💡 Inspiration

The inspiration for MindFlow came from observing a common struggle among students: the boredom and inefficiency of passive studying. We realized that while students have access to information, they lack tools to actively engage with it. We wanted to build a bridge between raw notes and active recall, using Generative AI to create a personal tutor that is fun, instant, and available 24/7.

⚙️ What it does

MindFlow is an intelligent study companion that transforms any topic into three distinct learning modes:

  1. Flashcards: Auto-generated cards with "flip" animations and text-to-speech audio for auditory learners.
  2. Quiz Mode: Gamified multiple-choice questions with instant scoring, feedback, and celebratory effects.
  3. Memory Hacks: A unique feature that generates catchy mnemonics (acronyms and rhymes) to help students memorize complex strings of information.

🔧 How we built it

We built the application using React and Vite for a lightning-fast frontend experience. The user interface was crafted with Tailwind CSS, utilizing a modern "glassmorphism" aesthetic with custom category themes.

The core intelligence is powered by the Groq API (Llama 3 model). When a user enters a topic, we send optimized system prompts to Groq, which returns structured JSON data containing the educational content. This data is then rendered dynamically into our interactive components. We used Framer Motion for fluid animations and the Web Speech API to give the AI a voice.

🛑 Challenges we ran into

  • Structured AI Output: Getting the AI to consistently return valid JSON without hallucinating extra text was difficult. We solved this by refining our strict system prompts and implementing robust error handling.
  • Mobile Responsiveness: Designing complex 3D-flipping cards that look good on mobile screens required careful CSS grid optimizations.
  • Voice UX: The default browser text-to-speech was too fast. We had to implement rate control (0.8x speed) and proper speech cancellation to ensure a smooth auditory experience.

🧠 What we learned

We learned a great deal about Prompt Engineering—how to guide Large Language Models to act as educators. We also deepened our understanding of React state management for handling complex user flows (switching between inputs, games, and results) and the importance of Accessibility features like voice support in educational tools.

Try Now

Website

Built With

Share this project:

Updates