Project Motivation
This project began with a simple observation: many Gujarati-speaking community members struggle to use AI for meaningful religious study. When I saw how difficult it was for loved ones to access trustworthy, vernacular answers from the “Vachanamrut,” I realized there was a gap between powerful technology and lived, cultural experience.
Building this for my family — and especially seeing my mother’s joy when she could ask questions naturally in Gujarati — made the goal clear: bridge culture with technology without compromising faithfulness to the text.
AI should be inclusive, accessible, and trustworthy — this project was my attempt to embody those values.
Technical Overview
System Architecture
Tech Stack
- Next.js App Router
- TypeScript
- Tailwind CSS + shadcn/ui
- Retrieval-Augmented Generation (RAG)
- Gujarati language QA over “Vachanamrut”
- Chunking, embeddings, and context retrieval
- Server Actions & Route Handlers
- Streaming responses for chat UX
- Secure environment variables (server-only)
- Accessibility-first content and contrast
- Mobile-first layout
- Gujarati script readability
About the Developer
I'm Nyal Dhirajlal Kakadia, a B.Tech student at VIT Vellore, passionate about Deep Learning and AI accessibility. I built this project to bridge culture and technology — bringing trustworthy, Gujarati-language answers to the “Vachanamrut.”
“I aim to continue building AI systems that respect linguistic, cultural, and ethical boundaries — a direction I wish to deepen through the MSAII program at Carnegie Mellon.”