G-SAN: Government Services Agent for Navigation

conceptual workflow

Inspiration

India has hundreds of central and state government schemes — scholarships, pensions, subsidies, healthcare benefits, and more. But for most citizens, especially in rural areas, figuring out which ones they are eligible for and how to apply is confusing and often frustrating. The information is scattered across portals, buried in complex rules, and rarely available in local languages.

As three college students, we asked ourselves: what if there was a simple way for any citizen to just tell their situation, and get clear, step-by-step guidance in their own language? That question inspired our idea for the Government Services Agent for Navigation or G-SAN, for short.

What it does (our vision)

G-SAN would:

Take a citizen’s profile (age, income, occupation, location).
Match them with relevant government schemes.
Provide guidance in local languages on how to apply.
Generate a simple summary (via SMS/WhatsApp/PDF).
Track applications and send status updates.

In short, it acts like a personal guide for government benefits, ensuring citizens don’t miss out on support they deserve.

How we plan to build it

We don’t have a working prototype yet, but we’ve mapped out how it could be built:

Granite LLMs → handle multilingual conversations and Q&A.

Agent Development Kit (ADK) → create multi-step workflows (profile → eligibility mapping → guidance → notifications).

Data preprocessing tools → structure scheme eligibility rules in a machine-readable way.

Challenges we expect

Government schemes have complex and varied eligibility criteria.
India’s language diversity makes it hard to cover all users.
Many portals are not standardized, which will make automation tricky.

What we’re proud of so far

Identifying a real and relatable pain point for millions of people.
Designing an end-to-end workflow that uses AI agents, not just simple chat.
Learning how IBM’s Granite + ADK could be applied to a socially meaningful use case.

What we learned (as a student team)

How to think about agentic AI beyond chat — focusing on workflows and automation.
The importance of accessibility and inclusivity when designing technology for the public.
How to take a huge, complex problem and break it down into an achievable first step.

What's next for G-SAN

Start prototyping a demo version that can handle a few central government schemes.
Expand to state-level schemes and more languages.
Explore simpler versions (like offline or IVR support) for citizens without smartphones.

Even though we’re still in the idea stage, this project has already helped us think deeply about how AI can solve India-centric challenges. We’re excited to take the next steps and turn this concept into a working solution.