Inspiration
When we first got into university, we quickly realized that most of learning happens when you're alone with your books.
That sounds fine until you're staring at a difficult topic like Quantum Mechanics, Advanced Mathematics, or Data Structures and wondering:
"How exactly am I supposed to explain this to myself?"
We used tools like ChatGPT and other AI assistants, and while they were helpful, they weren't designed to be true learning companions. We constantly had to remind them where we stopped, re-upload context, and become experts at prompt engineering just to get good educational responses. On top of that, many students around the world cannot afford multiple premium subscriptions.
For a long time, we wanted to build a platform where students could learn naturally, ask questions without fear, upload their own study materials, and continue learning exactly where they left off.
When we heard about the Google Rapid Agent Hackathon, we finally had the opportunity to turn that vision into reality and build something that could help not only ourselves, but students and lifelong learners everywhere.
What it does
EduAgent AI is an intelligent tutoring platform that transforms study materials into a personalized AI learning experience.
Users can:
- Chat with an AI tutor about any topic
- Upload study materials such as PDFs, notes, presentations, and images
- Ask the AI to explain concepts in simple terms
- Generate personalized study plans
- Continue previous learning sessions from any device
- Save and organize study conversations
- Learn from both uploaded materials and general knowledge topics
- Receive contextual tutoring instead of generic chatbot responses
EduAgent AI acts less like a search engine and more like a personal teacher that remembers your learning journey.
How we built it
We built EduAgent AI using:
Next.js 15 React TypeScript Tailwind CSS Supabase Authentication Supabase Database Google Gemini* **Elasticsearch Vercel
The platform combines conversational AI, persistent session storage, authentication, and intelligent document processing into a unified tutoring experience.
We implemented:
- Google OAuth authentication
- Session-based chat persistence
- Multi-session learning workspaces
- Responsive mobile and desktop interfaces
- AI-powered tutoring workflows
- Search and retrieval architecture using Elasticsearch
- Cloud deployment infrastructure for scalability
Challenges we ran into
This project pushed us technically in several areas.
Some of our biggest challenges included:
- Building reliable chat persistence across devices
- Integrating authentication with seamless session management
- Managing conversation memory without losing context
- Connecting Gemini with our tutoring workflows
- Handling deployment issues and production builds
- Designing a mobile experience that felt like a modern AI product rather than a traditional school portal
- Creating an architecture capable of supporting future document understanding and multimodal learning
We spent significant time debugging authentication flows, session synchronization, deployment configurations, and AI response pipelines.
Accomplishments that we're proud of
We're especially proud that we transformed an idea from our own educational frustrations into a working product.
Some highlights include:
- Building a functional AI tutoring platform from scratch
- Successfully implementing persistent learning sessions
- Creating a clean, modern AI-first user experience
- Integrating Google Gemini into a real educational workflow
- Implementing authentication and user management
- Designing a scalable architecture for future growth
- Deploying a working cloud-based application that can be accessed from anywhere
Most importantly, we're proud that EduAgent AI solves a problem we personally experience as students.
What we learned
This project taught us much more than just coding.
We learned:
- How to architect AI-powered applications
- Best practices for authentication and session management
- How retrieval systems improve AI experiences
- The importance of designing for real users rather than just building features
- How to deploy and manage production-ready applications
- How to collaborate effectively under hackathon time constraints
We also learned that educational AI becomes significantly more powerful when it remembers the learner's journey instead of treating every conversation as a fresh start.
What's next for EduAgent AI
This is only the beginning.
Our roadmap includes:
- Advanced document tutoring for PDFs, presentations, and notes
- Image and screenshot understanding
- Voice-based tutoring conversations
- AI-generated quizzes and flashcards
- Personalized learning paths
- Learning analytics and progress tracking
- Multi-language support
- Adaptive tutoring based on learning style
- Collaborative study spaces
- Mobile applications for Android and iOS
Our long-term vision is to build an AI learning companion that makes high-quality personalized education accessible to anyone, anywhere in the world.
Built With
- elasticsearch
- gemini
- next.js
- react
- supabase
- tailwindcss
- typescript
- vercel

Log in or sign up for Devpost to join the conversation.