Nothing leaves the sandbox without you.
When your child invents a new character, it lands in a private sandbox — visible only to them in that session. When they finish a story, it lands in your approval queue. You read every page. You approve, edit, or send back. Then — and only then — does anything export, share, or print.
Sandbox mode for new characters
When your child creates a character, that character is private to their session. They can play with it. It does NOT enter the shared character family until you approve it from your portal.
Page-by-page review
When a story is finished, you see every page side-by-side: text on the left, illustration on the right. You can approve the whole thing, edit a sentence, regenerate an image, or send the story back to your child for revision. Books in “awaiting parent” status never reach the printer.
Why it's the safety contract
We don't rely on AI moderation alone. Sparky is bounded, the image model is filtered, content moderation runs on every output — but ultimately, you are the gate. If anything slips past every other check, you catch it on review. That layered safety is the only kind worth claiming.