We present a retail-specific AMI environment developed in DFKI's Innovative Retail Laboratory (IRL) sponsored by Globus, a German hypermarket chain. The mission of IRL is the application of AI technologies (such as plan recognition, ubiquitous user modelling, multimodal interfaces, semantic web, virtual characters, sensor fusion, and adaptive information presentation) to the field of ambient retail intelligence. We present an overview of one of the flagship projects of IRL, the SemProm project funded by German Ministry of Education and Research with a budget of over 16 Million Euro. The Semantic Product Memory (SemProM) is a prerequisite for advanced retail intelligence. It provides a digital diary of the complete product life cycle that is embedded in the product itself using smart wireless micro-sensor technology.

The semantic product memory has the function of a black box in airplanes, and, like a flight recorder, records all relevant ambient parameters in digital form. Consequently, chocolate candy boxes in a refrigerated truck can "complain" to the air conditioning that their critical values for air humidity have been exceeded, and the air conditioning can automatically adjust itself. But most importantly, people can access these digital product diaries at any time. When such digital diaries are kept in the blue lid of the beluga caviar jar, in the bottom of the box of Belgium luxury chocolates and in the cork of a top French wine, the dealer as well as the end customer, can always check whether or not the premium product has been subjected to any ambient influences that would lower the quality. The micro-sensors note where and when the wine was not stored flat or was exposed to vibrations, daylight, or even strong temperature fluctuations. The end customer can then decide whether or not to purchase the product at the offered price, in spite of these shortcomings.

We introduce the concept of speaking products, which can present information from their digital diaries in a context-sensitive and anthropomorphic way. Customers can converse with products and can ask for product comparisons and assistance concerning allergies or dietary constraints. In addition, we show how AI methods can be used in instrumented retail environments to gain a deeper understanding of customer behaviour to enable better marketing decisions and more accurate demand forecasts.