Using AI responsibly

Concept analysis and enrichment — when you submit a story concept, the system analyses it in the context of your child's profile: their age band, interests, and anything you've asked to avoid. This context sets the tone, vocabulary level, emotional register, and thematic guardrails for the entire series before a single word of story is written.

Age-calibrated generation — stories are generated using age-band-specific rules that shape sentence complexity, narrative pacing, and appropriate themes. A story for a 3–5 year old is written very differently from one for an 8–10 year old, even if the core concept is the same.

Series continuity — all five chapters of a series are generated with awareness of each other. Characters, locations, and narrative threads are kept consistent so the series reads as a coherent whole rather than isolated episodes.

Visual consistency — each chapter illustration is generated using a visual style description derived from earlier chapters, helping characters and settings look recognisably consistent even though each image is generated independently.

Audio narration — chapter text is passed to an AI voice synthesis service to produce natural-sounding narration. The narration uses the same text that was approved by the safety pipeline.

Input screening — story concepts and child profile details are reviewed before they are used to generate content. Concepts that fall outside acceptable boundaries are blocked before generation begins.

Generation-time guardrails — age-appropriate constraints and topic restrictions are built directly into how stories are requested from the AI, applied consistently for every story, not just flagged cases.

Output moderation — generated story text is reviewed by an automated moderation system before it is stored or displayed. Content that does not pass moderation is not shown to users.

Image prompts are anonymised — the prompts used to generate chapter illustrations do not include your child's name or any personal details.

It doesn't know your child — the AI works from the profile you provide. It can personalise by age band, interests, and things to avoid, but it has no lived understanding of your child's individual personality or emotional sensitivities beyond what you tell it.

Repetition and familiar phrasing — AI language models can sometimes produce stories that feel slightly generic or use recurring phrases. Trying a more specific or unusual concept often produces a better result.

Illustration drift — while we work to maintain visual consistency, AI image generation does not guarantee pixel-perfect character continuity. Characters may vary slightly in appearance between chapters.

Stories are not facts — story content is fiction and should not be treated as factually accurate, educational, or medically appropriate without review.

Transparency — we will always be clear about when and how AI is used in creating your child's stories.

No child data in AI prompts — your child's name is never included in any prompt sent to an AI provider. Story concepts and age-band context are used, but not identifying details.

Continuous improvement — we review safety incidents and update guardrails when issues are found. This page will be updated to reflect any significant changes to how AI is used in the service.

Human oversight — significant changes to the generation pipeline or safety approach are reviewed by the founders before deployment.