LegalFlow

Inspiration

We built LegalFlow after realizing how time-consuming and scattered legal research still is. Lawyers often have to search through multiple documents, acts, and case files, while general AI tools often provide answers that aren't backed by reliable legal sources.

During the hackathon, we wanted to build something that solves a real problem instead of another chatbot. Our goal was to make legal research faster, more reliable, and accessible while keeping every answer grounded in authentic legal documents.

What We Built

LegalFlow is an AI-powered legal intelligence platform for both the public and legal professionals.

It provides two modes:

• General Mode – Ask legal questions and receive answers with citations from indexed legal documents.
• Lawyer Mode – Upload case details and receive structured multi-domain legal analysis.

Features

• Admin PDF upload and indexing
• RAG-based legal Q&A
• Multi-agent analysis for Criminal, Civil, Corporate, and Tax Law
• Conflict detection
• Structured legal reports
• Client workspaces
• Knowledge graph visualization
• Voice input and text-to-speech
• English, Hindi, and Marathi support

How We Built It

Tech Stack

• React + Vite
• Node.js + Express
• MongoDB
• Pinecone
• Cloudinary
• Groq LLMs
• Google Gemini

Workflow

1. Upload legal PDFs.
2. Extract, chunk, and embed documents.
3. Store embeddings in Pinecone.
4. Retrieve relevant context.
5. Multi-agent analysis.
6. Generate grounded responses with citations.

Challenges We Faced

• Keeping AI responses grounded in legal documents.
• Designing a scalable multi-agent architecture.
• Supporting scanned PDFs using OCR.
• Building multilingual and voice capabilities within the hackathon timeline.

What We Learned

• Building reliable RAG systems.
• Designing specialized AI agents.
• Working with vector databases and semantic search.
• Creating AI products users can trust.

Future Scope

• More legal domains.
• Better citation extraction.
• Improved OCR.
• Mobile applications.
• Team collaboration for law firms.
• Integration with live legal databases.

Built With

• cloudinary
• exceljs
• express.js
• google-gemini
• groq
• jwt
• langchain
• mongodb
• mongoose
• node.js
• pdfkit
• pinecone
• react
• react-force-graph-3d
• react-router
• tailwind-css
• tesseract.js
• three.js
• vite