StockPulse AI

A mid-market distribution company was losing millions annually to overstock and stockout cycles. Their demand planning relied on spreadsheets updated weekly by regional managers, each using different assumptions and formats. Seasonal demand spikes were consistently misjudged, leading to warehouse overflow in some regions and empty shelves in others.

Existing forecasting tools required data science teams they did not have, and the off-the-shelf solutions could not integrate with their legacy ERP and warehouse management systems.

We built a custom demand intelligence platform that ingests historical sales data, supplier lead times, and external signals like weather and regional events to generate SKU-level demand forecasts updated every six hours. The system integrates directly with their existing ERP via a REST API layer we designed.

A dashboard gives procurement teams real-time visibility into predicted stockouts, overstock risk scores, and automated reorder suggestions. The ML pipeline runs on AWS SageMaker with a feedback loop that improves accuracy as more data flows through the system.