LegalDocSummariser
Link to Live Project: https://github.com/Tanishq2527/legalDoc.git
We built an AI-powered Legal Document Assistant that helps users quickly understand and work with complex legal documents.
Problem:
Legal contracts, policies, and case laws are long, jargon-heavy, and time-consuming to read. Lawyers and businesses waste hours searching for key clauses, risks, or answers.
Our Solution:
An end-to-end AI pipeline that:
-
Ingests documents (PDF, Word, text) and preprocesses them.
-
Embeds & stores chunks in a vector database (FAISS/Pinecone).
-
Uses Retrieval-Augmented Generation (RAG) with Mistral-7B-Instruct to:
-
Summarize contracts into plain English.
-
Answer user questions with citations.
-
Highlight risky or critical clauses.
-
Auto-generate skeleton legal briefs with evidence-backed arguments.
-
-
Frontend (React + Tailwind) for file upload, queries, and results.
-
Backend (FastAPI + LangChain) handles embeddings, retrieval, and orchestration.
Why itβs impactful:
-
β± Saves hours β contracts summarized in minutes.
-
βοΈ Makes law accessible β non-lawyers can understand documents.
-
π° Reduces legal costs β automates first drafts & clause detection.
-
π Increases trust β answers come with real citations from documents.
Future scope:
-
Voice-based search & response.
-
Multi-language legal support.
-
Fine-tuning for specific domains (contracts, compliance, case law).
-
Integrations into enterprise document workflows.
This build was uploaded as a hackathon project