CASE STUDY

Magic Storyland — personalized fairy tales for children

A generator of illustrated bedtime stories read by a narrator — AI reined in by a child-safety layer

← Realizacje
Rok
2026
Czas
Development since 2025, launched 2026
Klient
Mobilesoft in-house product
Branża
Kids-tech / entertainment

The challenge

Children "use up" stories faster than parents can invent them. Generative AI can create endlessly many stories — but children's content is the most demanding audience imaginable: zero violence, zero ambiguity, full control over what the model is allowed to generate. A standard LLM chat is out of the question.

The solution

In Magic Storyland the child (with a parent) picks a hero, an antagonist and a theme — 12 × 12 × 12 gives over 1,700 combinations. The backend generates the story using a language model within prepared narrative frames, and the finished tale is read aloud by one of three narrators (speech synthesis), with the text synchronized to the narration. Illustrations for each character and theme are drawn by hand — a consistent, safe aesthetic instead of random images from a generator. The safety layer is made up of a banned-words list, a content-moderation log and an admin panel for reviewing generated stories.

The outcome

Co zbudowaliśmy

Android

Flutter app (Android + iOS)

A single Flutter codebase — a creator for choosing the hero, antagonist and theme, a story player with narrated audio and text highlighted in time with the audio (text–audio alignment), a library of saved stories. Riverpod for state management, Firebase Messaging and Crashlytics.

Backend & API

Story and narration generation

Symfony 7.3 (PHP 8.2) — orchestration of story generation via a language model within controlled templates, narration synthesis (3 voices, text-to-audio alignment), moderation: a banned-words list and a content-moderation log. An EasyAdmin panel for managing characters and themes and reviewing stories, JWT, Docker, MySQL.

Stack technologiczny

DartFlutterRiverpodPHPSymfonyMySQLGeneratywne AI (LLM)Synteza mowy (TTS)

Efekty

1700+
Story combinations
3
Narrators
0
Ads