What Talebrix is, and what it talks to

It sits on a shelf in your child's bedroom, and it is meant to live out rather than come out. Children speak their ideas and choose options on the touchscreen, with a parent nearby.

The writing and the drawing do not happen inside Talebrix. They happen on our servers, which is why it needs a wi-fi connection and will not make a book without one.

That is a real trade-off and it is worth saying plainly rather than in a footnote. A machine that did everything on the device would work in a cottage with no signal. It would also be a far more expensive one, and the books would be worse — the whole reason the same idea comes out differently for a four year old and an eight year old is that there is something substantial doing the writing.

They are generated, each time, by an AI model. There is no library of stories inside Talebrix and no set of templates with a gap for a name — which is the whole reason the shelf on the homepage has a book about the bin lorry on it. Nobody would ever print that book, because there is a market of exactly one for it.

What Talebrix sends is the answers your child gave and the ages it has been set to. What comes back is a finished book: the words, the pictures, and the cover. It then prints and binds it in the room with you.

It is worth being clear about what this does and does not mean for the child, because the two get run together constantly. There is nothing for a child to hold, unlock or log into, and no account that belongs to them. What there is, is a spoken exchange of a few questions, out loud, in the room, with you there, plus touchscreen choices. The aim is a finished book in minutes; end-to-end speed still needs validation.

You set up a profile for each child in the house, because a household usually has more than one and the right book has to go to the right person. You choose what to call a profile, and it does not have to be their real name — a nickname, what they call themselves, anything at all. Whatever you type is what goes on the cover, so plenty of people use the name only their family uses.

Choose an age or reading level and update it as your child grows. An exact birthday is not required.

The planned device sends speech to a cloud recognition service and uses cloud processing for the story and pictures. Audio deletion and no provider training are requirements we will verify with suppliers. We will publish the confirmed data flow before launch.

This website asks for none of it — see how we handle the one email address we ask for. The age demonstration on the homepage runs entirely in your browser: it stores nothing and sends nothing anywhere.

Most of the questions we get are about what Talebrix might quietly turn into once it is on the side. These are the answers, and they are design constraints rather than a position we might revisit when it suits us.