What Happens When Bedtime Stories Star Your Own Kids
“Read me a story about me,” my daughter says, for the third night in a row.
This is not a request I can fulfill with the bookshelf. We own plenty of picture books—dragons, princesses, a bear who refuses to hibernate—but none of them feature a six-year-old named Mila who just learned to tie her shoes and is very proud of it. So I wing it. I make up a story about a girl named Mila who finds a talking frog in the garden. The plot collapses by page two. The frog wants to go to the grocery store for some reason. I’m exhausted.
Then a friend mentions a father of three who got tired of the same problem and built an AI that writes bedtime stories where your own kids are the heroes. It’s in a free two-week beta. I sign up, mostly because I’m curious what a machine thinks a bedtime story is. I expect generic slop with our names typed in.
The first story it generates is about my son, Liam, and a lost robot in a forest. The robot has been separated from its family. Liam, the story version, helps it find its way back by remembering landmarks—the big red rock, the split tree, the stream that sounds like it’s laughing. The writing is fine. The names are correct. But what catches me off guard is how Liam reacts.
He sits up in bed. “The robot was scared?” he asks. He wants to know why the robot didn’t just use a map. He points out that he wouldn’t bring a flashlight on a forest adventure because the flashlight in his room is broken. He argues with the story. Then he asks if we can read it again.
He has never once asked to hear a store-bought book twice in one night.
There is something real here, and it’s not just the novelty of new technology. When a child sees their own name in a story, the stakes change. They’re not observing a character who happens to be brave or clever—they’re rehearsing being brave and clever themselves. It’s the same reason teachers use students’ names in word problems. A math question about apples is boring. A math question about whether you can fit enough apples in your lunchbox is suddenly personal. The effort level shifts.
What surprised me was how the AI handled my daughter. Mila’s story was about a girl who moves to a new room and is afraid of the dark. It was formulaic, honestly. The villain was a “shadow rabbit” that she found both spooky and silly. But the girl in the story doesn’t defeat the rabbit by being brave. She talks to it. It turns out the rabbit is scared of the flickering night-light.
Mila asked, “Why would the rabbit be scared?”
We ended up having a real conversation about what’s actually scary at night. The creaky floorboard. The way the hallway light turns the wall into shapes. The fact that the closet door was left open a few inches. None of this was in the AI’s story. But the story gave us an opening to talk about it without Mila having to admit she was scared of something directly. The therapy-speak version of this is “externalization”—putting a feeling outside yourself so you can examine it. But at our kitchen table, over scrambled eggs, it just felt like a conversation.
I’ve started applying this even without the tool. The beta is free for two weeks, but I’ve learned the trick is not the AI—it’s the rule of putting the child in the narrative. Now, before bedtime, I ask one question: “What’s something you did today that you want in the story?” Sometimes it’s a trip to the playground. Sometimes it’s a fight over a blue crayon. Whatever they say goes into the plot.
If Liam says he climbed a big hill at the park, the story version of him climbs a hill that’s just a little too steep, and he figures out how to get down without falling. If Mila says she brushed her teeth without being reminded, that becomes an act of secret-agent-level competence. The stories are not impressive. They’re barely plotted. But they don’t need to be. The child already knows the plot—they lived it. What they want is to see it rearranged, to see themselves handling things.
One thing I’ve learned: don’t fix the story when it’s wrong. The AI made mistakes. It gave Mila a sister named Lily who doesn’t exist. She was offended. “That’s not my sister,” she said, with real gravity. So I let her correct it. We rewrote that part together. She decided the nonexistent sister was actually a flood survivor who came to live with us. Which is a bigger problem than we were prepared for. But the point is she was editing—deciding what makes sense in a narrative, what sounds true and what doesn’t. That’s a real skill.
There are limits to all this. The AI’s stories can get repetitive. The language skews generic. Some nights my kids want the book about the bear who sleeps all winter, and no algorithm can compete with a bear they’ve known for three years. The novelty is already fading a bit, and we’re only on day ten. I’m not sure I’ll keep paying once the beta ends. Two weeks is long enough to see what this is, and what it isn’t.
But the habit underneath—putting my kids in the stories we tell—is not going anywhere. It’s barely more effort than making up a halfhearted tale on the spot. One question before we read, and suddenly they’re the ones remembering the plot, arguing about details, taking the narrative somewhere I didn’t expect.
The other night, Mila said, “The girl in the story was me, but she wasn’t really me.” Exactly. That’s the point. A story is a place to try on a version of yourself and see how it fits. Some nights the story is good. Some nights it’s not. Some nights it’s 9 p.m., I’m tired, and the best I can do is read the same bear book for the fifth time. The AI can’t fix that, and neither can I. But the conversation the next morning, when Mila tells me what she would’ve done differently in the shadow rabbit story—that’s where the actual magic happens. It’s just slow, and unglamorous, and hard to measure. Like most parenting things.
Please indicate: Thinking In Educating » What Happens When Bedtime Stories Star Your Own Kids