Home
Every citizen has a right (haq) to the welfare they're entitled to. We help them find and claim it.
YojanaSaathi is an agentic AI system that matches citizens to the government welfare schemes they qualify for — replacing manual search-and-filter with actual reasoning over eligibility rules. Built for the BharatCares × AICTE × IBM SkillsBuild AI Automation Internship 2026.
| Page | What's There |
|---|---|
| Getting Started | Clone, install, run locally — for new contributors |
| Architecture | System design, agent pipeline, data flow |
| Tech Stack | Every technology used and why |
| API Reference | Full request/response contract for the backend |
| Agent Design | How each of the 5 agents works internally |
| Roadmap | What's shipped, what's next, Phase 2 (mobile) |
| FAQ | Common questions about the project and its decisions |
| Team | Who's building this and what they own |
India runs 1,000+ central and state welfare schemes, but eligibility criteria are scattered across ministries, contradictory in places, and legally dense. The government's own discovery portal is search-and-filter — it doesn't reason about a citizen's actual situation, doesn't catch schemes they've never heard of, and doesn't help with the application itself. Millions of eligible citizens miss benefits simply because no one connects their real circumstances to the right scheme.
Rather than another chatbot wrapper, YojanaSaathi runs a deterministic, auditable rule engine for eligibility decisions — not an LLM guessing yes/no on legal entitlements. AI is used where it adds real value: language generation, application drafting, and friendly explanation — never for the eligibility decision itself. See Agent Design for the full reasoning.
- Live Demo (add link once deployed)
- Report a Bug
- Request a Feature
- Discussions
Maintained by Prem Rajesh Sargara (P.R.S) & Dvij. Last updated as part of the Jul 2026 build cycle.
Every citizen has a right (haq) to the welfare they're entitled to. We help them find and claim it.
YojanaSaathi is an agentic AI system that matches citizens to the government welfare schemes they qualify for — replacing manual search-and-filter with actual reasoning over eligibility rules. Built for the BharatCares × AICTE × IBM SkillsBuild AI Automation Internship 2026.
| Page | What's There |
|---|---|
| [Getting Started](Getting-Started) | Clone, install, run locally — for new contributors |
| [Architecture](Architecture) | System design, agent pipeline, data flow |
| [Tech Stack](Tech-Stack) | Every technology used and why |
| [API Reference](API-Reference) | Full request/response contract for the backend |
| [Agent Design](Agent-Design) | How each of the 5 agents works internally |
| [Roadmap](Roadmap) | What's shipped, what's next, Phase 2 (mobile) |
| [FAQ](FAQ) | Common questions about the project and its decisions |
| [Team](Team) | Who's building this and what they own |
India runs 1,000+ central and state welfare schemes, but eligibility criteria are scattered across ministries, contradictory in places, and legally dense. The government's own discovery portal is search-and-filter — it doesn't reason about a citizen's actual situation, doesn't catch schemes they've never heard of, and doesn't help with the application itself. Millions of eligible citizens miss benefits simply because no one connects their real circumstances to the right scheme.
Rather than another chatbot wrapper, YojanaSaathi runs a deterministic, auditable rule engine for eligibility decisions — not an LLM guessing yes/no on legal entitlements. AI is used where it adds real value: language generation, application drafting, and friendly explanation — never for the eligibility decision itself. See [Agent Design](Agent-Design) for the full reasoning.
- [Live Demo](#) (add link once deployed)
- [Report a Bug](../../issues/new?template=bug_report.md)
- [Request a Feature](../../issues/new?template=feature_request.md)
- [Discussions](../../discussions)
Maintained by Prem Rajesh Sargara & Dvij. Last updated as part of the Jul 2026 build cycle.