How AI is used

1. What is generated

• The text. At every level except the first, the lecture text is a retelling written by a large language model (at present, models from OpenAI). It follows the order, the conclusions and the results of the book in new words, with examples it chooses, sometimes from the examples of your own life that you gave. At the first level, “As written”, the book’s own text is used, cut into lectures.
• The drawings. Schemes are small programs written by a language model and drawn in the browser; painted pictures are made by an image model; sketchnote cards are composed by a model. No person draws them.
• The voice. By default your device reads the lectures aloud with its own voice. Studio narration, once available, is a synthetic voice from a speech provider, not a recording of a person, and never a clone of a real person’s voice.
• The analysis of your book (structure, language, what each part is about) and the price estimate are also produced by models.

2. What can be wrong

A retelling is not the book. The model can misstate a fact or a formula, leave out a qualification, add a detail that is not in the book, get a name or a date wrong, or draw a scheme that does not match the sentence. It can be fluent and confident while wrong.

The drawings are checked by machine for mechanical faults (text off the canvas, overlapping labels), but nobody reads every lecture before you do.

Use the lectures as a way into a hard book, and go back to the book for anything you rely on: an exam, a decision, a calculation, a citation. The Service gives no medical, legal, financial or other professional advice, whatever the book is about.

If you find a bad part, tell us at contact e-mail. It helps us improve the pipeline, and a job that failed on our side is charged only for what was made.

3. How it is marked

The site, the player and the download step say that lectures, drawings and voices are AI-generated, and files you download carry the same note in their metadata where the format allows. This follows the transparency duty of the EU AI Act (Article 50) and the terms of our AI providers. Please keep the note if you pass the files on.

4. Whose it is

The book stays yours. We claim no rights in what you upload and use it only to make your course (see the Terms of Service, section 3).

The lectures are not the author’s text, except at the first level, and are not endorsed by the author or the publisher.

We claim no copyright in the lectures, drawings or narration made for you. Whether AI output can be protected by copyright, and who holds it, is unsettled, and a course built on a copyrighted book may count as a derived work of it. Use the courses for your own study; before publishing or selling one, take advice.

5. What the models receive

To write and to draw, we send the AI providers extracts of your book’s text, your choices and the examples you typed, never your account details. Under the providers’ API terms this data is not used to train their models. Details in the Privacy Policy.