Privacy notice
This privacy notice is available in German and in English; both versions describe the same processing.
1 Controller
The controller responsible for processing personal data on this platform is Manuel Dolderer, Schraberg 22a, 58313 Herdecke, Germany, e-mail kontakt@transformeducation.de.
TransformEducation is an independent learning platform. It is not a service of Universität Witten/Herdecke or any other university, even when a lecturer uses it in a course.
Data protection officer: Klaus Dolderer, attorney (Rechtsanwalt), Odenwaldblick 3, 74722 Buchen, Germany (contact by post)
2 What the platform is
The learning platform supports students as they learn in their courses. At its core is Ava, an AI system that answers questions about course content, helps with reading course texts and asks review questions. Ava does not give grades and does not assess performance. Lecturers can also release an AI avatar: Ava with the academic perspective the lecturer has written themselves.
3 What data we process
- Account: username, name, e-mail address, degree programme and semester (both optional), password (stored only as a hash), time of e-mail confirmation, time of the confirmation "at least 18 years old" (no date of birth), courses joined, version, language and time of your acceptance of the Terms of Use. You can register with any e-mail address and a course's invite code.
- Learning: conversations with Ava and with a lecturer's AI avatar, marked passages and questions in the reading tutor, your highlights and notes in the text, reading progress, review cards with your answers and the review schedule calculated from them, notes in the Second Brain, exam dates you enter yourself, feedback and reported answers.
- Lecturers: name, e-mail address, course structure, sessions with notes, learning goals, uploaded texts with a rights declaration, approvals of sources; for lecturers who create an AI avatar, also its text with a version counter and its release status.
- Technical: the cookies in section 8. Neither the web server nor the application writes access logs with IP addresses. Error reports from the application go to the operator by e-mail; they describe the technical sequence and the account ID, but contain no form content, no IP address and no name.
No special categories of data: Do not enter information covered by Art. 9 GDPR about yourself or others on the platform, that is nothing about health, racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, sex life or sexual orientation, and no genetic or biometric data (§ 5 of the Terms of Use). The platform is not meant for this, and we have no legal basis to process such information. If you enter it anyway, it is stored like any other message and sent to the AI provider (section 6), and Ava's answer may repeat it in the nightly check (section 4). In that case write to the address in section 1; we will delete the messages concerned without undue delay.
4 Purposes and legal bases
| Purpose | Data | Legal basis |
|---|---|---|
| Providing the account, courses, Ava, AI avatars, the reading tutor, review and the Second Brain | Account, learning | Art. 6(1)(b) GDPR (contract of use under the Terms of Use) |
| Making a note visible to the course | Name, text of the note, passage | Art. 6(1)(b) GDPR (you share each note yourself and can take it back at any time) |
| Generating Ava's answers (transfer to the AI provider, section 6) | Message, conversation history, course context, matching Second Brain notes; for an AI avatar, the lecturer's text and passages from their course materials instead of the notes | as above |
| Creating an AI avatar and trying it out before release (lecturers) | The avatar's text, the answer in the optional part of the course setup conversation, test questions (not stored) | Art. 6(1)(b) GDPR (Educator Terms) |
| Nightly quality check: an AI model checks Ava's answers for fidelity to sources and rules and codes her teaching moves (paragraph below the table) | Ava's answer and the course sources she received for the answer (cited or not), not your message; data without text about each answer from Ava | Art. 6(1)(f) GDPR; legitimate interest: correct answers that are true to the sources and follow the rules; objection by switch |
| Reviewing reported answers | Copy of the reported answer, your reason and, if you leave the box ticked, your message before it | Art. 6(1)(b) GDPR |
| Aggregated figures for lecturers | Figures for the whole group, only from three students upwards; stored once a week as a history, with no data on individual persons | Art. 6(1)(f) GDPR; legitimate interest: adapting teaching to the group's progress without making individuals visible |
| Account e-mails (confirmation, password, getting started) | E-mail address, name | Art. 6(1)(b) GDPR |
| Answering requests sent through the contact form | Name (if given), e-mail address, message | Art. 6(1)(f) GDPR; legitimate interest: answering requests. Where the request concerns your contract of use, Art. 6(1)(b) GDPR |
| Protecting the contact form against abuse | Check value calculated from the IP address, message counter (one hour at most) | Art. 6(1)(f) GDPR; legitimate interest: protecting the contact form against abuse |
| Contract confirmation with the withdrawal notice by e-mail after each acceptance of the Terms of Use | E-mail address, name, version and language of the acceptance | Art. 6(1)(c) GDPR in conjunction with § 312f(2) BGB |
| Handling and documenting a withdrawal | Account number, time of the withdrawal, of the confirmation of receipt and of the deletion; name and e-mail address only in the confirmation of receipt | Art. 6(1)(c) GDPR in conjunction with §§ 355, 356a BGB |
| Checking whether a text carries an AI marking from the platform (AI marking) | The pasted text, only for the check; it is neither stored nor logged | Art. 6(1)(f) GDPR; legitimate interest: making the origin of a text verifiable |
| Security, abuse prevention, troubleshooting (usage limits per person, error reports, backups) | Account, technical data | Art. 6(1)(f) GDPR; legitimate interest: secure and stable operation |
Contact form: What you send through the contact form (your name if you give it, your e-mail address and your message) goes by e-mail to the operator's mailbox at ALL-INKL.COM (section 6); it is not stored in the platform's database. We use these details only to answer your request and delete them once the request is settled, at the latest six months after we receive it. They are kept longer only while the request is still being handled, a statutory retention obligation applies or we need them for a specific legal claim, and only for as long as that lasts. To limit abusive requests, we process a check value calculated from your IP address with a secret key and a message counter, only in memory and for one hour at most; the IP address itself is not stored. The legal basis for this is Art. 6(1)(f) GDPR; our interest is protecting the contact form against abuse.
Nightly quality check: Each night an AI model, through the AI provider in section 6, checks Ava's new answers, including those of AI avatars. It receives Ava's answer and, for part (a), the course passages she received for the answer, cited or not, never your message. Ava's answer can, however, repeat what you wrote, such as a name, your situation or your question; we therefore treat it as personal data. The check has two parts.
- (a) Sources and rules: The model checks whether the answer matches the cited sources and follows Ava's rules. An answer that its sources do not support loses the label "From course sources". If an answer is flagged, a copy of Ava's answer with the checking model's reasoning, without your question, goes to the operator for review. If Ava's answer reacts to possible distress, the operator also receives an e-mail with this reasoning.
- (b) Teaching moves: The model assigns codes to Ava's answer, such as explaining, asking back or giving a hint. For this the platform stores, without any text, for each answer from Ava: the mode, which message in the conversation it was, how long your preceding message was, how many seconds passed since Ava's previous answer, whether you asked about a marked passage and how many course sources Ava was given. The codes are evaluated in aggregate across all conversations, to improve Ava's rules. To check whether the model codes correctly, the operator may draw single answers from Ava as a sample and read and code them himself; the sample contains only Ava's answer and the mode, not your message and no account; Ava's answer can, however, repeat what you wrote. The files for this are kept encrypted on his device and deleted after coding, at the latest after three months. From the time of an objection, answers to that student are no longer included in any sample.
The check creates no profile of you and no assessment of your performance, and lecturers see none of it. The legal basis is Art. 6(1)(f) GDPR; our legitimate interest is correct answers that are true to the sources and follow the rules, and a tutor that helps you learn instead of handing out solutions. The data under (b) is kept until you object or your account is deleted; flagged answers like reported answers, and the e-mail to the operator like error reports (section 7). You can object to the check at any time (Art. 21 GDPR) with the switch "Object to the nightly check of my Ava answers" under Profile and settings. From then on, Ava's answers to you are no longer checked, the data under (b) is no longer recorded or coded, and the data already stored for it and open flags from the check are deleted. Answers you report yourself are still reviewed.
No research takes place on this platform. We do not use your data for advertising, we do not sell it, and we do not use it to train AI models.
5 Who sees what
- No lecturer and no one in your course can read your conversations with Ava. This also applies to conversations with a lecturer's AI avatar: the lecturer does not see them, not even in summary, and the avatar does not learn from them; it changes only when the lecturer changes their text. Between students, the database separates data per person (row-level security); to lecturers the platform shows only figures for the whole group, never individual conversations.
- Whoever operates the platform has technical access to the database. The operator sees conversations only when you report an answer (then that answer, your reason and, if you leave the box ticked, your message before it), when the nightly check flags an answer (then only Ava's answer, section 4), in case of a technical fault, or at your request.
- Highlights and notes in the reader are for you alone; lecturers and Ava do not see them. If your course uses shared notes, you can share a single note with the switch “Visible to the course”. Then all students and lecturers of this course see it at its passage, with your name. Lecturers of the course can hide a shared note from the course; you still see it. If you make it private again or delete it, it is gone for everyone at once, and the same happens when you leave the course.
- Lecturers see how many students have joined their course, but no names; further figures for the whole group only from three students upwards; never individual conversations, questions or learning progress. The exception are notes you make visible to the course yourself.
- The platform does not train any AI model with your conversations.
6 Recipients and processors
- IONOS SE, Montabaur: operation of the server that holds the application, database, search and backups. A data processing agreement with IONOS is in place; the data centre is in Germany.
- Own server, no third party: semantic search (embeddings with a local model) runs on the same server.
- Amazon Web Services EMEA SARL, Luxembourg (Amazon Bedrock, AI model Claude by Anthropic): for each answer from Ava, your message, the conversation so far, the course context and matching notes from your Second Brain go to Amazon Bedrock; for the nightly quality control only Ava's answer and the course sources she received for the answer, cited or not. Processing takes place only in data centres in the EU (Frankfurt, Stockholm, Milan, Spain, Ireland, Paris); without an EU region configured, the platform sends no request. Anthropic does not receive the content. The platform sends all requests to the AWS Region eu-central-1 (Frankfurt). There, Amazon Bedrock data retention is set to "none" for our account; according to the AWS documentation, AWS then writes no request or response data to durable storage and does not share it with the model provider (Amazon Bedrock, Data retention). Model invocation logging is switched off for our account. According to AWS, neither AWS nor the model providers use inputs to or outputs from Amazon Bedrock to train models (Amazon Bedrock FAQs). A data processing agreement with AWS is in place.
- ALL-INKL.COM (Neue Medien Münnich), Friedersdorf: sending the platform's e-mails. A data processing agreement with the provider is in place.
- Lecturers receive only the aggregated figures from section 5 and notes you make visible to their course yourself.
- Students of your course see only notes you make visible to the course yourself (section 5).
- University lecturers can use the platform in their courses on their own decision. Universities receive no personal data from the platform.
7 Retention
- Account and learning data: until your account is deleted, at most six months after your course ends if you do not keep using the account; we remind you by e-mail 30 days before.
- Deletion on request: You can ask for deletion at any time by e-mail to the address in section 1; a withdrawal from the contract through the withdrawal function also counts as a deletion request. From the platform's start, the operator handles deletion requests continuously by hand, without undue delay and within one month at the latest; this also applies to test accounts, early termination and data whose purpose has lapsed. Automatic deletion after the end of a course will be added later; until then the operator deletes these accounts as well.
- Backups: daily database backup on the same server, plus an encrypted copy of the database and the uploaded files kept by the operator on a device with an encrypted disk; both separate from live operation, kept for 14 days and then deleted automatically. Deleted data remains in the backups for that time but is not used from them. If a backup has to be restored, we re-apply all deletions made since it was created before the platform runs again.
- At Amazon Web Services: requests and answers are not stored permanently (data retention "none", section 6).
- Record of a withdrawal (account number and times, no content): kept as proof after the account is deleted.
- Highlights and notes in the reader: until you delete them, at the latest until your account is deleted; if a lecturer deletes a text, the notes on it are deleted with it.
- Reported answers: 90 days after the review is completed.
- Error reports in the operator's mailbox: 90 days.
- Server system logs: 30 days.
8 Cookies
The platform sets only cookies that are necessary for its operation (§ 25(2) no. 2 TDDDG): a session cookie that keeps you signed in (two weeks), a cookie against cross-site requests (one year) and, if you switch the language, a cookie that stores the chosen language for one year. There are no analytics, advertising or tracking tools, and the fonts are hosted on our own server.
9 AI notice and no automated decision-making
Ava is an AI system, not a human; her answers are generated by the AI model Claude by Anthropic, run via Amazon Bedrock in the EU. Answers can contain errors. Ava follows fixed teaching rules, which you can read under "How Ava works". An AI avatar is Ava with a lecturer's academic perspective. It is not the lecturer, says that it is an AI and does not speak for the lecturer on exams, grades or personal matters. The platform makes no decisions within the meaning of Art. 22 GDPR: there are no grades, no assessment of performance and no decision with legal effect. Ava's answers carry a machine-readable marking as AI-generated (Art. 50(2) AI Act): time, checksum of the text, name of the system, provider and, where applicable, the version of Ava's rules, and for answers of 400 characters or more also invisible Unicode characters in the text. The marking contains no personal data; how it is built and how to check it is explained under AI marking. The review schedule is calculated automatically; it is a learning aid and has no consequences for your studies.
10 Minimum age
The platform is for people aged 18 and over. You confirm this when you register; only the time of this confirmation is stored.
11 Your rights
You have the right of access (Art. 15 GDPR), rectification (Art. 16), erasure (Art. 17), restriction of processing (Art. 18) and data portability (Art. 20); you can export your Second Brain yourself at any time as Markdown or JSON. You can object to processing based on legitimate interest (Art. 21); for the nightly quality check you can do this directly with the switch under Profile and settings (section 4). You can withdraw consent at any time with effect for the future (Art. 7(3)). An e-mail to the address in section 1 is enough for any request.
You can lodge a complaint with a data protection supervisory authority (Art. 77 GDPR), in particular in the member state where you live. The authority responsible for the operator is the State Commissioner for Data Protection and Freedom of Information of North Rhine-Westphalia (Landesbeauftragte für Datenschutz und Informationsfreiheit Nordrhein-Westfalen), Kavalleriestraße 2-4, 40213 Düsseldorf, Germany.
12 Obligation to provide data
The platform cannot be used without an account. Only the invite code, username, name, e-mail address, password, the confirmation "at least 18 years old" and the acceptance of the Terms of Use are required; all other information, including degree programme and semester, is voluntary.
13 Version
Version of 16 November 2026. If data flows change, this notice will be updated.