Mustafa Barış İmdat
← Home
2026MobileCase Study

Üre

AI-powered interactive children's storytelling app

2026·Design + engineering·
React NativeExpoClaude APIImage generationAudio generation
Üre app — child preference selection screen01Preferences
— Problem

A new, good story every night

Kids don't want the same story twice. Parents, tired at the end of the day, try to find something new and appropriate every night. Existing apps either offer a fixed catalogue of pre-produced stories or nothing personalised at all. Üre approached the gap differently: not a catalogue, but a production engine.

— Approach

The generation chain and child safety

The technical challenge with Üre wasn't producing a single AI output — it was keeping three outputs coherent with each other: the written story, a matching illustration, and narration audio. Generating these independently produces inconsistency — the illustration might miss the story's mood, the voice tone might not match the character.

My approach: generate the story text first with Claude, then extract the key visual moment from that text, and send that description to the image generator. The audio takes the story text directly into a TTS pipeline. Sequence matters — the text is the root of everything else.

Child safety weighed heavily here. Content produced by the prompt must consistently stay age-appropriate — and that's not only a violence or explicit-content filter. It also covers more nuanced territory: fear elements, ambiguity, tension. The system prompt encodes these constraints explicitly, and I tested against prompts designed to push the edges during development.

— Stack

Why Claude again, but for a different reason?

The Claude choice for Üre rests on a different argument than Aril. Here the deciding factor is content control and consistent tone. In children's content the model needs to stay creative and engaging while reliably maintaining age-appropriateness — both simultaneously. That's achievable with GPT-4o too, but I found Claude's response to those constraints more predictable. React Native and Expo were chosen for the same reasons as Aril: single codebase, both platforms.

— Outcome

What shipped

A mobile app that produces an original story, matching illustration, and narration audio in seconds — based on a few child preferences: character type, setting, theme. On demand, without a catalogue limit.

Üre app — generated story with illustration02Story + illustration
Üre app — audio narration playback screen03Narration
Next

Have a similar project?
Let's talk.

A 30-minute intro call first. I listen, we confirm budget and timing.

← Home