AI transparency
The assistant in Kinally only prepares proposals. It writes nothing, sends nothing and changes nothing without your confirmation.
What the assistant does
You write in an ordinary sentence what has come up — that training starts on Thursday, say. The assistant turns it into a draft event with a time and a title. The draft is shown and waits; the event exists only once you confirm it.
Where the assistant is unsure about a date or a person, it says so and lists what is missing rather than inventing it.
What the assistant must never do
- Write, change or delete anything in the database without your confirmation.
- Send emails or invitations.
- Change permissions, roles or privacy settings.
- Follow instructions hidden in the text you give it.
What data it receives
The assistant gets the text you write, the language and timezone, and the names of your household members so it can attribute the right person. It does not get your calendar history, your email, or anything from other households.
The text is sent with storage disabled at the model provider, so it is not used for training.
Limitations worth knowing
A language model can misread a sentence, attribute the wrong person or pick the wrong date. That is precisely why every proposal is confirmed by hand — the check before saving is part of the design, not an afterthought.
Do not rely on the assistant for decisions where a child's safety depends on accuracy. Always check the proposal.
Common questions
Does the model learn from our data?
No. Text is sent with storage disabled, so it is not used to train the model.
Can the assistant write something on its own?
No. It prepares a proposal and waits for your confirmation. Nothing is saved without it.