ECHO: The AI Tutor

Inspiration

Most education software only cares if you get the answer right or wrong. But what if you got the right answer for the wrong reasons? We realized that true learning happens when you understand the logic behind a concept. We wanted to build an AI tutor that doesn't just grade you, but actually looks at how you think, catches bad habits early, and lets you safely stress-test your ideas.

What it does

ECHO is a next-generation predictive learning platform. Its core features include:

• Reasoning X-Ray & Future Mistake Forecast: The AI analyzes your problem-solving logic and predicts misconceptions you haven't even made yet.
• The Mistake Lab: A safe sandbox where you can test your hypothesis against edge cases (like adding air resistance or changing mass).
• Multiplayer 3D Simulations: Interactive physics environments (like our Particle Collider) where you can collaborate in real-time with peers.
• Voice-Enabled Copilot: A Socratic AI assistant you can actually talk to using your microphone, which speaks its answers back to you.

How we built it

We built the frontend using React, TypeScript, and Tailwind CSS for a sleek, modern, dark-mode UI.

• For the 3D environments, we used Three.js (via React Three Fiber).
• To make the simulations multiplayer, we set up a custom Node.js/Express backend with Socket.io for real-time WebSocket syncing.
• For the voice interactions, we tapped into the native Web Speech API for instant Speech-to-Text and Text-to-Speech capabilities.
• The brain of the operation is powered by the Gemini API, which handles the complex reasoning, X-Ray analysis, and Socratic tutoring.

Challenges we ran into

One of the biggest challenges was getting the real-time multiplayer 3D simulation to work perfectly. Syncing physics parameters (like particle count and kinetic energy) across multiple browsers without noticeable lag took a lot of fine-tuning with WebSockets. We also had to spend a lot of time perfecting the UI layout so that complex AI data (like the Mistake Forecasts) looked clean, readable, and not overwhelming.

Accomplishments that we're proud of

We are incredibly proud of how "alive" the platform feels. When you are moving your mouse in a 3D simulation, seeing your friend's cursor pop up on screen in real-time, and simultaneously talking out loud to the AI Copilot—it feels like a massive leap forward for EdTech.

What we learned

We learned a massive amount about real-time networking with Socket.io and managing 3D canvas states in React. Most importantly, we learned how powerful AI can be when you prompt it to analyze logic and reasoning rather than just asking it to spit out facts.

What's next for ECHO

Next, we want to add user authentication and cloud databases (like Firebase) so students can track their learning streaks over time. We also plan to build out a full "Teacher Dashboard" so educators can see a bird's-eye view of the most common mistakes their entire classroom is making.

Built With

• express.js
• gemini-api
• google-ai-studio
• lucide-icons
• node.js
• react
• socket.io
• tailwindcss
• three.js
• typescript
• vite
• web-speech-api
• websockets