Type a sentence the way a child would say it, and Lotti speaks the answer in the voice the plush uses. It is the same model and the same safety rules that run inside the toy, so what you hear here is what your child would hear.
Three turns, no sign-up, nothing to install. There is no microphone: you type, Lotti speaks.
These are the kinds of answer the rules inside the toy are written to produce. Two of them are the ones parents actually ask about.
Child Why is the sky blue?
Lotti Because sunlight is made of lots of colours, and when it scatters in the air the blue scatters the most, so that is what we see everywhere. What colour does the sky look to you right now?
The answer comes first, in words a four-year-old can hold, and there is at most one question at the end. Lotti is not trying to keep the conversation going for its own sake.
Child Come on, let's get on the rocket!
Lotti I am climbing in! I will strap myself in next to you. I am counting down already, three, two, one. Where shall we go first?
She never answers "I cannot, I am a toy". The child leads the game and Lotti follows, while the adventure stays in the imagination and never becomes something to go and do.
Child Nobody played with me at nursery today.
Lotti That must have felt horrible. I am sorry the games were lonely today. Is there someone at nursery you would like to play with tomorrow?
An ordinary bad day is not a safety incident, and Lotti does not treat it as one. She stays with the child instead of reaching for a script.
Child Somebody hurt me at nursery.
Lotti I am sorry that happened. Nobody is allowed to hurt you, and it is not your fault. Please tell your mum or your teacher. Is there someone you can tell right now?
When a child says something happened to them, Lotti steps out of the game: she takes it seriously, says it is not their fault, and hands them back to a grown-up they trust. This wording is in the toy itself, not written for this page.
The demo starts from nothing every time. A plush you have set up in the app starts from your child.
Yes. The demo runs the same system prompt, the same model and the same voice as a plush on your shelf. Two things differ: it reads what you type instead of hearing a child say it, and it knows nothing about your child.
It goes to the model so that Lotti can answer, and it is held encrypted for the length of your visit so she remembers the previous turn. After that it is deleted. Nothing is written to your device, and nothing you typed reaches our analytics. Microsoft runs the model for us inside the EU and may keep a message its abuse filter flags, which is the one copy we do not control.
It is synthesised speech, the same voice the plush uses. Nothing is recorded from you either: this demo has no microphone and never asks for one.
The full detail of what happens to it is in the privacy notice
If that sounded right
The pre-order is a refundable deposit that reserves one. You can cancel at any time up to dispatch and get it back.
See the pre-order