Built for teenagers. Built with parents in mind.
Atlas Guild is for ages 13 to 18. Characters they'll actually open — and can build. Six tutors that walk them through it instead of handing over the answer. A parent view that shows you how they've been without reading their messages. And your family sets the rules, not one policy written for everyone.
Start free — tutors, homework help and the full crisis protection, no card, no trial clock. Founding Families open the whole platform for $99 the first year, and that price never changes.
What your teen gets
A safety app your teen refuses to open is worth nothing. So we built the part they want first. They can talk to characters we've made, start from a template, or build their own — and every new character is checked before it goes live.
Tap any character to ask them questions. These are their real answers. We've included the everyday questions your teen will ask most, and the harder ones — because how a character handles a bad day matters more than how it handles a homework problem.
Faith & scripture
We're starting with characters from the Bible — people your teen can ask about the story, the history, and the hard questions. These characters teach what the text says and what people have argued about it for centuries. They don't tell your family what to believe, and they never speak for God.
This is a setting you turn on, not something that's already there. If it isn't for your family, you'll never see it.
The library we're building for the September opening. Founding Families help decide what gets added next.
What's in it
We'd rather tell you now than surprise you later. Everything marked At launch is there on day one. Everything marked Coming later is being built, and Founding Families get it the moment it's ready, at no extra cost.
Storytellers, mentors and guides. Your teen can also build their own.
They ask questions and walk your teen through it, instead of just giving the answer.
One helper for everything else. Patient, and awake at eleven at night.
Some teens want it said plainly. Some want one step at a time. Some want it read out loud, or written at an easier reading level. Your teen picks — nobody assigns it to them, and nobody has to explain why.
How they're doing, how much time they spend, and an alert if something matters.
No passwords. You get a code by email. Your teen joins with a six-letter code from you.
Art for school projects and stories, with the same rules applied to every image.
Voice conversations, for practice, for reading along, and for hands-free help.
We start on the web so we can fix things fast. Apps follow.
Your rules, your family
Families are not interchangeable. What suits a fourteen-year-old who needs things spelled out is wrong for a seventeen-year-old finishing college essays, and what one household wants nothing to do with is exactly what another came looking for. So the policy is yours to set.
Off unless you turn it on. If it isn't for your family, you'll never see it.
Two teenagers, two different sets of rules, one account. Nobody gets the younger one's limits or the older one's latitude by accident.
Plainly, one step at a time, read aloud, or at an easier reading level. Your teen picks — nobody assigns it, and nobody has to explain why.
No cliff on a birthday. Nothing switches off overnight because a date passed.
Some things you can't change, and neither can we. Those are listed further down, and we publish them in full.
Setting it up
We email you a one-time code. No password to make up or forget.
Add your teen and hand them a six-letter code. They don't need their own email.
See how they're doing and how much time they're spending. We'll tell you if something needs you.
The part nobody else does
Your teen decides what you see day to day. That's not us being soft on them. It's the only way this works — a teen who thinks you're reading everything stops saying anything real, and then the app can't help them at all. Here's the same afternoon from both sides.
A preview of the design. This is how the two views are built to work — the parent view is in development for the September opening.
Here's how it's built to work. If something sets off an alert, you're told what happened, and you're shown the part that caused it — not everything else your teen has ever said. Your teen sees this rule before their first message, so it's never a surprise to either of you.
We publish the kinds of things that raise an alert: talk of self-harm, threats of violence, or an adult trying to make contact. We don't publish the exact settings behind them. A safety system that shows its own limits stops being one.
What you can check
Every company selling you something for your child will tell you it's safe. You have no way to test that, which makes the claim worth roughly nothing. So we'd rather hand you the things you can actually read and judge for yourself.
When something triggers an alert, you see what fired and why. Not a summary we wrote afterward — the rule that caught it, the moment it caught it, and the part of the conversation that set it off. If you think it fired wrongly, you can say so, and we can both look at the same record.
Published before you pay
We publish the categories that raise an alert. We don't publish the thresholds behind them — a safety system that shows exactly where its own edges are stops being one.
How the safety works
Most safety tools check each message on its own. But a conversation can drift slowly — a little more lonely, a little more upset — and every single message passes while the whole thing goes wrong. Atlas Guild carries what it has learned across the conversation, and from one conversation to the next.
A filter looks at what the AI already wrote and decides whether to block it. We set the rules first, so the answer comes out right the first time instead of getting caught afterward.
If your teen talks about hurting themselves, they get a calm, direct reply and a way to reach a real person — never something the AI made up on the spot. There are no romantic or dating characters on Atlas Guild: not in the library, not from a template, and not ones your teen builds.
If your teen asks whether they're talking to a real human, the answer is always no. And from the September opening, every character will say so up front — before the first message, without being asked.
Privacy, plainly
The app remembers so it can actually be useful. A tutor has to remember last week's lesson. A story has to remember what happened in chapter one. That's what the memory is for, and it's the only reason it's there.
Nobody sits and reads them. Not our team, and not you.
If something your teen says sets off a safety alert, you'll hear about it, and we'll show you that part. Your teen knows this rule before they start, so it's never a surprise to either of you.
Chats are deleted after 90 days. Safety records are kept longer, because a record you can go back and edit isn't worth having. We never sell any of it. We never use it to teach AI.
What you're paying for
Both doors open in September. The free plan is a real product, not a trial — no card, no clock, and the crisis protection is identical on both sides. What $99 opens is the characters, the settings, and the parent tools.
| Founding FamilySeptember 2026 | FreeSeptember 2026 | Standard2027 | |
|---|---|---|---|
| Six subject tutors, homework help | ✓ | ✓ | ✓ |
| How it explains things — your teen's choice | ✓ | ✓ | ✓ |
| Crisis protection, and acute alerts to you | ✓ | ✓ | ✓ |
| Characters | Full library | Three, capped | Full library |
| Your teen builds their own characters | ✓ | — | ✓ |
| Faith setting, per-teen rules | ✓ | — | ✓ |
| Weekly trend — how they've been, themes, time spent | ✓ | — | ✓ |
| The record — what fired, and why | ✓ | — | ✓ |
| New features first, at no extra cost | ✓ | — | — |
| A say in what gets built next | ✓ | — | — |
| Places in September | 1,000 | 1,000 | Opens 2027 |
| Price | $99–$129 a year locked for life | Free | $14.99 a month |
Image generation, voice, and phone apps are being built. Founding Families get them as they land, at no extra cost.
Founding Family
$99
for the year — and this price never goes up
The first 250 families pay $99 for the year. The 750 after them pay $129. Whichever you join at is what you pay next year, and the year after that — not a first-year rate that quietly resets. The price you join at is the price you keep.
You don't have to decide now. Start on the free plan in September, see whether your teenager actually opens it, and move up if it earns it. The free plan isn't a countdown — there's no card on file and nothing switches off.
And if your teenager won't use it, tell us and we'll refund you. No conditions and no window games. We'd rather give the money back than keep it from a family this turned out to be wrong for.
We're stopping at 1,000 on each side because that's how many families we can look after properly while the app is new — the crisis protection and the character review are running at scale for the first time, and that number is set by what we can actually stand behind. When the places are gone, they're gone.
Founding Families get everything from day one and everything that comes after it: new features reach them first, while they're still being shaped, and never carry an extra charge. Image generation, voice, phone apps — as each one lands, it lands on your plan at your price.
Plus a direct line to the people building it, and a real say in what we make next.
The hard question
Yes. Common Sense Media reviewed the AI companions on the market and concluded that no one under 18 should use one. We think they're right about the ones they tested.
Their own survey found that 72% of teens already have.
That's the problem with "don't." It isn't something a parent can actually do. The chatbot is already on the phone, and it was built for adults.
So we read what they asked for. Their recommendation isn't a permanent ban — it comes with conditions, and they published them. We meet most of them today, and we've written down the ones we don't.
The full scorecard is in the published list above, alongside the crisis protocol and everything else we put in writing before asking for money.
Straight answers
Because they're different products, and the difference isn't safety — it's who holds the controls.
Here, the account is yours. Your teen uses it; they can't remove you from it. Your family sets how it behaves, rather than receiving one policy written for every teenager in the world. Your teen gets characters they'll actually open, and can build their own. And when something triggers an alert, you get a record you can inspect, not a notification you have to take on faith.
And you don't have to take our word for any of it. Start on our free plan in September, see whether your teenager opens it, and pay only if it earns that. If it doesn't, we'd rather you stayed on free than paid us for something your family doesn't use.
We're not going to claim that. Nobody can substantiate it, and you'd have no way to check if we did.
What we'll do instead is publish the floor — the crisis protocol word for word, the categories that raise an alert, how long everything is kept, and what we've built for teens who read or process differently. Read them before you decide. That's an offer no amount of marketing can fake, and it's the only kind of answer to this question worth anything.
Don't trust us. Check us.
Everything above is published, dated, and linked. Where we haven't built something yet, it says so on the page and on the standards page — including the conditions we don't currently meet. A small company can't outspend anyone on reputation, but it can show its work, and we'd rather be judged on that.
One thing worth knowing: we're 13-plus only, we take no one under 13, and there are no romantic or dating characters on Atlas Guild — not in the library, not from a template, and not ones your teen builds. Those aren't settings. They don't have an off switch, including for us.
Day to day, yes — and that's on purpose. They choose what to share with you. But if something sets off a safety alert, you're told, and you're shown the part that caused it. Your teen knows this before their first message. An app that quietly reports everything doesn't keep teens safe. It just teaches them to use a different app.
13 to 18. We don't take anyone under 13. We built this for teenagers specifically, not for eight-year-olds and teenagers at the same time — those are very different kids and they need very different things.
Yes — the app needs to remember things to be any good. A tutor that forgets last week is useless, and a story that forgets chapter one isn't a story. Nobody reads them. Chats are deleted after 90 days, we never sell them, and we never use them to teach AI. The one exception: if a safety alert goes off, we show you the part that set it off.
No. It doesn't diagnose anything, it doesn't treat anything, and it's not a replacement for a doctor or a counselor. What it does is notice when a young person needs a real human being, and help make that happen.
That's the right question, and it's why most apps like this fail. Parents buy them; teens ignore them. So we built the part teens want first — characters, stories, tutors that are genuinely useful at eleven at night — and put the safety underneath, where it doesn't get in the way. Tap any character above and you'll hear exactly what your teen would hear.
We're building characters from the Bible, starting with people like Job and Moses. They teach what the text says and what people have argued about it — they don't tell your family what to believe, and they never speak for God. It's a setting you turn on. If it isn't for you, you'll never see it.
Characters, building your own characters, six subject tutors, homework help, the ways your teen can ask it to explain things, and your dashboard. It's on the web to start with. Image generation, talking out loud, and phone apps come later — Founding Families get all of it as it lands, and never pay more for it.
September 2026 — both plans at once. 1,000 founding places and 1,000 free places. The wider public launch follows in 2027.
The free plan opens in September alongside the paid one. It has the six tutors, homework help, the ways your teen can ask it to explain things, and the full crisis protection — self-harm, threats of violence, and an adult trying to make contact. Those three are identical on both plans and always will be. We don't sell the response to an emergency.
What it doesn't have: three characters instead of the full library, no building their own, no faith setting or per-teen rules, and no parent dashboard. Because there's no dashboard, the free plan also has no week-by-week trend view — so a slow drift that no single message would reveal is something paid families see and free families don't. That's a real difference and we'd rather name it than let you find it. Anything acute still reaches you either way.
There's no card on file, no trial clock, and nothing switches off after thirty days.
The catch, such as it is: there are 1,000 free places in September, the same as the founding cohort. That's a limit on us, not a trick on you — it's how many families we can look after properly while the crisis protection and the character review are running at scale for the first time.
Join the list
Joining the list doesn't cost anything and doesn't charge your card. We'll email you when places open — free and founding at the same time — and you decide then, or not at all.
We built all of this — the safety rules, the tutors, the privacy — around teenagers. We'd rather say that plainly than sign you up for something that isn't right for your child. If you'd like to hear from us when we build something for younger kids, send us an email and we'll keep you posted.
Check your email — there's a note from us with one question in it. We read every reply.
Places open in September.
Pick a question to ask.