Generative AI Financial Data Agent

A comprehensive automated system designed to ingest financial documents (such as Annual Reports or Bank Statements), extract key financial metrics using Generative AI, and store structured data in a MongoDB database. The project features a Flask-based web interface for file management and data review.

Features

• PDF Ingestion: Automatically parses text from PDF documents using PyMuPDF.
• AI-Powered Extraction: Utilizes OpenRouter (DeepSeek model) to intelligently locate and extract:
  • Financial Summaries: Key metrics like Net Profit, Deposits, and Advances.
  • Balance Sheets: Detailed breakdown of Assets, Liabilities, and Equity.
• Two-Stage Processing: Implements a "Finder" agent to locate relevant pages and an "Extractor" agent to parse data into strict JSON schemas.
• Database Integration: Stores extracted financial data in MongoDB for persistent record-keeping.
• Web Interface: A user-friendly Dashboard built with Flask to upload files and review extracted data.
• Authentication: Secure login system to protect data views.

Tech Stack

• Language: Python 3.x
• Web Framework: Flask (with Flask-Login)
• Database: MongoDB
• AI/LLM: OpenAI Client (via OpenRouter targeting deepseek-r1t2-chimera)
• PDF Processing: PyMuPDF (fitz)

Installation

1. Clone the repository

   git clone <repository-url>
   cd Generative_AI_Financial_Data_Agent

2. Install Dependencies

   pip install -r requirements.txt

3. Set up MongoDB Ensure you have MongoDB installed. This project is configured to look for MongoDB on port .

   # Example command to start mongo on the specific port
   mongod --port <your port>

4. Environment Configuration Create a .env file in the root directory and add your OpenRouter API key:

   OPENROUTER_API_KEY=your_api_key_here

Usage

1. Run the Application

Start the Flask server:

python app.py

You should see output indicating the server is running on http://127.0.0.1:5000.

2. Access the Web Interface

• Upload: Select a PDF financial report and click "Upload". The system will process it in the background.
• Login: To view extracted data, navigate to the Login page.
  • Default Username: admin
  • Default Password: password123 (Note: These credentials are hardcoded in app.py for demonstration purposes).
• Review: Once logged in, you can see a JSON-formatted breakdown of the data extracted from your uploaded documents.

Project Structure

• app.py: The main Flask application entry point. Handles routing, authentication, and file uploads.
• ingest_processor.py: The core logic engine. It handles PDF chunking, calls the AI agent, and manages MongoDB operations.
• ai_agent.py: A CLI script (called by the processor) that interfaces with the AI API to perform specific tasks (Find/Extract).
• test_*.py: Utility scripts to verify your environment:
  • test_api.py: Checks connectivity to OpenRouter.
  • test_db.py: Checks connectivity to MongoDB (Port 27018).
  • test_imports.py: Verifies all Python dependencies are installed.

Testing

To ensure your environment is set up correctly before running the main app, you can run the provided test scripts:

# Test API Connection
python test_api.py

# Test Database Connection
python test_db.py

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.