NITISARAL

What Inspired Us India has over 1 billion citizens, yet most government policies are written in dense legal language that the average person cannot understand. We saw farmers unable to understand MSP schemes, students missing out on scholarships, and citizens unaware of their RTI rights — simply because the information was inaccessible. Team Futuristic Four wanted to fix that. What We Built A web-based AI policy simplifier built for Indian citizens, powered by Claude API (Anthropic). Users can paste any government policy text and instantly get:

A plain-language summary in 12 Indian languages (Hindi, Hinglish, Bengali, Tamil, Telugu, Marathi, Gujarati, Punjabi, Kannada, Odia, Malayalam) Audience-specific analysis (farmer, student, business owner, senior citizen) TLDR / Deep / Standard analysis modes Action steps, eligibility checker, and RTI guidance WhatsApp share button for viral spread across India Voice input (Hindi support) and read-aloud output Mobile-first design with offline caching

How We Built It

Frontend: Pure HTML, CSS, JavaScript — no framework, works on low-end Android phones AI: Claude API (Anthropic) for intelligent policy analysis, multilingual output, and bias detection Performance: localStorage caching so repeat queries are instant Design: Dark theme, sticky analyze button, haptic feedback on Android, iOS safe area support

Challenges We Faced

Getting Claude to consistently return structured JSON with all required fields — solved with strict prompt engineering and array safety guards Making the UI work smoothly on cheap Android phones with small screens and slow networks Handling 12 languages in a single prompt without quality degradation Balancing response depth (TLDR vs Deep mode) while keeping API costs low

What We Learned

How to engineer prompts for structured, reliable AI outputs The importance of mobile-first design for Indian users How Claude API handles multilingual tasks far better than expected Real-world deployment with Netlify in under 2 minutes