Twelve warehouses, one 15-year-old ERP, and $50k in inventory losses every month.
Mid-size DistributorMid-size distributor with 12 warehouses, frequent inventory mismatches, $50k monthly loss.
Business Impact
62% → 97%
Inventory Accuracy
$50k → $8k
Monthly Loss
3 days → 4 hrs
Purchase Order Cycle
The Problem
Inventory counts were wrong because products were shelved incorrectly, stockouts weren't caught until too late, and purchase orders were created manually.
Our Approach
A computer-vision + AI agent system. Employees scan shelves with a mobile app. AI identifies misplaced items, low stock, and incorrect shelving — then auto-generates purchase orders with manager approval guardrails.
How we solved it
Live inventory visibility · AI demand forecasting · ERP integration
Integrated with the client's existing infrastructure without a rip-and-replace.
Twelve warehouses and a 15-year-old ERP made inventory accuracy a monthly write-off. We engineered a computer-vision agent that catches misplaced SKUs and stockouts on the floor, then generates purchase orders behind manager approval guardrails — so automation stays auditable.
The Impact
62% → 97%
Inventory Accuracy
$50k → $8k
Monthly Loss
3 days → 4 hrs
Purchase Order Cycle
Stack: Python · TensorFlow Lite · FastAPI · OpenCV · SAP BAPI integration
Read about our guardrail system →