The ability to be able to accurately predict wrong inventory records through AI offers up many interesting use cases, with the potential to improve loyalty, sales and productivity. Over the last 18 months, the academic team have been using data shared by the eight participating retailers, to test the accuracy and useability of their predictions..