LegalMind AI – Your AI-Powered Legal Research Assistant

LegalMind AI – Transforming legal research with LLMs. Ask, analyze, and get answers—fast, accurate and lawyer-friendly

Inspiration

Legal knowledge is often locked behind paywalls, jargon, or requires expert consultation. We were inspired to build something that empowers everyday people and small businesses to access legal insights easily—just by asking questions naturally.

What it does

LegalMind AI helps users:

Ask legal questions and receive LLM-powered, context-aware answers.
Upload contracts or legal documents to extract summaries and detect potential risks.
Search through legal case laws and statutes using semantic understanding.
Interact in a chat interface that remembers past queries and references.

How we built it

Frontend: React.js + Tailwind CSS
Backend: Flask + OpenAI GPT-4 APIs
Document Parser: PyMuPDF and PDFplumber
RAG Stack: LangChain + FAISS
Hosting: Vercel (frontend), Render (backend)
Authentication: Firebase

Challenges we ran into

Designing accurate prompts for legal use-cases.
Managing large PDF contract context within LLM limits.
Ensuring the AI output remains unbiased and clearly states limitations.
Fine-tuning search results using embeddings and FAISS.

Accomplishments that we're proud of

Successfully built a working contract analyzer and Q&A system.
Integrated vector-based retrieval to enhance LLM responses.
Delivered a clean and accessible UI for non-technical users.

What we learned

The importance of Retrieval-Augmented Generation (RAG) in real-world use.
Balancing UX design with trust and transparency in legal AI.
Managing LLM context limits and optimizing embedding queries.

What's next for LegalMind AI – Your AI-Powered Legal Research Assistant

Integrating a citation engine that links to verified case laws or statutes.
Supporting jurisdiction-based filtering (e.g., U.S. vs. India).
Adding voice input and multilingual support.
Partnering with law students and legal aid services for beta testing.

Built With

chroma
faiss
fastapi
firebaseauth
flask
google-cloud
gpt-4
hugging-face-inference-api
javascript
langchain
llama3
openai
openai-api
pdfplumber
pymupdf
python
react.js
render
tailwindcss
vercel

Submitted to

Qloo LLM Hackathon

Created by

I developed the backend authentication system and integrated the AI model for legal query understanding. It was my first time combining NLP with React and Flask APIs and I gained a deep understanding of secure login flow, user session handling and deploying AI services. I also worked on connecting the frontend with backend endpoints using fetch requests.

Arnab Chakraborty
As a core contributor to LegalAid Genie, I led the integration of voice-enabled functionality using OpenAI's Whisper API, enabling users to interact with the assistant through natural speech in their native languages. I also played a key role in developing the multilingual pipeline using Hugging Face translation models to ensure accurate and culturally appropriate responses. On the backend, I helped design the Retrieval Augmented Generation system using LangChain and FAISS to deliver context-aware legal insights from localized datasets.

Trishita Maity
Gracy Kumari
Sohan Saha