ai personalized learning path
ai assistant result
video player with ai agent
ai quick qna
ai solve image
ai smart note summary
multi person collaboration
livesession
ai search post summary
ai embedding recommendation

🎓 Studify — AI-Powered Learning Platform

🌟 Inspiration

As students, we realized that education is full of information — but not intelligence.
Modern learners face information overload: multiple classes, notes, part-time work, and endless online resources.
Instead of learning efficiently, students often scroll aimlessly through disorganized materials.

We wanted to fix that — to create a system that not only stores knowledge, but understands and organizes it intelligently.
That’s why we built Studify, an AI-powered learning platform that transforms raw materials into structured, searchable, and reusable knowledge units.

🚀 What It Does

Studify integrates AI into every layer of the learning experience:

🎥 AI Video Assistant

Converts passive watching into active learning.
Students can ask:

“What did the professor just say about Newton’s Law?”
and instantly receive an accurate, timestamped answer.
Powered by dual-model embeddings (384d + 1024d) and parallel retrieval, it achieves over 90% response accuracy.

🧠 AI Summary & Recommendation System

Tackles information overload by summarizing community posts into concise overviews and recommending personalized content using a 60% rule-based + 40% AI-based hybrid model.

💬 Community

An AI-powered social space where students post, chat, and learn from personalized insights instead of random feeds.

🏫 Classroom

A real-time collaborative classroom powered by LiveKit, Liveblocks, and Supabase Realtime, enabling tutors and students to learn, interact, and share assignments anytime.

🔍 AI Content Detection

Ensures academic integrity with a Hugging Face–hosted DivEye model that classifies submissions as human-written, AI-generated, or AI-rewritten — with confidence scores and risk levels.

🌍 Internationalization

Supports English and Chinese, with a scalable multilingual design for future expansion.

🧩 How We Built It

Frontend:

Next.js (App Router)
React 19
TailwindCSS 4
Framer Motion
Radix UI

Backend:

Supabase (PostgreSQL + Realtime)
Upstash Redis
QStash

AI Stack:

LangChain
OpenAI Embeddings
Hugging Face Whisper & DivEye models

Real-time & Collaboration:

LiveKit (WebRTC SFU)
Liveblocks (CRDT Sync)
Supabase Realtime (WebSocket)

Payments:

Stripe Integration

Deployment:

Vercel
Hugging Face Spaces
Capacitor (Android APK)

⚙️ Challenges We Ran Into

Integrating dual embedding models (384d + 1024d) to balance cost and semantic precision.
Building a fully asynchronous pipeline for AI video transcription, segmentation, and embedding.
Maintaining real-time data synchronization in collaborative whiteboards with dozens of active users.
Achieving seamless communication between AI tools, user actions, and backend event triggers.

🧭 What We Learned

We learned how to combine AI reasoning, RAG retrieval, and real-time collaboration into a cohesive platform that actually feels intelligent.
More importantly, we discovered that AI in education shouldn’t replace learning — it should amplify understanding.

🔮 What’s Next for Studify

Expanding our AI Video Assistant to support multilingual understanding.
Improving AI recommendations using reinforcement learning.
Opening the Studify API for schools and institutions to integrate their own materials.
Adding mobile-first interactive dashboards for learners and tutors.