JanaSetu-AI
A multilingual AI benefits agent that links scheme rules to citizen evidence and prepares explainable, human-reviewed next steps.
Built at AgentX Hackathon 2026
Team CHAKRAVYUHA: Putta Shankar Prasad Y N, Raksha Bhat.L, RAHUL GOWDA S CSE
Built with Next.js, React, TypeScript, Supabase, PostgreSQL, Web Crypto API, IndexedDB, Vercel, Playwright and ClamAV tooling.
About the project
Citizens across India often discover government welfare schemes but struggle to navigate fragmented portals, complex conditions, and required paperwork. Generic chatbots fail to preserve the relationship between official rules, citizen evidence, and concrete preparation steps.
JanaSetu AI is an India-wide, multilingual benefits-readiness workspace designed to bridge this gap. Supporting English, Hindi, Kannada, Telugu, Tamil, and Malayalam, it enables citizens to organise reusable profiles, explore source-linked schemes, inspect preliminary eligibility, identify document gaps, and prepare application drafts.
Core Innovation: The Benefit Readiness Graph
The core innovation is an inspectable graph connecting each scheme condition to:
Official sources and review statuses.
Citizen-stated facts and evidence states.
Identified information gaps and recommended next actions.
The agent-oriented workflow coordinates scheme retrieval, deterministic rule evaluation, evidence mapping, and draft preparation. Crucially, unknown information remains marked as unknown rather than being hallucinated by generative models, ensuring explainable, reviewable outputs.
Architecture & Privacy Boundaries
Built using Next.js, React, TypeScript, and Supabase/PostgreSQL, JanaSetu AI prioritises local-first privacy:
Local Storage & IndexedDB: Manages documents directly in the browser with Web Crypto encrypted vault export/import.
Non-Government Scope: Outputs are preparation aids, not official government submissions or legal eligibility decisions. External actions require authorized departmental channels.
Current Scope & Target Users
The working public preview includes 142 catalogue references, with four rule-screened prototype schemes awaiting independent review. It does not claim full verified scheme coverage, live department API integration, DPDP certification, or automated submission; final decisions rest with the responsible authorities.
Target Users: Students, farmers, women, senior citizens, and households across India seeking public welfare benefits.