LegalEase AI - Legal Document Simplification Platform

Making Legal Documents Accessible Through AI

🚀 Overview

LegalEase AI is a comprehensive platform that transforms complex legal documents into plain English using Google's cutting-edge Gemini AI. The platform features document analysis, real-time meeting transcription, interactive AI chat, and voice-enabled features to make legal information accessible to everyone.

🎯 Key Features

📄 Document Analysis: Upload contracts, agreements, or legal documents for AI-powered analysis
🎙️ Real-time Meeting Transcription: Chrome extension for Google Meet with live legal analysis
💬 Interactive AI Chat: Ask questions about your documents with context-aware responses
🧠 RAG-Powered Search: Retrieval-Augmented Generation for intelligent document querying
🔍 Vector Search: Advanced semantic search using embeddings for precise legal clause matching
🗣️ Voice Features: Speech-to-text input and text-to-speech output
⚖️ Legal Risk Assessment: Automatic identification of high, medium, and low-risk clauses
🌍 Multi-language Support: Multiple language options for global accessibility
📊 Smart Embeddings: Document vectorization for similarity search and contextual understanding

🏗️ Architecture

Core Technologies

Frontend: Next.js 15.2.4 with App Router & TypeScript
AI Engine: Google Gemini 2.0 Flash (via @google/genai SDK)
Vector Database: Google Vertex AI Vector Search for enterprise-grade similarity search
Embeddings Model: Google Universal Sentence Encoder & text-embedding-004
RAG Architecture: Retrieval-Augmented Generation for context-aware responses
Authentication: NextAuth.js with Google OAuth
Database: Firebase Firestore with Vertex AI vector index integration
Real-time Communication: WebSocket with Socket.IO
Chrome Extension: Manifest v3 for Google Meet integration
Voice Processing: Google Speech-to-Text & Text-to-Speech APIs
Search Engine: Hybrid search combining Vertex AI vector similarity and traditional indexing

Project Structure

google-hackathon/
├── 📁 app/                          # Next.js app router pages
│   ├── 📁 api/                      # API routes
│   │   ├── 📄 analyze/              # Document analysis endpoint
│   │   ├── 📄 chat/                 # AI chat functionality
│   │   ├── 📄 speech-to-text/       # Voice recognition
│   │   ├── 📄 text-to-speech/       # Voice synthesis
│   │   ├── 📄 transcribe/           # Meeting transcription
│   │   └── 📄 websocket/            # Real-time connections
│   ├── 📄 upload/                   # Document upload interface
│   ├── 📄 dashboard/                # Main dashboard
│   ├── 📄 meet/                     # Google Meet integration
│   ├── 📄 chat/                     # Chat interface
│   ├── 📄 voice-test/               # Voice features testing
│   └── 📄 settings/                 # User settings
├── 📁 chrome-extension/             # Chrome extension for Meet
│   ├── 📄 manifest.json             # Extension configuration
│   ├── 📄 background.js             # Background script
│   ├── 📄 content.js               # Content script for Meet
│   ├── 📄 popup.html/js            # Extension popup
│   └── 📁 build/                   # Built extension files
├── 📁 lib/                         # Utility libraries
│   ├── 📄 document-processor.ts     # Gemini AI document processing
│   ├── 📄 gemini.ts                # AI client configuration
│   ├── 📄 firebase.ts              # Firebase configuration
│   ├── 📄 vertex-vector-store.ts   # Vertex AI Vector Search operations
│   ├── 📄 embeddings.ts            # Text embedding generation and processing
│   ├── 📄 rag-engine.ts            # RAG implementation for contextual retrieval
│   ├── 📄 search-engine.ts         # Hybrid search functionality
│   ├── 📄 languages.ts             # Language support
│   └── 📄 websocket.ts             # WebSocket utilities
├── 📁 components/                   # React components
│   ├── 📄 chat-interface.tsx        # AI chat component
│   ├── 📄 transcript-analyzer.tsx   # Meeting analysis
│   ├── 📄 document-viewer.tsx       # Document display
│   └── 📁 ui/                      # UI components
├── 📁 hooks/                       # React hooks
│   ├── 📄 use-voice-recording.ts    # Voice input hook
│   └── 📄 use-text-to-speech.ts    # Voice output hook
└── 📁 docs/                        # Documentation
    ├── 📄 websocket-integration.md  # WebSocket setup
    ├── 📄 voice-chat-features.md    # Voice features
    └── 📄 google-cloud-transcription.md

🔧 Installation & Setup

Prerequisites

Node.js 18+ and pnpm
Google Cloud API Key (Gemini AI & Embeddings)
Google Cloud Project with Vertex AI enabled
Firebase Project
Chrome Browser (for extension)

1. Clone & Install

git clone <repository-url>
cd google-hackathon
pnpm install

2. Environment Setup

Create a .env file with the following configuration:

# 🤖 Gemini AI Configuration
NEXT_PUBLIC_GEMINI_API_KEY=your_gemini_api_key

# 🧠 Google Cloud Vertex AI & Embeddings
GOOGLE_CLOUD_PROJECT_ID=your_gcp_project_id
GOOGLE_CLOUD_LOCATION=us-central1
VERTEX_AI_INDEX_ENDPOINT=your_vertex_ai_index_endpoint
VERTEX_AI_INDEX_ID=your_vertex_ai_index_id
GOOGLE_EMBEDDINGS_API_KEY=your_google_embeddings_key
GOOGLE_APPLICATION_CREDENTIALS=path/to/service-account.json

# 🔥 Firebase Configuration
NEXT_PUBLIC_FIREBASE_API_KEY=your_firebase_api_key
NEXT_PUBLIC_FIREBASE_AUTH_DOMAIN=your_project.firebaseapp.com
NEXT_PUBLIC_FIREBASE_PROJECT_ID=your_project_id
NEXT_PUBLIC_FIREBASE_STORAGE_BUCKET=your_project.firebasestorage.app
NEXT_PUBLIC_FIREBASE_MESSAGING_SENDER_ID=your_sender_id
NEXT_PUBLIC_FIREBASE_APP_ID=your_app_id

# 🔐 Google OAuth
NEXT_PUBLIC_GOOGLE_CLIENT_ID=your_google_client_id
NEXT_PUBLIC_GOOGLE_CLIENT_SECRET=your_google_client_secret

# 🔑 NextAuth Configuration
NEXTAUTH_SECRET=your_nextauth_secret
NEXTAUTH_URL=http://localhost:3000

# 🗣️ Voice Services
GOOGLE_SPEECH_TO_TEXT=your_gemini_api_key
GOOGLE_TEXT_TO_SPEECH_KEY=your_gemini_api_key

3. Run Development Server

# Start the development server
pnpm dev

# Build for production
pnpm build

# Start production server
pnpm start

Visit http://localhost:3000 to see the application.

🔌 Chrome Extension Setup

Development Installation

Navigate to Extensions: chrome://extensions/
Enable Developer Mode: Toggle in top-right corner
Load Extension: Click "Load unpacked" → select chrome-extension directory
Verify Installation: Extension appears in your extensions list

Usage in Google Meet

Join a Google Meet: Start or join any meeting
Activate Extension: Click the LegalEase icon in your browser toolbar
Connect to Server: Click "Connect to Server" in the popup
Start Transcription: Click "Start Transcription" when ready
View Analysis: Real-time transcript and AI analysis appear on your dashboard

📋 Features Deep Dive

🔍 Document Analysis

Direct AI Processing: Upload PDFs, images, or documents directly to Gemini AI
Vector Embeddings: Automatic document vectorization using Google's text-embedding-004
Smart Text Extraction: Superior accuracy compared to traditional OCR
Semantic Search: Find similar clauses across your document library using Vertex AI vector similarity
Legal Risk Assessment: Automatic identification of risky clauses with severity levels
Plain English Translation: Complex legal jargon simplified for easy understanding
Interactive Q&A: Context-aware chat about specific document content with RAG enhancement
Multi-format Support: PDFs, Word docs, images, and text files
Knowledge Base: Build a searchable Vertex AI vector database of your legal documents

🎤 Real-time Meeting Transcription

Google Meet Integration: Seamless Chrome extension integration
Live Audio Capture: Real-time audio processing during meetings
WebSocket Communication: Instant data streaming to your dashboard
Legal Analysis: AI identifies legal topics and risks in conversations
Privacy-First: Audio processed securely, no local storage

💬 Intelligent Chat System

RAG-Enhanced Responses: Retrieval-Augmented Generation for accurate, context-aware answers
Vector Search Integration: Chat queries automatically search your document embeddings
Context-Aware: AI remembers your uploaded documents and their semantic content
Legal Expertise: Specialized prompts for legal document analysis with knowledge retrieval
Voice Support:
- Speech-to-Text: Speak your questions naturally
- Text-to-Speech: Listen to AI responses
History Management: Save and retrieve past conversations with semantic indexing
Multi-language: Support for multiple languages with cross-lingual embeddings
Smart Suggestions: AI suggests relevant follow-up questions based on document content

🗣️ Voice Features

High-Quality Recognition: Google Speech-to-Text API
Natural Voice Output: Google Text-to-Speech with natural voices
Smart Processing: Optimized for legal terminology
Error Handling: Comprehensive fallback mechanisms
Browser Compatibility: Works across modern browsers

🛠️ API Endpoints

Endpoint	Method	Description
`/api/analyze`	POST	Analyze uploaded documents with Gemini AI
`/api/chat`	POST	Interactive AI chat with RAG enhancement
`/api/embeddings`	POST	Generate document embeddings for vector storage
`/api/vector-search`	POST	Semantic search across document library
`/api/rag-query`	POST	RAG-powered document querying
`/api/speech-to-text`	POST	Convert voice recordings to text
`/api/text-to-speech`	POST	Convert text responses to speech
`/api/transcribe`	POST	Process meeting transcripts from extension
`/api/websocket`	WebSocket	Real-time communication for live features
`/api/latest-transcript`	GET	Retrieve latest meeting transcript
`/api/chat-history`	GET/POST	Manage chat conversation history
`/api/document-similarity`	POST	Find similar documents using vector search

🧪 Testing

Test Gemini AI Integration

# Verify Gemini AI connectivity
node test-new-gemini.js

# Expected output: "Yes, the new Gemini SDK is working!"

Manual Testing Checklist

✅ Document Upload: Test various file types (PDF, DOC, images)
✅ Vector Embedding: Verify documents are properly vectorized and stored
✅ Semantic Search: Test vector-based document similarity search
✅ RAG Functionality: Verify contextual retrieval in chat responses
✅ Chat Features: Try document-specific questions with RAG enhancement
✅ Voice Input: Test microphone recording and transcription
✅ Voice Output: Test text-to-speech playbook
✅ Meet Integration: Join Google Meet with extension enabled
✅ Real-time Sync: Verify WebSocket communication
✅ Embedding Quality: Test cross-document legal clause matching

Voice Features Test

Visit /voice-test to test speech-to-text and text-to-speech functionality.

🚀 Deployment

The application is ready for deployment to:

Recommended Platforms

Vercel (optimal for Next.js)
Netlify
Railway
Render

Build Verification ✅

Your successful build shows:

✅ All routes compiled successfully
✅ Static pages pre-rendered (30/30)
✅ Bundle sizes optimized (~169-221 KB first load)
✅ No critical compilation errors

Environment Variables for Production

Ensure all environment variables from .env are configured in your deployment platform.

📚 Key Components

🤖 Gemini AI Integration

Model: gemini-2.0-flash-exp (latest experimental model)
Direct File Upload: Superior to traditional OCR approaches
Context Management: Maintains document context across conversations
Auto-cleanup: Files automatically deleted after processing for privacy

🧠 Vector Database & RAG Architecture

Vector Store: Google Vertex AI Vector Search for enterprise-grade similarity search
Embedding Model: Google Universal Sentence Encoder (text-embedding-004)
RAG Implementation: Retrieval-Augmented Generation for enhanced responses
Hybrid Search: Combines Vertex AI semantic vector search with traditional keyword matching
Document Chunking: Smart text segmentation for optimal embedding quality
Similarity Threshold: Configurable relevance scoring for precise retrieval
Auto-scaling: Vertex AI handles scaling and performance optimization automatically

🌐 Chrome Extension Architecture

Manifest v3: Latest Chrome extension standard
Content Script: Integrates seamlessly with Google Meet UI
Background Script: Manages WebSocket connections and audio capture
Popup Interface: User-friendly control panel

🔐 Authentication & Security

Google OAuth: Secure user authentication
Firebase Integration: User data and chat history storage
API Key Protection: Secure environment variable handling
CORS Configuration: Proper cross-origin request handling

📖 Advanced Configuration

Gemini AI Model Configuration

// lib/gemini.ts
const genAI = new GoogleGenAI({
  apiKey: process.env.NEXT_PUBLIC_GEMINI_API_KEY,
});

// Using the latest experimental model
model: "gemini-2.0-flash-exp";

Vector Database Configuration

// lib/vertex-vector-store.ts
import { VertexAI } from "@google-cloud/vertexai";
import { MatchServiceClient } from "@google-cloud/aiplatform";

const vertex = new VertexAI({
  project: process.env.GOOGLE_CLOUD_PROJECT_ID,
  location: process.env.GOOGLE_CLOUD_LOCATION,
});

const matchClient = new MatchServiceClient();

// Document embedding and storage
const embedDocument = async (text: string, metadata: any) => {
  const embedding = await generateEmbedding(text);

  // Store in Vertex AI Vector Search
  await matchClient.upsertDatapoints({
    index: `projects/${process.env.GOOGLE_CLOUD_PROJECT_ID}/locations/${process.env.GOOGLE_CLOUD_LOCATION}/indexes/${process.env.VERTEX_AI_INDEX_ID}`,
    datapoints: [
      {
        datapointId: metadata.docId,
        featureVector: embedding,
        restricts: [
          { namespace: "userId", allowList: [metadata.userId] },
          { namespace: "docType", allowList: [metadata.docType] },
        ],
      },
    ],
  });
};

RAG Implementation

// lib/rag-engine.ts
export const ragQuery = async (question: string, userId: string) => {
  // 1. Generate query embedding
  const queryEmbedding = await generateEmbedding(question);

  // 2. Vertex AI Vector similarity search
  const searchResponse = await matchClient.findNeighbors({
    indexEndpoint: process.env.VERTEX_AI_INDEX_ENDPOINT,
    queries: [
      {
        datapoint: {
          featureVector: queryEmbedding,
        },
        neighborCount: 5,
        restricts: [{ namespace: "userId", allowList: [userId] }],
      },
    ],
  });

  // 3. Construct context from retrieved documents
  const context = searchResponse[0].nearestNeighbors
    .map((neighbor) => neighbor.datapoint.metadata.text)
    .join("\n\n");

  // 4. Generate response with context
  return await generateContextualResponse(question, context);
};

WebSocket Configuration

// Real-time meeting transcription
const socket = io(SERVER_URL, {
  transports: ["websocket", "polling"],
  upgrade: true,
});

Voice Configuration

// Configurable voice settings
{
  voice: "en-IN-Standard-A",
  languageCode: "en-IN",
  speed: 1.0
}

🔍 Troubleshooting

Common Issues

🎤 Microphone Not Working

Check browser permissions for microphone access
Ensure microphone isn't being used by other applications
Try refreshing the page

🔌 Extension Not Connecting

Verify the Next.js server is running on localhost:3000
Check Chrome extension permissions
Try reloading the extension

📄 Document Upload Failing

Ensure file size is under 10MB
Check supported file formats (PDF, DOC, images)
Verify Gemini API key is configured correctly

🔍 Vector Search Not Working

Check Google Cloud project ID and Vertex AI API enablement
Verify Vertex AI Vector Search index exists and is deployed
Ensure service account has proper Vertex AI permissions
Check index endpoint configuration

🧠 RAG Responses Inaccurate

Verify Vertex AI embedding model configuration
Check similarity threshold settings in Vector Search
Ensure document chunking is optimized for legal content
Review index deployment status and health

🔊 Voice Features Not Working

Check browser audio permissions
Ensure speakers/headphones are working
Verify Google API keys are set correctly

📝 Documentation

Additional documentation available in the docs directory:

WebSocket Integration: Real-time communication setup
Voice Features: Speech recognition and synthesis
Google Cloud Transcription: Meeting transcription setup
Integration Guide: Gemini AI integration details
Vector Database Setup: Vertex AI Vector Search configuration and embedding strategies
RAG Implementation: Retrieval-Augmented Generation with Vertex AI architecture

🤝 Contributing

Fork the repository
Create a feature branch (git checkout -b feature/amazing-feature)
Commit your changes (git commit -m 'Add amazing feature')
Push to the branch (git push origin feature/amazing-feature)
Open a Pull Request

⚖️ Legal Disclaimer

Important Notice: This application provides information and analysis for educational purposes only and does not constitute legal advice. Always consult with qualified legal professionals for important legal matters. The AI-generated content should be verified independently.

📄 License

This project is part of a Google Hackathon submission and follows applicable hackathon guidelines.

🎯 Project Stats

Build Time: ~2-3 seconds
Bundle Size: 101-221 KB (optimized)
API Routes: 20+ endpoints
Vector Dimensions: 768 (text-embedding-004)
Max Document Size: 10MB per upload
Embedding Speed: ~500ms per document chunk
RAG Retrieval: Top-5 similar documents
Chrome Extension: Manifest v3 compliant
AI Model: Google Gemini 2.0 Flash (Experimental)
Languages Supported: 10+ languages
File Formats: PDF, DOC, images, text

🎉 Built with ❤️ for the Google Hackathon

Making legal documents accessible to everyone through the power of AI

Live Demo • Chrome Extension • API Docs

🚀 Quick Start Commands

# Install dependencies
pnpm install

# Start development server
pnpm dev

# Build for production
pnpm build

# Test Gemini integration
node test-new-gemini.js

# Package Chrome extension
cd chrome-extension && ./package-extension.sh

Ready to deploy! 🎯 Your LegalEase AI platform is now ready for production deployment.

Name		Name	Last commit message	Last commit date
Latest commit History 38 Commits
app		app
chrome-extension		chrome-extension
components		components
contexts		contexts
docs		docs
docs1		docs1
hooks		hooks
lib		lib
public		public
styles		styles
test-scripts		test-scripts
types		types
.gitignore		.gitignore
INTEGRATION_COMPLETE.md		INTEGRATION_COMPLETE.md
README.md		README.md
components.json		components.json
download-test-audio.js		download-test-audio.js
next.config.mjs		next.config.mjs
package-lock.json		package-lock.json
package.json		package.json
pnpm-lock.yaml		pnpm-lock.yaml
postcss.config.mjs		postcss.config.mjs
test-google-api-key.js		test-google-api-key.js
test-google-speech.js		test-google-speech.js
test-new-gemini.js		test-new-gemini.js
test-transcription.js		test-transcription.js
tsconfig.json		tsconfig.json

Folders and files

Latest commit

History

Repository files navigation

LegalEase AI - Legal Document Simplification Platform

🚀 Overview

🎯 Key Features

🏗️ Architecture

Core Technologies

Project Structure

🔧 Installation & Setup

Prerequisites

1. Clone & Install

2. Environment Setup

3. Run Development Server

🔌 Chrome Extension Setup

Development Installation

Usage in Google Meet

📋 Features Deep Dive

🔍 Document Analysis

🎤 Real-time Meeting Transcription

💬 Intelligent Chat System

🗣️ Voice Features

🛠️ API Endpoints

🧪 Testing

Test Gemini AI Integration

Manual Testing Checklist

Voice Features Test

🚀 Deployment

Recommended Platforms

Build Verification ✅

Environment Variables for Production

📚 Key Components

🤖 Gemini AI Integration

🧠 Vector Database & RAG Architecture

🌐 Chrome Extension Architecture

🔐 Authentication & Security

📖 Advanced Configuration

Gemini AI Model Configuration

Vector Database Configuration

RAG Implementation

WebSocket Configuration

Voice Configuration

🔍 Troubleshooting

Common Issues

📝 Documentation

🤝 Contributing

⚖️ Legal Disclaimer

📄 License

🎯 Project Stats

🚀 Quick Start Commands

About

Topics

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages