Inspiration BharatWasi was inspired by the need for an AI mediator that understands and respects the complex, relationship-driven social contracts and hierarchy of Indian culture. What it does It analyzes unstructured audio/text of a local dispute (family, landlord, neighbor, workplace) and generates a Root-Cause Graph, Cultural Reasoning (Samajh), and Three Practical Resolution Strategies (Samadhan).

How I built it The project was built using a Python Flask backend and a TailwindCSS/JavaScript frontend, powered by the Google Gemini 1.5 Flash API with strict JSON schema prompting to ensure structured, predictable advice.

Challenges I ran into The primary challenge was prompt engineering to prevent generic advice and ensure the AI consistently embedded deep, specific Indian cultural concepts (Izzat, Log Kya Kahenge, Badappan) into its analysis and strategies.

Accomplishments that I'm proud of I am most proud of creating a functional JSON-driven cultural AI engine that successfully translates emotional, qualitative data into objective, actionable advice that is both technically complex and socially nuanced.

What I learned I gained deep expertise in using Gemini's structured output mode and learned to leverage large language models to bridge the gap between advanced technology and highly specific, sensitive sociocultural mediation.

What's next for BHARATWASI Future plans include integrating Speech-to-Text for direct Hindi/Hinglish audio input and developing templates for formal legal/social documentation based on the generated resolution strategies.

Built With

Share this project:

Updates