InvoiceFlow - Document Matching & Reconciliation Platform

Automated three-way reconciliation for Purchase Orders, Invoices, and Delivery Notes using AI-powered document extraction and intelligent matching.

Problem & Solution

Problem: Manual reconciliation of purchase orders, invoices, and delivery notes is time-consuming, error-prone, and doesn't scale. Finance teams struggle to:

• Match documents across different formats and vendors
• Detect discrepancies in quantities, prices, and tax calculations
• Track currency mismatches and missing line items
• Generate audit-ready reconciliation reports

Solution: InvoiceFlow automates the entire reconciliation workflow:

• AI-Powered Extraction: Uses Azure Form Recognizer and LLM to extract structured data from PDFs
• Intelligent Matching: Matches documents by PO number or vendor name with fuzzy matching
• Discrepancy Detection: Automatically flags quantity mismatches, price changes, currency differences, tax errors, and missing items
• Three-Way Reconciliation: Compares PO ↔ Invoice ↔ Delivery Note in a single workflow
• Report Generation: Produces PDF, JSON, and CSV reports with detailed discrepancy analysis

Screenshots

The dashboard shows detailed discrepancy analysis, including tax rate mismatches, tax amount differences, and comprehensive matching results with confidence scores.

The analytics page provides insights into document processing metrics, matching statistics, and reconciliation trends.

The reconciliation report provides detailed three-way matching results with line-by-line comparison, discrepancy flags, and audit-ready documentation.

High-Level Architecture

graph TB
    subgraph "Frontend"
        UI[Next.js UI]
        Components[React Components]
    end

    subgraph "Backend API"
        API[FastAPI REST API]
        Services[Business Logic Services]
        Extractors[Document Extractors]
    end

    subgraph "AI Services"
        Azure[Azure Form Recognizer]
        LLM[OpenAI GPT-4]
    end

    subgraph "Storage"
        DB[(PostgreSQL)]
        MinIO[MinIO S3]
    end

    UI --> API
    API --> Services
    Services --> Extractors
    Extractors --> Azure
    Extractors --> LLM
    Services --> DB
    Services --> MinIO

    style UI fill:#3b82f6
    style API fill:#10b981
    style Azure fill:#f59e0b
    style LLM fill:#8b5cf6
    style DB fill:#ef4444
    style MinIO fill:#06b6d4

Loading

Project Structure

InvoiceFlow/
├── backend/             # FastAPI backend
│   ├── src/
│   │   ├── api/         # API endpoints
│   │   ├── core/        # Config & DB
│   │   ├── models/      # DB models
│   │   ├── services/    # Business logic & extractors
│   │   └── main.py      # App entry
│   ├── alembic/         # Migrations
│   ├── scripts/         # Utilities
│   └── Dockerfile
│
├── frontend/            # Next.js frontend
│   ├── src/
│   │   ├── app/         # App router & pages
│   │   ├── components/  # React components
│   │   └── lib/         # Utilities
│   └── Dockerfile
│
├── docs/                # Documentation
├── docker-compose.yml   # Docker orchestration
└── README.md

Tech Stack

Layer	Technology	Purpose
Frontend	Next.js 14, React, TypeScript	Modern React framework with SSR
UI Framework	Tailwind CSS, shadcn/ui	Utility-first CSS, component library
Backend	FastAPI, Python 3.11	High-performance async API
Database	PostgreSQL 16	Relational data storage
Object Storage	MinIO	S3-compatible file storage
AI/ML	Azure Form Recognizer	Document OCR & extraction
AI/ML	OpenAI GPT-4 (via Instructor)	LLM-enhanced extraction
PDF Processing	PyMuPDF (fitz)	PDF parsing & validation
Matching	RapidFuzz	Fuzzy string matching
Infrastructure	Docker, Docker Compose	Containerization
Package Managers	uv (Python), npm (Node)	Dependency management

Quick Start

Prerequisites

• Docker & Docker Compose
• Azure Form Recognizer credentials
• OpenAI API key (optional, for enhanced extraction)

Setup

1. Clone repository

   git clone https://github.com/stephaneWamba/invoiceflow
   cd InvoiceFlow

2. Configure environment

   cp backend/env.example .env
   # Edit .env with your Azure and OpenAI credentials

3. Start services

   docker-compose up -d

4. Run migrations

   docker-compose exec backend alembic upgrade head

5. Access application

   • Frontend: http://localhost:3100
   • Backend API: http://localhost:8100
   • MinIO Console: http://localhost:9101 (minioadmin/minioadmin)

Documentation

• User Guide - How to use InvoiceFlow
• Architecture - System design & components
• API Reference - REST API documentation
• Frontend Guide - Frontend architecture
• Local Setup - Development setup guide

License

MIT