Part 3/11:
Unlike retailers, manufacturers like Kinu do not operate physical stores but supply products to various retail outlets. They aim to leverage predictive analytics to maintain ideal shelf stock levels across over 15,000 stores in North America and Canada, encompassing more than 1,500 SKUs.
However, the data received from retail partners is often inconsistent:
Varying data quality and schemas
Delayed updates due to sales cycles
Different data granularities—daily, weekly, or monthly
Inability to access real-time store conditions (e.g., via cameras)
Given these constraints, AI systems are designed from the data point of view rather than relying on intrusive hardware.