Warehouse management AI agent

An AI agent that autonomously manages business warehouses by tracking inventory levels, detecting low-stock products, communicating with suppliers, and placing orders.

Comment

Warehouse Management AI Agent: Automating Reorders for Small Businesses

Inspiration We were inspired by the small businesses in Hong Kong and Kazakhstan that struggle to manage stock levels due to limited staffing and tools, resulting in shortages and operational stress. This project empowers these teams with simple automation to keep shelves stocked and communities served efficiently.

What it does The system ingests product stock from Google Sheets, persists it in MongoDB, and applies inventory logic (safety stock and reorder points) to compute purchase quantities. It generates summarized reports and composes clear, supplier-ready HTML emails that list products, quantities, unit prices, and total cost, then sends them automatically.

How we built it We used Python for rapid development, Google Sheets as a friendly data source, and MongoDB to store products, suppliers, inventory, and order history. The agent computes reorder needs using safety stock and moving-average style signals to smooth variability and avoid both stockouts and over-ordering.

Challenges we ran into Designing a clean architecture between the inventory ingestion (Sheets), the data store (MongoDB), and the ordering workflow took iteration and learning. Getting indexes and document models right for fast lookups and updates was a key hurdle as we adopted MongoDB practices for performance and maintainability.

Accomplishments that we're proud of We delivered an end-to-end system that measurably reduces stockouts and the associated waste, especially for perishables. By automating purchasing with urgency indicators, we help teams focus on customers while cutting stress and reducing spoilage.

What we learned We learned how to apply engineering fundamentals to a real operations problem: designing data models, validating inputs, coordinating services, and turning formulas into robust workflows. We also practiced writing maintainable Python with configuration via environment variables and production-friendly logging.

What’s next for Warehouse Management AI Agent Next, we’ll deepen demand forecasting with adaptive models, seasonal adjustments, and vendor lead-time variability. We’ll build a simple web interface for configuration and dashboards, add multi-warehouse routing, and integrate vendor APIs for order confirmations and shipment tracking to close the loop.

Share this project:

Updates