut lists are created in your browser and saved directly to your device. The exception is Lumi Artist videos, which are rendered on a server as described above.

Emails from Lumenwerk

We send sign-in codes, the waitlist confirmation, your beta key and the link to the feedback form through Resend (Resend, Inc., USA) from the address noreply@mail.lumenwerk.com. For this Resend receives your email address and the content of the message and stores them in its EU region (Ireland); delivery data such as time and delivery status is kept by Resend for a limited time. We have a data processing agreement with Resend under Art. 28 GDPR; where data reaches the USA, transfers are based on the EU-US Data Privacy Framework or the EU Standard Contractual Clauses. The legal basis is Art. 6 (1) (b) GDPR. Replies to these emails go to our Google Workspace mailbox.