LLM Arena_Thundercats

Inspiration

The idea for RightoStudy was born out of witnessing how how many international students in the U.S. face complex, often opaque immigration rules, legal exploitation, and abrupt visa revocations. Inspired by real-world cases and stories-including friends and students whose lives were upended-we wanted to build a tool that could help navigate these challenges with clarity, confidence, and access to trusted support.

With rising political scrutiny, inconsistent enforcement of student visa laws, and the recent spike in revocations, it felt urgent to create something that is not only helpful but potentially life-changing.

What it does

RightoStudy is an AI-powered assistant for international students in the U.S. that:

Answers visa-related questions with plan-language explanations
Analyzes uploaded documents (e.g., SEVIS letters, contracts, legal notices) and highlights key information -Provides step-by-step guidance for high-stress situations (e.g., Notice of Intent to Terminate, ICE interactions) -Suggests resources like a nearby legal aid, university contacts, and embassy hotlines -Operates through a clean, privacy-focused chat interface with optional crisis mode

How we built it

We also aggregated a mini-database of immigration FAQs, U.S. legal policies (e.g., INA 212(a)(3)(C)), and institution-level visa data to power our retrieval-augmented generation (RAG) system.

Challenges we ran into

Data availability: Comprehensive and up-to-date student visa datasets were difficult to find due to privacy and policy restrictions. We had to rely on aggregating partial public reports.
Legal sensitivity: Ensuring that we did not cross the line into offering legal advice required careful phrasing and disclaimers.
Model hallucination: LLMs occasionally generated inaccurate or exaggerated responses. We tackled this by refining prompts and adding document-grounded retrieval.
UX under stress: Designing a user interface that feels calm, empowering, and safe—especially for students in crisis—was more challenging than expected.

Accomplishments that we're proud of

Developed a working RAG pipeline that pulls legal clauses directly into student Q&A answers
Enabled document parsing + explanation in real-time for complex legal text
Created a focused, empathy-driven UI for vulnerable populations
Built an MVP that balances automation with clear calls to human resources when needed

What we learned

That tech can be truly impactful when rooted in empathy and a real-world problem
The power of careful UX design in helping users feel seen, supported, and less overwhelmed
The ongoing tension between automation and human oversight—especially in legal contexts

What's next for RightoStudy

Expand document understanding to include school policies, leases, and employment contracts
Multilingual support, starting with Arabic, Hindi, Spanish, and Mandarin
Voice-to-text and mobile support to improve accessibility
Legal aid partner integration so students can escalate serious cases to real professionals
Data feedback loop to improve LLM responses and detect emerging visa issues

RightoStudy is just the beginning—we hope to scale this into a broader platform for immigrant rights and student protections, worldwide.