The inspiration for JanaVani came from seeing how difficult it is for citizens to access government information due to language barriers, confusing websites, and long processes. I wanted to build something that allows people to simply speak or type in their own language and instantly get clear information about schemes and public services. Vernacular AI felt like the right solution to make digital governance more accessible.

I built JanaVani as an AI-driven conversational system where users can ask questions through voice or text. The system uses multilingual NLP to understand queries, detects intent, and generates responses using an LLM. I connected it with structured data about government schemes and service workflows. The prototype interface was designed to work on web or WhatsApp. The core idea is simple: convert complex procedures into easy, conversational steps.

Through this project, I learned how Vernacular AI can reduce digital inequality, how to create user-friendly flows for non-technical users, and how to structure large amounts of government data. I also gained experience in prompt engineering, multilingual design, and working under time pressure.

Some challenges I faced included handling accurate translations, simplifying lengthy government processes, ensuring reliable responses, and maintaining consistency across different languages and queries.

Built With

  • base44
  • vibecoding
Share this project:

Updates