mindbridge

MindBridge is an AI-powered mental wellness companion that gives every student in India a safe, judgment-free space to talk about how they feel  at any hour, for free, with zero friction.

Comment

🧠 MindBridge

AI Mental Health Companion for Students

🔗 Live App: https://mindbridge-psi-nine.vercel.app 🏁 Hackathon: AI For Good Hackathon 2026 👤 Built by: Anirudh Lohiya (Solo, 48 hours)

🚀 Elevator Pitch

MindBridge is an AI-powered mental wellness companion that gives every student in India a safe, judgment-free space to talk — anytime, completely free, with zero friction.

Built on Meta LLaMA 3.1 and powered by Groq, it begins with a simple mood check-in and evolves into a warm, CBT-based conversation that helps students understand their emotions and build coping strategies.

When signs of crisis appear, it automatically surfaces verified Indian helpline numbers — because some situations need more than AI.

50 million Indian students struggle silently every year. MindBridge is the bridge between suffering alone and getting real help.

💡 Inspiration

I’ve seen friends struggle with exam stress, anxiety, and loneliness — often silently, because of stigma around mental health in India.

When I learned that over 50 million students face mental health challenges but fewer than 10% seek help, I knew I had to build something meaningful.

“No student should feel alone at 2 AM during exam season with nobody to talk to.”

⚙️ What It Does

MindBridge provides a simple, frictionless support experience:

  1. No signup — open and start instantly
  2. Select your mood from 6 emotional states
  3. AI begins a warm, empathetic conversation
  4. Identifies stress triggers
  5. Offers CBT-based coping techniques
  6. Detects crisis signals and shows helpline support
  7. Continues conversation as long as needed

🛠️ How I Built It

Layer Technology
Frontend React + Vite
Backend FastAPI (Python)
AI Model LLaMA 3.1 (8B) via Groq API
Deployment Vercel + Render

The hardest part wasn’t coding — it was making the AI feel human, warm, and safe, instead of robotic or clinical.

⚠️ Challenges I Ran Into

  • CORS issues in GitHub Codespaces
  • Balancing empathy vs. safety boundaries
  • Designing a calm, non-clinical UI
  • Fine-tuning crisis detection without false positives

🏆 Accomplishments

  • Built and deployed a fully working app in 48 hours (solo)
  • Implemented real-time crisis detection + helpline support
  • Designed a UI users described as “safe to open up in”
  • Achieved zero signup friction (instant access)
  • Made it completely free and accessible

📚 What I Learned

  • Prompt engineering can completely transform LLM behavior
  • Mental health AI requires strict ethical boundaries
  • Free-tier deployment can support real-world apps
  • CBT techniques can be embedded naturally in conversation

🔮 What’s Next

  • Mood tracking dashboard
  • Regional language support (Hindi, Marathi, Tamil)
  • College counselor integration
  • Anonymous peer support groups
  • Offline-first PWA support

🔗 Links

❤️ Built for AI For Good Hackathon 2026

Solo project — designed and shipped in under 48 hours.

Built With

Share this project:

Updates