What happens when something bigger comes up.
This page describes the current product design ahead of public release. MyHommi sits with children while they work, so occasionally a child may type something that is not about homework at all.
MyHommi is educational support. It is not a counsellor, not a crisis service and not emergency help. If a child is in danger, contact your local emergency number or a crisis service directly.
What it is designed to do
When a message suggests a serious concern, MyHommi stops behaving like a tutor. It responds calmly, does not press for detail, and points the child towards an adult they trust.
The check runs before the coaching itself. A safety response is held until the whole response has been checked rather than streamed word by word.
Two things run: a fixed check on the child’s own device, and the AI itself, which is also asked to notice when something is wrong. The device check is the one that works with no connection.
What parents see
The current design does not give parents a transcript. Where MyHommi records that a concern arose, it records the kind of concern and the date, not what the child said.
Limits we will not dress up
- No automated system catches everything, and MyHommi will miss things.
- It cannot reliably tell the difference between a joke and a real disclosure.
- It has not been independently validated by an external safeguarding body.
- It is not monitored by a human. Nobody is reading messages as they arrive.
- It works best in English. MyHommi supports ten languages, and the checks that run on the child’s device are strongest in English; in the other nine they are thinner and cover fewer kinds of concern.
For schools
MyHommi does not replace any part of a school’s safeguarding policy. The complete product and school arrangements will be reviewed before public classroom use.
- Safety questions
- hello@myhommi.com
Product information checked 23 August 2026