Synopsis Presentation on
AI-Lawyer
Presented by
Yash Raj Sharma(202216024)
Anish Acharya(202216041)
Maanas Tiwari(202216034)
Under the Guidance of
Under the Guidance of
Mr.Gaurav Pradhan, Assistant Professor, CA Department
Dept. of Computer Application, SMIT, Majhitar, East Sikkim
Introduction
• Indian Legal system backlashes critical accessibility challenges -
• 70% of Indians cannot afford legal representation, making justice inaccessible.
• Complex legal terms alienates non-English speakers, especially in rural areas.
• 50M+ pending cases lead to decade-long judicial delays.
Solution: AI-Powered Legal Assistance
Built on Indian-LawyerGPT framework, leveraging Falcon-7B & Llama2 fine-tuned on Indian legal
datasets with (4bit - quantisation) QLoRA (reduce computational cost) for efficiency and
maintaining accuracy.
Key Features:
Legal Analysis & Assistance – Uses InLegalBERT, trained on 5.4M+ legal documents for precise legal
insights.
Document Interpretation & Generation – Automates legal drafting with accuracy.
Multilingual Support – Enables access to justice in regional languages.
Regulatory Compliance – Generates standardised legal documents adhering to emerging
regulations such as the Digital Personal Data Protection Act, 2023.
Dept. of Computer Application, SMIT, Majhitar, East Sikkim
Aim and Objectives :
Develop an AI-powered legal assistant that provides multilingual legal
assistance, understands legal queries, and fetches relevant laws or case
references using AI & NLP, bridging the gap between individuals and legal
professionals.
Objectives:
• NLP Model Training: Fine-tune Indian legal datasets for legal Q&A and
judgment prediction (e.g., Aalap, Mistral 7B) .
• Multilingual Chatbot with Conversational interface : Support English,
Hindi, and extend to Tamil, Marathi, etc. using LangChain & FAISS for
semantic search .
• Legal Database Integration: Connect to Indian Kanoon for real-time case
law and statute updates. This integration guarantees the platform will remain
current with evolving legal standards.
• Ethical AI & User Feedback: Implement expert reviews and bias mitigation
for transparency .
• Document Analysis & Compliance: Use of OCR (Tesseract) for digitisation,
Dept.detection,
AI-driven clause of Computer Application,
and DPDP SMIT,Act
Majhitar, East Sikkim
compliance
Problem Definition
The Indian legal system spans over diverse legal domains, including family,
property, labor, and criminal laws. Yet, a huge portion of the population remains
in the dark unaware of their legal rights and entitlements. This lack of legal
literacy limits access to justice and prevents people from fully utilising the legal
provisions designed for the benefit of them.
1.High Costs: Legal advice is unaffordable for low-income groups .
2.Complex Documentation: Manual reviews delay case resolutions .
3.Language Exclusion: Rural populations lack Hindi/regional resources .
4.Compliance Gaps: Startups risk penalties from outdated contracts .
General Overview of Problem
Overview
• AI-powered legal assistant providing quick, accessible legal insights.
• Analyzes user queries on rights, contracts, and legal procedures.
• Bridges the gap between individuals and legal professionals.
Key Features
Understands Legal Queries – Analyzes user questions & legal conte
Finds Relevant Information – Fetches laws, case studies & referenc
Provides Basic Guidance – Offers general legal insights (not a lawye
replacement).
User-Friendly – Designed for all, no legal expertise needed.
Privacy & Ethics – Ensures data security & ethical AI use.
Empowering users with simplified legal access & guidance!
Dept. of Computer Application, SMIT, Majhitar, East Sikkim
Proposed Solution Strategy
omain-specific models (e.g., Aalap, InLegalBERT) to boost legal
R to digitize paper documents and apply NLP for clause detectio
ploy a multilingual chatbot with semantic search and context-aw
s: Develop a continuous monitoring module with predictive ana
ces: Connect to Indian legal databases (e.g., Indian Kanoon) fo
rporate legal expert evaluations to ensure ethical, transparent
Feasibility Study
Technical Feasibility: Proven success of domain-
specific models like InLegalBERT and RAG-based
legal chatbots confirms the effectiveness of
advanced NLP in Indian legal applications.
Economic Feasibility: Automation of routine legal
tasks reduces legal fees by 60–80%, making legal
services more affordable.
Operational Feasibility: Collaboration with legal
experts ensures accurate, practical, and reliable
outputs.
Legal Feasibility: Strong data privacy measures
and compliance checks align with the DPDP Act,
2023, ensuring legal adherence.
Literature Survey
AI-powered legal assistants transform legal research, compliance, and
accessibility through NLP, OCR, and machine learning. Models like
InLegalBERT and LawyerGPT enhance legal text interpretation and
case law retrieval using Indian legal datasets . RAG-based
approaches further improve legal text analysis .
To address India's linguistic diversity, multilingual legal chatbots
powered by Aalap and Mistral 7B provide legal assistance in Hindi,
Tamil, Marathi, and more, while LangChain and FAISS optimize
semantic search for better contextual responses. AI-driven contract
review and compliance verification leverage OCR to detect
missing clauses and regulatory risks, ensuring compliance with laws
like the DPDP Act, 2023.
However, challenges like ethical concerns and bias necessitate
human-in-the-loop oversight to ensure fairness, transparency, and
accountability . Future research should focus on expanding regional
language support, refining risk analytics, and strengthening
AI ethics frameworks .
Project Plan
• Team Structure
GUIDE
Mr. Gaurav Pradhan
Assistant Professor I
Dept. of Computer
Application,
SMIT
Yash Raj Sharma Anish Acharya Maanas Tiwari
202216024 202216041 202216034
Gnatt Chart
Duration Of the Project
ACTIVITY 16/01/2025 07/02/2025 01/03/2025 29/03/2025 19/04/2025
to to to to to
07/02/2025 01/03/2025 29/03/2025 19/04/2025 26/04/2025
Literature Survey
Problem
Identification
Design
Implementation
Testing &
Validation
Documentation
Red : To be completed ; Green : Achieved
Conclusion
The AI-Powered Legal Assistant is designed to bridge the gap
between citizens and legal professionals by offering free, AI-
driven legal assistance. By automating legal research, document
analysis, and compliance monitoring, the platform will ensure
affordable and accessible justice for all. Future improvements
will focus on expanding regional language support, enhancing
predictive analytics, and strengthening AI ethics, contributing to
a more inclusive and efficient legal system
