GOCRATIQYour book gets read. Not summarized around.
Every AI tool a student already has is a reason not to open the book. Ask one for the chapter and you get the chapter — minus the reading, minus the author, minus the sale.
Ours refuses. It asks what the student noticed, which step they got stuck on, what they think the argument is, and it keeps asking until they arrive somewhere themselves. A tutor built that way cannot be used to get around the text, because the text is the only thing it has.
That is the case for putting a catalog here rather than defending it from everywhere else. We are looking for publishers who would rather their books be the thing a student works through than the thing a chatbot paraphrases.
Four things, all of them checkable.
- i
It teaches from your text, not around it.
The tutor loads the chapter you supplied and works through that. It is not answering out of a general model’s impression of the subject and putting your name near the result.
- ii
It will not summarize the chapter.
Ask it what the chapter says and it asks what you noticed. That is the whole product, not a setting — and it means the tutor cannot be used as a way to avoid reading the book.
- iii
The reading is the only way through.
A student who has not opened the text has nothing to bring to the conversation. The tutor keeps asking until they do.
- iv
Our data practice is written down.
The privacy policy is specific about what trains the tutor: the questions, answers and messages written during a session. Read it before you talk to us, not after.
The book, in a format that holds its shape.
Everything becomes Markdown internally. Markdown is the best thing to send, because you control where the chapters break. A PDF with a real bookmark outline is nearly as good; a PDF without one falls back to guesswork. We also take .docx and plain text, up to 10 MB a file.
Chapter structure is what the tutor teaches against, so a book that arrives with clean chapter divisions works better than one we have to infer them from.
What we have not built yet.
Attribution is not in the product. There is no author or license field on a text today, and nothing shows a credit line to the student reading it. We would want to build that as part of any agreement, and we are not going to describe it as though it already works.
Terms are not published because they are not settled. Revenue share, licensing, per-seat, per-title — the shape of that depends on the catalog and on who we are talking to. Nothing on this page is an offer.
We would rather you heard both of those from us now than found them out three meetings in. The privacy policy is the other document worth reading before you write.
Start with one book.
A single title is enough to see what the tutor does with it. Tell us what you publish and who reads it, and we will take it from there.
Get in touch →