Inspiration
Mental wellness is often undervalued, especially in high-pressure environments where people struggle to understand their feelings or find balance. We created Green Minds to offer a safe, intelligent, and deeply supportive space—a digital companion that doesn’t just listen, but understands and helps. With the rise of generative AI, we saw a powerful opportunity to combine language models like Groq (LLaMA 3) and Google Gemini with real psychological value to build a holistic mental health toolkit.
What it does
Green Minds is a full-stack, AI-powered mental health app offering:
📝 Secure multi-user journaling
💬 Dual-model sentiment & emotion analysis (GoEmotions + BERTweet)
📚 Personalized stories, activities, and mental wellness tips
🌐 Multi-language support
🎮 AI-generated cognitive games (riddles, scrambles)
🧘♀️ Mindfulness tools like breathing pacer and body scan
🕉️ Spiritual aid through daily Bhagavad Gita shlokas
📈 Emotional trend tracking with interactive charts
🏆 Gamification via daily streaks and personal scorecards
How we built it
Frontend: Streamlit for rapid UI development and interactivity
Backend: Python with Google Firestore (cloud database & real-time sync)
Sentiment Analysis Models:
GoEmotions for multi-label emotion classification
BERTweet-base-sentiment-analysis for sentiment detection
LLMs:
Groq API (LLaMA 3) as the primary model for content generation
Gemini API as a fallback for reliability
Features:
Text-to-Speech:gTTS for narrating AI-generated stories
Spiritual Feature: Bhagavad Gita REST API integration
Visualization: Plotly & Matplotlib for trend charts
Deployment: Streamlit Community Cloud
Security: Managed via secrets.toml for API keys and credentials
Challenges we ran into
Model selection: Initial BERT model lacked accuracy; switching to fine-tuned BERTweet drastically improved sentiment results
AI reliability: Relying on a single LLM caused outages—we added fallback support using Gemini
Prompt engineering: Designing effective prompts for clean, mood-aligned responses was harder than expected
Streamlit session state: Managing dynamic components like games required careful key and state management
Accomplishments that we're proud of
Successfully integrated two transformer models to deliver accurate emotional insights
Built a robust LLM system with failover logic between Groq and Gemini
Created a truly holistic wellness platform that blends AI, psychology, and spiritual wisdom
Developed a secure, user-friendly, and accessible journaling experience with multi-language support
What we learned
Fine-tuned models outperform generic ones in niche tasks like sentiment analysis
System redundancy (Groq + Gemini) improves application reliability and UX
Psychological tools + spiritual content = a uniquely impactful wellness experience
Streamlit, while simple, can scale to complex multi-page, interactive apps with careful state handling
What's next for GreenMind – AI-Powered Sentiment & Mood Journal
🎙️ Add voice journaling with live transcription
🧠 Integrate therapy-based exercises (e.g., CBT techniques)
🌍 Expand*regional language* support
🤖 Add mood-aware chatbot for interactive conversations
📊 Enhance emotion analytics with weekly/monthly wellness reports
🛡️ Launch on mobile platforms with offline support
Log in or sign up for Devpost to join the conversation.