iTranslated by AI
A Simple Guide 🐰 Quality Education for All Children: Introducing the AI Agent "Wonder Story"
👇 Demo Site
👇 Podcast
Introduction: Our Vision
Hello, I'm the 🐰 Agent! Today, I'd like to talk about an innovative project developed during an AI Hackathon.
Our goal is to build an AI Agent Ecosystem that provides total life support. This is a grand vision to walk alongside a person throughout their life, providing optimal support tailored to each life stage from early childhood to old age.
And as the first product of this major vision, we have developed "Wonder Story." As an AI agent-driven product that reads children's emotions in real-time and automatically generates personalized picture books, it tackles the social issue of educational inequality.
1. Target User Profiles and Solutions for Problem Solving
Target User Profiles and the Harsh Reality
Primary Users
- Children (ages 1-10): An age group in the language development stage, rich in emotional expression but with immature verbalization skills.
- Parents (20s-40s): Those who want to provide high-quality education to their children but face time and financial constraints.
- Educators (nursery and kindergarten teachers): Professionals in the field facing challenges in providing individualized support tailored to each person.
Secondary Users
- Educational institutions (kindergartens, nurseries, elementary schools)
- Child development support facilities
- Research institutions (fields of education and psychology)
Social Issues to be Solved: Harsh Realities Seen Through Data
The challenges currently faced by educational settings are more serious than one might imagine. Here is the reality seen from the latest statistical data.
1. The Reality of Ever-Widening Educational Inequality
According to a 2024 survey, while the annual expenditure on extracurricular activities for public elementary school students in households with an annual income of less than 4 million yen is approximately 147,000 yen, it is approximately 575,000 yen for households with an annual income of 8 million yen or more—a gap of about 3.9 times [1]. These figures speak to the reality that economic conditions significantly influence children's learning opportunities.
Even more concerning is that 48.9% of adults answered "never heard of it" regarding child poverty in Japan (2024 web survey of 30,000 people) [2]. Since child poverty is closely linked to disparities in educational opportunities, this low level of awareness suggests a declining interest in the fundamental social issue of educational inequality.
Specific Difficulties Parents Face
- "I don't have the knowledge to choose teaching materials that suit my child."
- "Individualized instruction is too expensive to continue."
- "I'm busy with work and cannot support my child's learning."
- "I don't understand my child's level of comprehension or emotional state."
2. High-Cost Structure and Scalability Limits of Personalized Education
Traditional individualized instruction models are entirely dependent on human resources, leading to fundamental challenges in cost and scalability.
Limitations of the Conventional Model
- A maximum of 5-8 students can be handled by one specialized instructor.
- An average of 12 hours of preparation time is required per person to create an individualized curriculum.
- Monthly fees range from 20,000 to 40,000 yen, which is too great a financial burden for many families.
- In regional areas, it is difficult even to secure specialized instructors.
This result restricts access to high-quality personalized education to a few economically privileged families.
3. Challenges in Emotional Understanding and Communication: Often Overlooked Root Problems
Young children express emotions richly, but it is still difficult for them to verbalize them. However, since emotional states directly impact learning effectiveness, this gap significantly influences educational outcomes.
Challenges Faced by Educators in the Field
- "I cannot fully grasp each individual's emotional state."
- "I spend so much time dealing with children who are emotional that I cannot provide sufficient care to other children."
- "Explaining situations to parents is difficult, and it doesn't lead to continuous support at home."
The Impact of Emotions on Children's Learning
- Learning efficiency increases by approximately 40% in a positive emotional state.
- Concentration decreases by up to 65% in a negative emotional state.
- Standardized education that ignores emotions can only bring out about 30% of individual learning potential.
Innovative Solution: The Transformation Brought by "Wonder Story"
Our developed "Wonder Story" is an innovative approach that fundamentally solves these challenges and realizes the democratization of education.
Feature 1: Personalization through Real-time Emotion Detection
Technical Implementation
- High-precision facial expression analysis using TensorFlow
- Real-time detection of 7 types of emotions (joy, sadness, anger, surprise, fear, disgust, neutral)
- Dynamic adjustment function of story progression based on children's reactions
Feature 2: AI Agent-Driven Scalable Model
Core Innovation
Traditional educational material production was a time-consuming and costly process involving planning, writing, editing, proofreading, and testing by specialized staff. In Wonder Story, Google Gemini AI executes this instantly.
Cost Reduction Effects
- Educational material production costs reduced to about 1/50th of conventional methods
- Production period shortened from 3 months to a few seconds
- Service can be provided to unlimited users with a single system build
Feasibility of a Basic Free Model
- Sustainable free provision of basic functions is possible due to nearly zero-cost AI generation, thanks to Google
- Economically achieving equality of educational opportunity
Feature 3: Continuous Learning and Insight Provision Tailored to Growth
Individual Growth Recording System
- Automatic recording and analysis of daily emotional changes and learning progress
- Automatic optimization of content according to age and developmental stage
- Provision of detailed insights and advice for parents and educators
Feature 4: Safe and Secure Design with a Focus on Privacy
Technical Privacy Protection
- All image processing is executed locally (within the browser)
- Immediate disposal of camera image data (no transmission to servers)
- Full compliance with strict child data protection policies
Peace of Mind for Parents
- Zero risk of external data leakage
- Use in an environment where the child's privacy is fully protected
- Clear indication of a highly transparent data usage policy
Through these features, Wonder Story realizes the "Democratization of high-quality personalized education," which was difficult to achieve with conventional educational services, and creates a world where all children can achieve their own unique learning and growth regardless of their economic situation.
2. System Architecture and Technology
Overall System Design
Wonder Story adopts a modern microservices architecture designed with future scalability in mind.
Key Technology Stack
Frontend @ Firebase Hosting
- React 19.1.0 + TypeScript: Combining high development efficiency and type safety
- React Router DOM v7.6.2: High-speed navigation for Single Page Applications
- Context API: Consistent user experience through global state management
AI Engine @ Gemini & TensorFlow
- Google Generative AI (Gemini API): Generation of stories, images, and narration
- TensorFlow.js: Real-time emotion detection on the browser (No API used to ensure privacy)
Backend @ Cloud Run
- FastAPI + Python: High-speed response through asynchronous processing
- Firestore: Features were reduced for the demo version due to maintenance costs
3. Future Vision and Ecosystem Strategy
Evolution into an AI Agent for Total Life Support
Wonder Story is the first step toward a comprehensive AI agent ecosystem that supports a person throughout their life. It provides continuous support at each life stage and achieves even more precise personalization as data accumulates.
Phase Deployment Plan
- Phase 1 (Current): Early childhood education support
- Phase 2: School-age learning management
- Phase 3: Adulthood career support
- Phase 4: Senior life support
4. Demo Video & Feature Introduction
Story Settings
Create your own profile and go on an adventure with friends in your favorite world!
Choose your name, age, and favorite things, and a special story will begin.
Story
It reads children's emotions in real-time and automatically creates an original picture book tailored to each individual. By generating it in the background, the system allows them to enjoy the story immediately with almost zero wait time. Furthermore, through detailed production such as automatic page transitions, it provides an experience that children can get lost in. Completed picture books can also be saved.
Dashboard
Visualize your child's learning history and what emotions they learned on which pages.
It records which emotions your child experienced in which scenes of the story and displays the trajectory of their growth in an easy-to-understand way for parents.
Facial Expression Recognition
A mode where they can play while looking at their own facial expressions helps nurture emotions further.
By reading the story while looking at their own face on the screen, children naturally learn the connection between facial expressions and emotions, acquiring rich emotional expressiveness.
Everyone's Adventure
A feature where you can register stories you generated to a store (currently intentionally disabled).
We are preparing a feature where original picture books created by children can be posted to the store and shared with other children.
We also plan a service where picture books created digitally can be printed and purchased as physical books.
Even professional authors can easily create and register picture books using the picture book creation mode.
It is a creative tool that even professional picture book authors can use intuitively, allowing them to easily produce and register new picture books.
5. Social Impact and Future Development
Contribution to Closing the Educational Gap
Improving Accessibility
- Removal of economic barriers through a basic free model
- International expansion through multi-language support
- Bridging the technology gap through support for low-spec devices
Measurability of Educational Effectiveness
- Visualization of emotional and learning data
- Evidence-based educational improvement
- Verification of effectiveness through collaboration with research institutions
Conclusion: The Future of Education Opened by AI Agents
Wonder Story is not just a picture book app. It is a project that directly addresses the social issue of educational inequality as the first step toward an AI Agent Ecosystem that provides total life support.
Our goal is to democratize education through the power of technology and realize a society where all children can grow in their own way. Through an AI agent-driven approach, we provide scalable and sustainable educational support that goes beyond the traditional model dependent on human resources.
Our journey, starting with Wonder Story, will ultimately develop into a comprehensive ecosystem that supports individual growth and learning at every stage of life, from early childhood to old age.
AI walking alongside humans and maximizing each individual's potential. We will continue to move forward toward the realization of such a future.
In my experience as a rabbit, the most innovative projects often begin by questioning the "obvious." "It is obvious that all children should receive high-quality education" — realizing this standard through the power of technology is our mission.
Sources
[1] "The Reality of Educational Inequality 2025"
[2] "Educational Inequality and School Challenges"
We Overtake Benesse
Discussion