My Little Star Story Inc.

An AI-based startup in Seoul, founded in 2026. We build My Little Star Storybook, where a pipeline of generative AI models turns one photo of a child into a finished picture book starring that child: story, illustrations, narration, music and video.

Company facts

What we make

Parents upload one photo of their child and tell us what the child loves. In a few minutes our AI pipeline delivers a complete, personalized storybook:

• Story — a language model writes an age-appropriate story with the child as the main character, and can include family members, pets or a favorite toy.
• Illustrations — the photo becomes a consistent character sheet, and every page is illustrated in one of eight art styles while the child keeps looking like themselves.
• Narration and music — each page is read aloud, optionally in a parent's own recorded voice, over instrumental background music.
• Video and PDF — the book is rendered as a narrated video and a printable PDF, and can be shared by a link that can be turned off at any time.
• Editing — any page can be redrawn or rewritten with a short instruction.

Sample pages

The same page drawn in different art styles by the service.

How we build with Generative AI

Every book is made by a chain of generative AI steps, and the chain is built around Anthropic's Claude: Claude does the writing and directing, and the image, voice and music models work from what it writes. Each step below is a separate language-model call that returns structured JSON validated against our own schema, so a 12-page book stays consistent from the first page to the last.

1. Story plan — from the parent's short idea, the cast and up to five reference photos (read with vision), the model picks a title, a four-part story arc, and how the book opens and closes.
2. Pages and art direction — for every page the model writes the text a parent reads aloud, and also the scene, the camera framing and which characters appear. That art direction becomes the prompt for the illustration model, so the language model decides what every picture shows.
3. Continuity — the model lists the props, outfits and recurring things (a cat, a teddy bear, a red umbrella) that must look the same across pages, and writes the mood brief for the background music.
4. Narration script — the model marks up each page with emotion, pacing and breath cues for the voice model, without changing a single letter of the text, so the narration sounds like a parent reading at bedtime.
5. Rewriting on request — a parent can ask in plain words ("make every page one sentence", "end each line with a gentle tone") and the model rewrites the text of the whole book in one pass, leaving the pictures as they are. One tap undoes it.

With Claude we use structured outputs, vision and prompt caching: the long system prompt with our writing rules is cached across the steps of a book, which keeps each book fast and affordable to make.

The rest of the AI pipeline

• Character sheet — an image-generation model turns the child's photo into a character sheet in the chosen art style. After that, only the sheet is used, never the photo.
• Illustrations — the same image model draws each page from the language model's art direction plus the character sheets of everyone in the scene.
• Voice — text-to-speech and voice-cloning models read the narration script, optionally in a parent's own recorded voice.
• Music — a music-generation model composes an instrumental track from the mood brief.
• Video and PDF — our own rendering service puts pages, narration and music together into a video and a printable PDF.

Child safety and privacy

A child's photo is used only to create the character sheet; after that, only the sheet is used. Uploaded photos are visible only to the parent who uploaded them and are never shown through share links. See our privacy policy and terms of service.