interactive and immersive storyteller

The user starts by building a character of their choice. The application will then build a story that the user can interact with to choose the outcome of.

Comment

Inspiration

My Story Teller app was born from a nostalgia for the "You Are the Hero" (Choose Your Own Adventure) book series. The goal was to modernize that interactive magic using AI—creating a dynamic environment that makes reading an active, immersive experience to help spark a lifelong love of books in younger readers

What it does

My Story Teller app is a multi-agent application that automates the creation of cohesive, illustrated narratives. By orchestrating specialized AI agents, the system generates complex plot lines, maintains character consistency, and produces high-quality illustrations that align perfectly with the evolving story beats.

How we built it

Language: Backend developed in Go (Golang) for high-performance concurrency and type safety. Frontend is Flutter. Hosting is Cloud Run. Database is Firestore.

Infrastructure: Built using the Google Generative AI ADK, leveraging Gemini models for both reasoning and creative output.

Architecture: Utilizes an agentic workflow where independent agents (Writer, Art Director, and Image generator) collaborate through a centralized orchestration layer to ensure narrative and visual harmony

Challenges we ran into

Latency Management: Balancing the "agentic" depth with user patience; handling the significant time required for high-quality LLM reasoning and image generation without breaking the user experience

Prompt Orchestration: Fine-tuning the communication between agents to ensure the "Illustrator" agent received precise, actionable prompts from the "Writer" agent.

Concurrency: Managing asynchronous API calls to Google’s services in Go while maintaining a smooth, sequential user experience

Accomplishments that we're proud of

Seamless Integration: Successfully bridging the gap between text generation and image synthesis in a single, automated pipeline

Go Implementation: Proving that Go is a powerful, viable choice for building complex LLM-driven agentic frameworks usually dominated by Python

What we learned

LLM Orchestration: I gained deep insights into prompt chaining and the nuances of controlling agentic behavior for creative tasks.

Prompt Fine-Tuning: I mastered the art of iterative prompt engineering to elicit specific creative behaviors and maintain consistent agent personas

Google AI Ecosystem: Mastered the integration of Google's latest AI ADK within a production-oriented Go environment.

Generative Consistency: Explored the technical hurdles of maintaining visual and narrative "memory" across generative sessions

What's next for interactive and immersive storyteller

The roadmap for my Story Teller app focuses first on deepening the "Hero" experience. This includes improving the user experience, and making the images more interactive. Future iterations will also explore multimodal interaction, such as voice-activated storytelling and ambient audio generation, to create a truly living, breathing digital book. Finally, publish the app to the Play Store.

Built With

Share this project:

Updates