When an AI writes your child’s bedtime story, the first question worth asking isn’t how clever the AI is. It’s: what is it not allowed to write? Here’s our answer in plain language: what we filter, what we tune, and what only you control.
The hard lines
Some things never make it into a chapter, for any child, at any setting: violence beyond gentle storybook peril; horror; anything romantic; real-world heavy topics arriving uninvited (a chapter will never spring death, divorce, or the news on your child); anything hateful, ever. These aren’t model preferences, they’re rules written by people and checked on every chapter, and they don’t bend because a plot would be more exciting on the other side of them.
The tuned layer: how scary is scary enough
A story with zero tension has no pulse; kids come to stories partly for a small, safe flutter of it. The question is size. A creaky bridge at night is a five-year-old’s worth of scary; getting briefly lost in a market might suit an eight-year-old. Tension scales with your child’s age, and one rule holds at every age: whatever fluttered mid-chapter is fully resolved by the end. Everyone safe, everyone home. The scare never travels to sleep. (This is the same principle as the gentle cliffhanger: a door left ajar, never an alarm.)
What only you control
You choose what the story leans toward: themes, values, the things worth celebrating. And you choose what it gently helps with. A new baby coming, a house move, a rough patch at school: tell the saga, and it carries the theme sideways, through the characters, without putting your child on the spot (how that works). That’s steering we wouldn’t want to automate. Nobody knows what your child needs from a story better than you do.
The check before your child hears anything
Every chapter passes review before it reaches a bedroom: within the rules, tension the right size, consistent with the world the story has built. A chapter that fails gets fixed or rewritten, and the rules sharpen a little each time something slips through to a human reader and makes them wince. That loop never ends, and we’d rather say so than pretend the system is finished.
The quiet safety feature: no sermons
It would be easy to end every chapter with “and THAT’s why we always tell the truth,” and it would be wrong. Kids smell a sermon instantly, and it curdles the story. Values ride inside what characters do, never in a speech at the end. We count that as safety too: it keeps the story a place your child trusts.
The companion question, what the AI knows and remembers about your child, has its own plain-language answer. And the theme controls are real product, not marketing: see how it works.