StoryWeave

A typical page from one of our stories
We allow each parent to make their own account, save stories and child profiles, and log back in
Creating a child profile, with name, age, story pacing, and favorite themes
Creating a new story, with custom themes, characters, moods, and lengths

Inspiration

1 in 6 children are neurodivergent. Traditional bedtime stories, made for a mass audience, use generalized pacing, language, and sensory input, which often fails neurodivergent children. Parents spend hours searching for appropriate content, often settling for repeated stories which don't always work for their kids. We're creating StoryWeave for those users, so the kids can have curated stories which fit their needs!

What it does

StoryWeave generates personalized bedtime stories that adapt to each child's cognitive profile, sensory needs and emotional state. The system creates fundamentally different story structures based on ability profiles, not just adjusting formatting or presentation, but restructuring the narrative architecture itself.

How we built it

For the backend, we used AWS Bedrock (Claude) for generating stories which are perfectly curated for children with ADHD, autism, or anxiety. Parents can select preferences based on their children's needs and their likes and dislikes to generate a specialized bedtime story. We used Google's Gemini API to generate custom images for the stories and ElevenLabs Generative Voice AI for converting text to speech. We used AWS Bedrock models to customize our ElevenLabs voices for better emotional responses in their narration. For the frontend, we used React and Tailwind CSS for displaying the website. Finally, we integrated the frontend and backend using Flask API and also used AWS DynamoDB for storing the created user profiles and stories.

Challenges we ran into

Connecting the frontend and backend while also integrating the DynamoDB database as we did not have much experience with setting up a database using AWS console.

Accomplishments that we're proud of

We were able to seamlessly integrate our AI models from several different providers for story generation and create a working MVP, and we're proud of how we used AWS Bedrock models to enhance our ElevenLabs narration with more emotion - demonstrating the utility of being flexible in our providers and synergistically improving performance over using any one service individually. We're happy we were able to adapt to new technologies like AWS DynamoDB, Bedrock and Flask. The website UI also turned out great!

What we learned

We learned more about neurodivergence in children, and how we can use AI models such as claude in AWS Bedrock and Google's Gemini to make an impact in generating stories for each kid's profile. We were also amazed by the capabilities and versatility of these models and look forward to exploring them further.

What's next for StoryWeave

We would like to conduct further research regarding the specific differences in needs for neurodivergent kids, and make our profiles more accurate and helpful after collecting feedback from our target audience, specifically the kids and their parents. The scope and abilities of our product can only get better as models improve and become more steerable with time.