Invoice Chatbot

Ask questions about your invoices in plain English and get instant answers. No complex commands, no waiting.

✨ What Makes It Special

🧠 Understands Natural Language

• "How many invoices are less than $2000?"
• "Show me overdue invoices from Microsoft"
• "What's our total spend with tech vendors?"

Uses AI to understand your questions, then processes everything locally for speed. Works great even without an API key.

🚀 Lightning Fast

Sub-10ms responses on thousands of invoices. Local pandas processing means no waiting for external APIs.

💬 Beautiful Interface

Colored output, emojis, and clear formatting that's actually pleasant to use.

🚀 Quick Start

Installation

git clone https://github.com/codingislub/chat.git
cd chat
python -m venv .venv

# Activate environment
# Windows: .venv\Scripts\activate
# Mac/Linux: source .venv/bin/activate

pip install -r requirements.txt

Usage

# Interactive mode
python -m src.chatbot --data data/invoices.json

# Single question
python -m src.chatbot --data data/invoices.json --q "How many invoices are overdue?"

# Debug mode
python -m src.chatbot --data data/invoices.json --verbose

📖 Example Questions

Get Overview:

python -m src.chatbot --data data/invoices.json --q "summary of all invoices"

Filter & Count:

# Small invoices
--q "how many invoices less than 2000"

# Vendor activity
--q "count invoices from Microsoft"

# Large tickets
--q "invoices over 5000"

Analyze Spending:

# Vendor totals
--q "total from Amazon"

# Overdue payments
--q "show me overdue invoices"

🗂️ Data Format

Simple JSON with your invoice data:

[
  {
    "invoice_number": "INV-001",
    "vendor": "Microsoft Corporation",
    "amount": 2500.00,
    "due_date": "2025-10-15",
    "status": "pending"
  }
]

Required: amount, vendor
Optional: invoice_number, customer, due_date, issue_date, status

⚙️ Configuration (Optional)

Create .env for enhanced AI understanding:

OPENAI_API_KEY=your_key_here
OPENAI_TEXT_MODEL=gpt-4o-mini
LOG_LEVEL=INFO

No API key? No problem – smart fallback patterns work perfectly.

📈 Performance

• Speed: < 10ms per query
• Scale: Tested with 10,000+ invoices
• Efficiency: Minimal memory usage
• Smart: AI only for understanding, not data processing

🆘 Troubleshooting

"No OpenAI API key found"
Just a warning – everything works fine with the fallback system.

"Invalid JSON format"
Validate with: python -m json.tool your_file.json

Slow performance?
Use --verbose to diagnose. System is optimized – slowness usually means memory issues.

🤝 Contributing

Fork → Make changes → Pull request. We'd love your help!