🧾 LangGraph Finance Analyzer

A production-ready LangGraph workflow demonstrating a multi-agent financial analysis system that automatically loads, cleans, normalizes, analyzes, and generates insights from raw financial CSV data.

📊 Architecture

The system uses a supervisor pattern with specialized agents:

• Supervisor Agent - Intelligent routing and orchestration
• Data Quality Agent - Validates input data
• Data Processing Pipeline - Load → Clean → Analyze
• Analysis Agents - Visualization, Forecasting, Anomaly Detection, Risk Analysis, Report Generation

🚀 Quick Start

Prerequisites

• Python 3.8+
• Azure OpenAI API credentials

Installation

# Clone the repository
git clone https://github.com/rmagnin88/fin-agent.git
cd fin-agent

# Install dependencies
pip install -r requirements.txt

# Configure environment
cp env.example .env
# Edit .env with your Azure OpenAI credentials

Run the Demo

Option A: LangGraph Studio (Recommended)

langgraph dev

Open the URL shown in the output (or go to https://smith.langchain.com) and interact with the graph visually.

Option B: Command Line

python run_demo.py

Option C: Programmatic Usage

from graph import app

result = app.invoke({
    "raw_csv": csv_content,
    "csv_path": "your_financials.csv"
})

print(result["insights"])

📁 Project Structure

fin-agent/
├── graph.py              # Main LangGraph workflow definition
├── run_demo.py           # Interactive demo runner
├── test_workflow.py      # Unit tests
├── export_graph.py       # Graph visualization exporter
├── sample_financials.csv # Sample financial dataset
├── requirements.txt      # Python dependencies
├── env.example           # Environment variables template
├── langgraph.json        # LangGraph configuration
└── graph/                # Graph visualizations
    └── finance_analyzer.png

🔧 Configuration

Create a .env file with your Azure OpenAI credentials:

AZURE_OPENAI_API_KEY=your_azure_openai_api_key_here
AZURE_OPENAI_ENDPOINT=https://your-resource.cognitiveservices.azure.com
AZURE_OPENAI_MODEL=gpt-5-mini
AZURE_OPENAI_DEPLOYMENT=gpt-5-mini
AZURE_OPENAI_API_VERSION=2024-12-01-preview
AZURE_OPENAI_MAX_TOKENS=4096

Optional: For LangGraph Studio and LangSmith tracing:

LANGSMITH_API_KEY=your_langsmith_api_key_here
LANGGRAPH_API_KEY=your_langgraph_api_key_here

🌟 Features

• ✅ Intelligent Data Cleaning - Automatically removes formatting, handles missing values
• ✅ Smart Column Detection - Identifies sales, profit, country, segment columns automatically
• ✅ Comprehensive Analysis - Computes totals, averages, top performers, breakdowns
• ✅ AI-Powered Insights - Generates narrative summaries with business recommendations
• ✅ Multi-Agent Architecture - Specialized agents for visualization, forecasting, anomaly detection, risk analysis
• ✅ Flexible Input - Works with CSV files or raw CSV strings
• ✅ Error Handling - Graceful fallbacks if data is malformed

📚 Resources

• LangGraph Documentation
• LangGraph Studio Guide

📝 License

MIT License - see LICENSE file for details