Invoice Manager - Production-Grade AI System

A full-stack, enterprise-grade Invoice Manager application leveraging modern AI orchestration (LangChain/LangGraph), RAG-based document querying, MCP servers for modular tool execution, and a React TypeScript frontend.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│                    React Frontend (TypeScript)                   │
│         Dashboard │ Documents │ Global Chat │ Doc Chat          │
└─────────────────────────────────────────────────────────────────┘
                                 │
                                 ▼
┌─────────────────────────────────────────────────────────────────┐
│                      FastAPI Backend                             │
│  /upload-invoice │ /documents │ /validate │ /chat/global        │
└─────────────────────────────────────────────────────────────────┘
                                 │
              ┌──────────────────┼──────────────────┐
              ▼                  ▼                  ▼
┌──────────────────┐  ┌──────────────────┐  ┌──────────────────┐
│   LangGraph      │  │   MCP Servers    │  │   MongoDB        │
│   Agent          │  │  ─────────────   │  │  ───────────     │
│  ─────────────   │  │  • Validation    │  │  • documents     │
│  • Intent Node   │  │  • RAG Query     │  │  • embeddings    │
│  • Tool Router   │  │  • Chat History  │  │  • chats         │
│  • Fallback      │  │  • Doc Listing   │  │  • validations   │
└──────────────────┘  └──────────────────┘  └──────────────────┘
              │                  │
              ▼                  ▼
┌──────────────────────────────────────────┐
│           Groq LLM Pool (Load Balanced)  │
│   llama3-70b │ mixtral-8x7b │ llama3-8b  │
└──────────────────────────────────────────┘

✨ Features

• Invoice Upload & Validation - Upload PDF/image/text invoices with AI-powered validation
• Per-Document RAG - Ask questions about specific invoices using vector search
• Global Chatbot - LangGraph-orchestrated agent for invoice management
• MCP Servers - Modular, structured tools for each capability
• Model Rotation - Load distribution across Groq models with fallback

📁 Project Structure

AgenticAI_FA3/
├── backend/
│   ├── app/
│   │   ├── main.py              # FastAPI entry point
│   │   ├── config.py            # Environment configuration
│   │   ├── api/routes/          # API endpoints
│   │   ├── core/
│   │   │   ├── llm/             # Groq client with rotation
│   │   │   ├── langchain/       # Tools, RAG, embeddings
│   │   │   └── langgraph/       # Agent orchestration
│   │   ├── mcp/                 # MCP servers
│   │   ├── db/                  # MongoDB models & repos
│   │   ├── services/            # Business logic
│   │   └── utils/               # Text extraction
│   ├── requirements.txt
│   └── .env.example
│
└── frontend/
    ├── src/
    │   ├── components/          # Reusable UI components
    │   ├── pages/               # Page components
    │   ├── services/            # API client
    │   └── types/               # TypeScript interfaces
    └── package.json

🚀 Quick Start

Prerequisites

• Python 3.11+
• Node.js 18+
• MongoDB (running locally or connection string)
• Groq API Key

Backend Setup

cd backend

# Create virtual environment
python -m venv venv
venv\Scripts\activate  # Windows
# source venv/bin/activate  # Linux/Mac

# Install dependencies
pip install -r requirements.txt

# Configure environment
copy .env.example .env
# Edit .env and add your GROQ_API_KEY

# Run server
uvicorn app.main:app --reload --port 8000

Frontend Setup

cd frontend

# Install dependencies
npm install

# Run development server
npm run dev

Access the Application

• Frontend: http://localhost:5173
• Backend API: http://localhost:8000
• API Docs: http://localhost:8000/docs

📡 API Endpoints

| Method | Endpoint | Description | |--------|----------|-------------| | POST | /api/upload-invoice | Upload invoice file | | GET | /api/documents | List all invoices | | GET | /api/documents/{id} | Get invoice details | | DELETE | /api/documents/{id} | Delete invoice | | POST | /api/validate-invoice/{id} | Validate invoice | | POST | /api/chat/global | Global chatbot | | POST | /api/chat/document/{id} | Per-document RAG chat | | GET | /api/chats/global | Get global chat history | | GET | /api/chats/document/{id} | Get document chat history |

🤖 LangGraph Agent

The agent uses conditional routing based on intent classification:

User Message → Intent Classification
                      │
    ┌─────────────────┼─────────────────┐
    ↓                 ↓                 ↓
Validate         RAG Query         List Docs
    │                 │                 │
    └────────────────┼─────────────────┘
                      ↓
              Response/Fallback

Supported Intents:

• validate_invoice - Validate document correctness
• query_document - RAG query on specific invoice
• list_documents - List all invoices
• get_document_details - Get invoice metadata
• general_chat - General conversation

🔧 MCP Servers

| Server | Tools | Purpose | |--------|-------|---------| | validation_server | validate_invoice, get_validation_rules | Invoice validation | | rag_server | query_document, get_document_context | Per-document RAG | | chat_server | get_chat_history, save_message | Chat persistence | | document_server | list_documents, get_document_metadata | Document CRUD |

💬 Example Prompts

Global Chat:

• "List all my invoices"
• "Validate invoice XYZ"
• "Which invoices are pending validation?"
• "Help me understand invoice management"

Document Chat (RAG):

• "What is the total amount?"
• "Who is the vendor?"
• "Is tax included?"
• "Summarize this invoice"

🛠️ Technology Stack

• Backend: FastAPI, LangChain, LangGraph, MongoDB, Groq
• Frontend: React 19, TypeScript, Bootstrap 5
• AI/ML: sentence-transformers, FAISS, Groq LLMs
• Document Processing: pypdf, pdfplumber, Pillow

📝 License

MIT License