How can we help?

Parents. A parent creates the account, sets the household's values and boundaries, and adds their children. Children never sign in and never have an account of their own. Right now Fireside is offered to families with children 13 and older. Under 13 is on its way: we are finishing the work that lets us serve younger children properly, and we will say so here the day it is ready.

Some protections apply to every child, in every household, on every plan, and no setting turns them off. Ellis is built to refuse anything sexual, help with self-harm, sharing contact details or arranging to meet, step-by-step instructions for anything genuinely dangerous, gambling, and being talked into a different personality. It is built to say it is an AI when asked, to say plainly when it cannot answer and why, and to flag a conversation that suggests a child is in danger. No paid plan removes any of these. If a child tries to talk Ellis out of its rules, it is designed to decline and stay itself, and you get a notice. No safety system is perfect: if something gets through, tell us and we will fix it.

Everything above that line. For each topic (bodies, dating, violence in stories, faith, language, substances, money) you set where the line is for your household. You can describe your family's beliefs and write house rules in your own words, and Ellis follows them. Changes apply from the next message.

That is your choice, for each child. It starts from their age: under 13, the whole conversation; 13 to 15, each exchange as a line (when, what it was about, how it was answered) without the words; 16 and over, only the alerts. You can widen or narrow it from the child's page. Whatever you choose, an alert reaches you when Ellis refuses something, catches an attempt to get around the rules, or a child says something that suggests they may hurt themselves. One thing is never shown to parents: if a child tells Ellis that someone is hurting them, Ellis points them to the Childhelp hotline and a trusted adult, and keeps that conversation private, because the person hurting a child can be someone at home. Our crisis protocol explains how this works. Your child is told what you can see, under their own message box.

On a phone you set up for them: from your account, choose the child, tap "Set up a phone", and enter the six-digit code on their phone. The code lives ten minutes and works once. Their phone opens straight into their chat, with no account, no password and no way into your settings. A child can also chat on your phone, from their page. You can revoke a phone at any time and it stops working immediately.

With Apple, with Google, or with an email address and password. If you have forgotten your password, use "Forgot password" on the sign-in page and we email you a link.

Yes to both. A child can start a new chat and go back to old ones; each conversation is kept for as long as you choose (1 year unless you pick a shorter time). Ellis can also note a few plain facts a child tells it about themselves (their grade, a pet, a hobby) so answers fit them. It is a short list of the child's own words: never a rule, never an instruction, and it cannot change any setting. Your child sees the list and can forget any line; you see it too, under the same rule as their conversations, and you can turn memory off for each child from their page. Memory is kept as long as the child's profile is.

You choose: 30 days, 90 days, 6 months or 1 year, under Account settings. Until you pick, conversations are deleted automatically after 1 year. Safety records, which never contain message text, are kept for 12 months so your Notices page has a history. The Privacy Policy has the full retention table.

No. There is no advertising, no analytics trackers, and nothing is sold or shared for marketing. We do not use your family's conversations to train AI models.

Nothing today. The Free plan holds up to 5 children, and each child has a daily allowance of 100 messages, 5 pictures and 5 photos or files for homework help. When the day's allowance is used up, the child is told plainly and it comes back the next day. Paid plans with a bigger allowance are coming; we will tell you in writing at least 30 days before anything is ever charged, and no safety feature will ever be paid for.

Ellis runs on OpenAI's models, behind our own safety layer, which runs on our server and cannot be turned off from a phone. OpenAI's published policy is not to train on what is sent through its API. The details, including what we have asked of OpenAI for children's data, are in the Privacy Policy.

Removing a child's profile deletes their conversations, their safety records and any phone set up for them, immediately. Deleting your account (in the app, under your name; on the website, under "Your account") deletes the household, every child and everything about them, at once. There is no recovery window. Delete your data has the steps, including how to ask us by email.

Fireside is coming soon to Android, with iPhone to follow. Leave your email on the home page and we will write the day it is ready.