AWS released a replenishment automation solution on its blog, saying it streamlines inventory management by automating order placement for demand surges. It is the company's first major update to its machine learning tools since the launch of the MMF framework.
The solution uses Databricks Genie and Amazon Quick to create a loop that detects demand surges, decides which supplier to use, and places orders automatically. It is designed for retail and supply chain operations, with availability starting immediately for AWS customers.
"The forecast sits in a governed data platform. Supplier availability lives in a separate operational feed," said an AWS representative. The system ensures that the forecast and the action live in different systems, with the forecast in a governed data platform and supplier availability in an operational feed.
The announcement follows the release of the MMF framework, which AWS said changed the demand forecasting process by predicting demand across an entire catalog with no per-item tuning.
The solution runs in four stages: Databricks handles the forecast, while Amazon Quick repeats the last three in the loop on a schedule.
It runs on Databricks and Amazon Quick, with the companion repository and accelerator’s fresh_retail_net example providing everything needed to reproduce it.
AWS did not say how the solution will handle low-volume SKUs, and it raised the question of whether the system can scale to larger inventories. The solution runs in four stages, with the forecast generated by Databricks and the loop managed by Amazon Quick. The system escalates to a human only when no rule fits.
Source: awsml
