Fujitsu launches retail AI trial for seven Japanese firms

Fujitsu has launched a trial environment for four AI agents designed to support retail operations and management decision-making. The company will test the system with seven retail companies in Japan.

The agents focus on sales structure analysis, customer loyalty analysis, sales planning and store manager support. They are designed to analyse retail and customer data and recommend practical measures, including changes to product ranges, inventory allocations and shelf displays.

The rollout reflects Fujitsu's effort to push AI use in retail beyond reporting and data analysis towards recommendations staff can act on. The trial environment is being offered through Uvance for Retail, Fujitsu's retail and distribution offering, alongside an execution platform that links data sources and business functions.

Retailers have amassed large volumes of sales and customer information, but many still struggle to use that data quickly in planning and day-to-day operations. The system is intended to address delays caused by manual reporting, specialist analytical work and the long gap between spotting a problem and responding to it.

Four functions

The first agent examines sales structure. It breaks down sales performance into component parts, highlights imbalances, and identifies areas where a retailer may be losing margin or missing growth opportunities. It can also suggest steps to improve profit and estimate the impact of those changes.

The second agent focuses on customer loyalty. It analyses customer data to identify behavioural patterns, purchase motivations and factors linked to attrition that may not be visible through standard measures such as visit frequency or loyalty-point use.

The third agent covers merchandise planning. It is intended to help merchandisers prepare product-level and store-level sales plans more quickly while accounting for changes in customer demand, market conditions and competition. It can also recommend store-by-store inventory allocations and later suggest inventory transfers to avoid stock shortages and missed sales.

The fourth agent is designed for store managers. It combines sales data with external information, including local market conditions, store characteristics, weather and social media activity, to recommend changes to product assortments and shelf layouts.

Platform layer

Behind the agents is a common execution platform that handles data storage, processing and access control. This shared layer allows the agents to draw on the same information base and business systems while supporting tasks such as analysis, demand forecasting and recommendation generation.

Fujitsu plans to expand the number of agents over time for different retail and distribution use cases. Its longer-term ambition is to extend AI-led recommendations beyond retail into other parts of the supply chain, including manufacturing, wholesale and logistics.

For Fujitsu, the retail push adds to a broader regional strategy in Australia and New Zealand, where it has operated for more than 50 years. The group said its acquisition of GK Software strengthened its retail offering with software for point-of-sale and omnichannel operations, adding to its existing capabilities in data, AI, cloud, security and managed services.

The retail AI agents and platform are scheduled for commercial availability in Australia and New Zealand in June 2027. That timeline gives Fujitsu time to test the system in Japan before a wider launch in the Oceania market.

Carl Marx, Head of Industry, Enterprise & Commercial, Fujitsu in Oceania, described the intended role of the software in retail decision-making.

"The practical value of AI agents is their ability to shorten the distance between insight and action. For retailers in Australia and New Zealand, that could mean helping teams respond faster to changing demand, customer behaviour and stock requirements, while keeping people responsible for the final decisions," said Marx.