Papyri

Papyri: Transforming facts into living adventures.

Comment

Inspiration

Passive learning is boring, so we created Papyri. Inspired by classic adventure books, our tool uses AI to turn educational content into a personalized quest. The name represents our goal: connecting old-school wisdom with the digital scrolls of tomorrow.

What it does

Papyri is a dynamic educational storyteller. Users provide their name, age, and a topic they are curious about (or even upload their own reference documents). The app then generates an interactive narrative in real-time. Each "verse" of the story includes:

  • Contextual Storytelling: Tailored to the user's specific age and topic.
  • Generative Visuals: Custom-painted illustrations that bring the scene to life.
  • Interactive Choices: Three distinct paths that let the user decide how the story—and the lesson—unfolds.
  • Synced Narration: An AI storyteller voice that reads the text aloud, highlighting sentences in real-time to assist with reading and engagement.

How we built it

We built Papyri using the following stack:

  • React & Tailwind CSS: For a responsive, "paper-like" aesthetic and smooth transitions.
  • Gemini 3 Pro/Flash: Used for the high-reasoning task of generating structured JSON that maintains educational accuracy and narrative flow.
  • Gemini 2.5 Flash Image: For real-time generation of vibrant, kid-friendly illustrations.
  • Gemini 2.5 Flash TTS: Powers the warm "Kore" voice narration.
  • IndexedDB: Implemented via a custom service to store "Chronicles" locally on the user's device, ensuring privacy and persistence.

Challenges we ran into

One of the primary challenges was the Multimodal Sync. Synchronizing the AI narration with the visual text highlighting required manual PCM audio decoding and a sequential pre-fetching strategy. To keep the experience fast, we had to fetch audio for sentences individually while the user was already listening, ensuring a gapless, low-latency storytelling experience. Another challenge was fine-tuning the System Instructions to ensure Gemini consistently returned valid JSON even when processing complex, user-uploaded document context.

Accomplishments that we're proud of

We are particularly proud of the Seamless Multimodal Loop. Seeing a child’s name integrated into a generated story, seeing a unique image of that specific scene, and hearing a voice read it back—all within seconds—feels like magic. We also succeeded in creating a "Library" system that feels premium, making the "Chronicles" page feel like a real archive of personal growth.

What we learned

Building Papyri taught us the immense potential of High-Reasoning Models for educational software. We learned how to handle raw PCM audio data in the browser and discovered that Gemini’s ability to "reason" over provided documents makes it a far superior tutor than traditional RAG systems. We also learned how important "Immersive UX" —the "story-font" and paper textures are just as important for engagement as the model itself.

Our journey with Gemini 3 Pro has shown us that AI’s greatest potential lies not in replacing human stories, but in amplifying our capacity to tell them. As we look ahead, Papyri remains committed to making every interactive adventure a discovery, every image a window into a new world, and every word a spark for a lifelong love of learning.

What's next for Papyri

The future of Papyri includes:

  • Multi-Speaker Dialogue: Using Gemini's multi-speaker voice config to allow characters within the story to actually speak to each other.
  • Gamified Knowledge Ranks: Expanding the "Explorer Profile" with actual quizzes based on the story history to earn badges.
  • Collaborative Scribing: A "Pass the Scroll" mode where two students can take turns making choices in the same story.
  • Pre-generating Scribes: Enable schools to leverage Papyri for pre-generating age-based scribes for students. To enhance engagement, integrate Gemini Veo 3.1 to produce short transition videos—such as a spaceship traveling between planets during a solar system lesson—creating a more immersive and interactive learning experience
  • AI-Tutor Integration: Evolving the platform from passive storytelling to active tutoring by utilizing Gemini’s "Thinking" mode for real-time Socratic learning.

Final thoughts

  • To WOW you, try to explore Domain Driven Design for 8 years old kid.

  • While testing the 'custom document upload' feature, I tried uploading my own CV. The result was incredible—it turned my career history and company experience into an epic adventure story written for a 10-year-old. Seeing my professional journey through that lens actually sparked a new idea: a tool specifically designed to transform standard CVs into immersive adventure stories. Check out these screenshots of my 'quest'.

Built With

  • ai-studio
  • gemini-2.5-flash-image
  • gemini-2.5-flash-tts
  • gemini-3-pro
  • indexeddb
  • react
  • tailwind-ccs
Share this project:

Updates