LexiBrief AI ⚖️🚀

An automated, high-fidelity legal document analysis platform that extracts critical legal insights, compliance risks, and core deadlines from complex legal PDFs with persistent cloud history orchestration.

LexiBrief AI transforms dense legal PDFs into four structured intelligence panels—Summary & Context, Core Legal Issues, Risk Obligations, and Deadline Alerts—backed by a cyberpunk-grade React dashboard and a resilient Node.js analysis pipeline.

---

✨ Key Architectural Features

Capability	Description
Gen AI Parsing Core	Powered by Google Gemini 2.5 Flash for semantic document analysis and intelligent text-chunk processing across multi-page legal filings.
Fault-Tolerant Retry Engine	Dual-attempt recovery with automatic model fallback (2.5-flash → 2.0-flash → 1.5-flash) to bypass temporary 503 Service Unavailable spikes without breaking the UI.
Direct Shard-Pool Connectivity	Custom multi-seed MongoDB URI layout bypasses Windows local router DNS issues and mongodb+srv SRV lookup blockades.
Dynamic Sliding Dashboard	Interactive PDF dropzone with glassmorphism UI, typewriter stream animations, and a cloud-synchronized History Logs drawer with per-entry purge controls.

---

🛠 Tech Stack

Frontend

• React (Vite) — SPA shell & component architecture
• Tailwind CSS — utility-first cyberpunk theming
• Framer Motion — slide-in history panel & micro-interactions
• Axios — REST orchestration to Express API
• Lucide React — premium iconography

Backend

• Node.js + Express — REST API gateway
• Multer — in-memory PDF ingestion
• pdf-parse-new — multi-page text extraction
• dotenv + cors — secure env & cross-origin config

Database

• MongoDB Atlas with Mongoose ODM — persistent scan history, metadata indexing, and targeted record purging

AI Framework

• Google Gen AI SDK (@google/genai) — structured 4-section legal synthesis engine

---

📁 Project Directory Layout

legal-saas-project/
├── README.md                 # Project documentation (this file)
├── backend/
│   ├── .env.example          # Environment variable template
│   ├── .env                  # Local secrets (git-ignored)
│   ├── package.json
│   ├── server.js             # Express API, Gemini engine, MongoDB layer
│   └── node_modules/
└── frontend/
    ├── index.html
    ├── vite.config.js
    ├── package.json
    ├── public/
    │   └── favicon.svg
    └── src/
        ├── main.jsx
        ├── App.jsx           # LexiBrief AI dashboard & history drawer
        ├── App.css
        └── index.css

---

🚀 Quick Installation & Setup

Prerequisites

• Node.js 18+
• npm 9+
• MongoDB Atlas cluster (with Network Access configured)
• Google AI Studio API key (Get API Key)

1. Clone & install dependencies

git clone <your-repository-url>
cd legal-saas-project

# Backend
cd backend
npm install

# Frontend
cd ../frontend
npm install

2. Configure environment variables

Create backend/.env from the template:

cd backend
cp .env.example .env   # Windows: copy .env.example .env

Edit backend/.env with your credentials:

PORT=5001
GEMINI_API_KEY=your_google_gemini_api_key_here
MONGO_URI=mongodb://username:encoded_password@shard-00-00.example.mongodb.net:27017,shard-00-01.example.mongodb.net:27017,shard-00-02.example.mongodb.net:27017/legal_saas_db?ssl=true&replicaSet=atlas-example-shard-0&authSource=admin&retryWrites=true&w=majority

Password tip: If your MongoDB password contains @, URL-encode it as %40 in the connection string.
Example: password @MyPass123 → %40MyPass123

Variable	Description
PORT	Express server port (default 5001)
GEMINI_API_KEY	Google Gemini API key from AI Studio
MONGO_URI	Direct multi-shard MongoDB Atlas connection string

3. Run locally

Open two terminals:

Terminal 1 — Backend

cd backend
node server.js
# or: npm run dev   (with nodemon)

Expected output:

Server is LIVE and running on http://localhost:5001
🍃 MongoDB Database Engine Connected Successfully!

Terminal 2 — Frontend

cd frontend
npm run dev

Open the dashboard at http://localhost:5173

---

📡 API Endpoints

Method	Endpoint	Description
POST	/api/analyze	Upload PDF → Gemini analysis → persist to MongoDB
GET	/api/history	List recent scan metadata (fileName, timestamp)
GET	/api/history/:id	Fetch full historical analysis payload
DELETE	/api/history/:id	Purge a specific history record

---

🧠 Analysis Output Matrix

Every successful scan returns four structured sections:

1. 📋 Summary & Context
2. ⚖️ Core Legal Issues & Clauses
3. ⚠️ Key Risks & Critical Obligations
4. 📅 Detected Deadlines & Timeline Alerts

The frontend parser streams each block into dedicated glassmorphism cards with graceful empty-state fallbacks.

---

🔁 Retry & Resilience Model

On each Analyze Document action:

• Up to 3 Gemini models are attempted in sequence
• Each model receives up to 2 attempts on 503 overload errors
• Maximum theoretical calls per scan: 6 (typically 1 on success)
• MongoDB writes use in-memory fallback only when Atlas is unreachable

---

🖥 UI Highlights

• LexiBrief AI neon gradient hero branding
• Drag-and-drop PDF upload with pulse feedback
• Real-time typewriter + glitch stream rendering
• Right-side History Logs flyout (Framer Motion)
• One-click history restore & animated delete purge

---

⚠️ Troubleshooting

Symptom	Likely Cause	Fix
503 UNAVAILABLE from Gemini	Temporary Google model overload	Wait 1–2 min; retry (auto-fallback models enabled)
Invalid URL MongoDB error	Special chars in password (@)	URL-encode password (%40)
querySrv ECONNREFUSED	Router DNS blocks SRV lookups	Use direct shard MONGO_URI (see .env.example)
bad auth MongoDB	Wrong username/password	Reset Atlas DB user password & update .env
History empty after restart	MongoDB was offline during scan	Confirm 🍃 MongoDB Connected in terminal

---

📄 License

This project is provided for educational and portfolio demonstration purposes. Configure your own API keys and database credentials before any production deployment.

---

LexiBrief AI ⚖️🚀
Precision legal intelligence. Structured. Persistent. Production-ready.