🚀 WOFO.AI

Enterprise-Grade Offline-Capable RAG Knowledge Assistant

Secure. Private. Offline. Intelligent.

Features • Architecture • Tech Stack • Installation • API • Contributors

---

📖 Overview

WOFO.AI is a state-of-the-art Retrieval-Augmented Generation (RAG) system engineered for enterprises. It allows organizations to securely ingest internal documentation—policies, technical manuals, legal contracts—and empowers employees to query this knowledge base using natural language.

Unlike public AI tools, WOFO ensures total data privacy by running a localized RAG pipeline where your sensitive data never leaves the controlled environment until it's processed for your specific query.

Why WOFO?

• � Offline & Air-Gapped Ready: Can be deployed in completely disconnected environments using local LLMs.
• �🔐 Enterprise Security: Role-based access control (RBAC) ensures data is only accessible to authorized personnel.
• 🧠 Context-Aware AI: Uses advanced vector embeddings to understand the meaning of your queries, not just keywords.
• ⚡ High Performance: Powered by Qdrant for millisecond-latency vector searches.
• 📂 Universal Ingestion: Support for PDF, DOCX, TXT, and XLSX formats.

---

✨ Key Features

Category	Features
🧠 Intelligent Core	• RAG Pipeline: Advanced retrieval system combining semantic search with LLM generation. • Smart Chunking: Automatically processes and chunks documents for optimal context. • Hybrid AI Engine: Switch between cloud models (Gemini) and fully offline local models (LLaMA/Mistral) for zero-data-leakage environments.
🛡️ Security & Access	• Fully Offline Capable: Run entirely on your local infrastructure without internet access. • Role-Based Access: Granular permissions for Admins, Employees, and Users. • Data Isolation: Enterprise-grade data separation protocols. • Secure Auth: JWT-based authentication with encrypted password storage.
⚡ Performance	• Vector Search: High-speed similarity search using Qdrant (Runs Locally). • Scalable Architecture: Dockerized services for easy horizontal scaling. • Optimized Backend: Node.js & Express for non-blocking I/O.
🖥️ User Interface	• Modern Dashboard: React-based UI with Tailwind CSS. • Chat History: Persistent conversation memory. • File Management: Drag-and-drop uploads with real-time status.

---

🏗 Architecture

The system follows a modular microservices-ready architecture:

graph TD
    User[👤 User / Employee] -->|Queries| Frontend[💻 React Frontend]
    Admin[👔 Admin] -->|Uploads Docs| Frontend

    subgraph "Application Layer"
        Frontend -->|API Requests| Backend[⚙️ Node.js/Express Server]
        Backend -->|Auth| AuthDB[(🍃 MongoDB Users)]
    end

    subgraph "RAG Pipeline"
        Backend -->|1. Extract Text| Parser[📄 Document Parser]
        Parser -->|2. Chunk & Embed| Embedder[🧠 HF Sentence Transformer]
        Embedder -->|3. Store/Retrieve vectors| VectorDB[(🦁 Qdrant Vector DB)]
        
        Backend <-->|4. Get Context + History| VectorDB
        Backend -->|5. Send Context + Prompt| LLM[🤖 Gemini / LLaMA]
        LLM -->|6. Answer| Backend
    end

Loading

---

🛠 Tech Stack

Frontend

• Framework: React 18
• Styling: Tailwind CSS, FontAwesome
• Animations: Framer Motion
• State/Routing: React Router DOM

Backend

• Runtime: Node.js
• Framework: Express.js
• Database: MongoDB (Metadata, Chat Logs), Qdrant (Vector Embeddings)
• AI Integration: @google/generative-ai, @xenova/transformers
• File Processing: pdf-parse, mammoth, multer

DevOps & Tools

• Containerization: Docker
• Version Control: Git
• API Testing: Postman

---

🚀 Installation & Setup

Follow these steps to deploy WOFO locally.

Prerequisites

• Node.js (v18+)
• Docker Desktop (for Qdrant)
• MongoDB (Local or Atlas URI)

1️⃣ Clone the Repository

git clone https://github.com/your-org/wofo-ai.git
cd wofo-ai

2️⃣ Backend Setup

cd backend
npm install

Create a .env file in the backend directory:

PORT=5000
MONGO_URI=mongodb://localhost:27017/wofo_ai
JWT_SECRET=your_super_secret_key_here
GEMINI_API_KEY=your_google_gemini_api_key

3️⃣ Frontend Setup

cd ../frontend
npm install

4️⃣ Start Vector Database

Start Qdrant using Docker:

docker run -d -p 6333:6333 -p 6334:6334 \
    -v $(pwd)/qdrant_storage:/qdrant/storage:z \
    qdrant/qdrant

5️⃣ Run the Application

Backend:

# In backend terminal
npm start
# Server will run on http://localhost:5000

Frontend:

# In frontend terminal
npm start
# Client will open at http://localhost:3000

---

🔌 API Documentation

Detailed documentation is available in backend/API_DOCUMENTATIONI.md.

Core Endpoints

Method	Endpoint	Description
POST	/api/upload/ingestion	Ingest: Upload PDF/DOCX, chunk, embed, and store in Qdrant.
POST	/api/search/retrieval	Query: Retrieve relevant context and generate an AI answer.
POST	/api/chatbot/query	Chat: Conversational endpoint with memory focus.
GET	/api/files	List: Get all uploaded documents (Admin).

---

Built with ❤️ for the Future of Enterprise Knowledge by Ahad Dangarvawala