The implementation of AI tools has increased across the retail sector in 2025 with 64 per cent of respondents to KPMG’s 2025 Intelligent Retail report saying they are deploying generative AI tools and 47 per cent stating the technology has become core to their business.
“Retailers have moved from pilots to production,” says Anya Minbiole, business strategy leader of retail and consumer goods at Microsoft. “Twelve to 18 months ago, we were still kicking the tyres. We asked, ‘can generative AI write copy?’ and then we watched as it created a meeting transcript, just to get a feel as to whether it might be able to help us. Now, we have gone beyond using generative AI as an assistant towards enabling combined human-agent teams to do much more.
“Secondly, we’re seeing generative AI moving from digital-only scenarios to physical execution. Multinational retail chain Walmart, for example, is using in-store agents to help store floor associates with product placement, restocking and shelf optimisation. Finally, we’re seeing brands become confident enough to let AI speak in their own voice – for instance Ralph Lauren’s ‘Ask Ralph’ digital stylist offers styling advice that is authentic and personalised to the individual customer.”
Anya Minbiole is business strategy leader of retail and consumer goods at Microsoft
This shift is the beginning of a new era in retail, one where the most forward-looking companies are combining human expertise with intelligent systems to operate faster and smarter. Microsoft coined a new phrase for these companies in its 2025 Work Trend Index report: ‘frontier firms’. The report describes these organisations as “businesses that are human-led and agent-operated”.
According to Minbiole, frontier firms in retail empower their employees to design and oversee processes while delegating repetitive or analytical tasks to AI-powered agents. “They build systems with key performance indicators (KPIs) to measure how well their agents perform, and they use digital orchestration to coordinate work between humans and agents,” she explains. “These organisations aren’t just using generative AI to be more productive; they’re using it to boost their competitive advantage, and they are set up to use their KPIs to learn quickly and experiment. Experimentation is critical to using generative AI.
“I’d argue that 12 to 18 months ago, retailers were starting to experiment from scratch. Now, they are building on what they’ve learned and refining use cases. Each experiment strengthens the next. It is never about technology, it’s about what you can do with it to create value. The real benchmark for AI progress is whether it makes a difference in people’s lives.”
Microsoft and its partner ecosystem are working to help retailers accelerate AI adoption. “We have account teams working closely with customers to create a strategy for exploration, sometimes starting with a workshop where customers identify the AI agents they need to create and onboard,” says Minbiole. “One example is American multinational food and beverage company Kraft Heinz, which partnered with Microsoft to create a real-time factory platform. In less than two years, the platform has reduced supply chain waste by 40 per cent, increased forecast accuracy by 20 per cent and saved over $1 billion in efficiencies. Direct-to-consumer retailer brands could deploy AI in a similar way.”
Meanwhile, Microsoft partner EPAM is collaborating with Albert Heijn, the largest supermarket chain in the Netherlands, to improve the store employee experience. It has done this by creating a conversational assistant that answers questions related to their tasks or customer requests, as well as a customer-facing app offering personalised recipe suggestions and shopping lists. “This app even knows what’s in your refrigerator via photos,” says Minbiole.
Accenture/Avanade worked with MediaMkt, a German chain of electronic stores, to help associates stay engaged with customers by asking an AI assistant called MyBuddy about everything from the price of a warranty on a dishwasher to a comparison of several TVs. Another Microsoft partner, Neudesic, has a Control Tower solution that provides retailers with a real-time digital twin view of their stores. Control Tower is an AI-powered store operations platform that connects people, processes and data to run a store, such as planogram compliance, equipment exceptions, promotion suggestions, enabling thousands of smarter decisions.
Avanade and Microsoft helped develop a hands-free AI assistant for MediaMarktSaturn employees to enhance in-store customer experiences
Microsoft’s ecosystem and its AI-powered platforms are empowering retailers to make better, data-driven decisions. “The real value of retail data lies in connecting to dynamic, learning agentic systems,” says Minbiole. “Retailers can now deploy intelligent agents that interpret data in real time and act on it, compressing the time between insight and execution. Loyalty programmes were originally designed for customers to ‘earn and burn’, but they’re now becoming AI-powered engagement engines. By integrating purchase and engagement data, AI can detect who is buying what and why, and thereby anticipate what will keep customers loyal.”
Coffee house chain Starbucks uses its ‘Deep Brew’ platform to personalise offers and recommend products to customers. Every tap, order and redemption updates the model, allowing Starbucks to tailor suggestions for customers based on the weather, time of day and even what’s in stock locally. “This closed-loop intelligence transforms loyalty from a marketing function into an adaptive, data-driven operating system, enabling retailers to move from reactive decision-making to real-time orchestration of customer experience and value,” says Minbiole.
“There are also emerging partners creating specialised agents that handle distinct parts of the value chain,” she adds. “Nimble’s market intelligence agent flags competitor promotions in real time; Ydistri models stock redistribution across stores; and Omnistream updates planograms automatically to highlight relevant products. Together, these agents compress the cycle from insight to execution to mere hours, creating a living ecosystem of intelligence where each agent becomes an expert that can act on what it knows best and learns through collaboration.”
Every tap on the Starbucks’ ‘Deep Brew’ platform updates the model allowing the coffee chain to personalise offers and recommendations to its customers
Agentic AI is emerging as the next step in global AI transformation across multiple industries, but how does Minbiole define it in a retail scenario?
“I’d like to describe scenarios that are generative AI but not agentic AI, so the difference is easier to grasp,” she says. “When a store associate at Tractor Supply Company uses a headset to ask a digital assistant about a warranty on a product, that associate is asking their generative AI-enabled tool questions and getting answers. The ‘human plus assistant’ phase is the first phase of generative AI in many companies. This phase has a lot of value by growing the company’s knowledge base as it helps associates.
“Agentic systems are the next step. These systems might span multiple departments, essentially breaking down the company definitions of what is ‘marketing’ or ‘operations’, or ‘sales’; agents run 24/7 to perform interrelated tasks, such as reviewing pricing or assortments or ingesting competitive information. Transforming work itself from a role-driven to an outcome-driven model also helps dissolve departmental barriers, morphing the decades-old ‘org chart’ to a new organisational construct we call ‘work chart’.”
With retail technology specialists readying themselves for the NRF 2026: Retail’s Big Show in New York, USA, in January 2026, Minbiole predicts the conversations around AI in retail to continue to focus on “exploration and value – what are companies trying, what’s working and what’s not.” And with 67 per cent of respondents to KPMG’s 2025 Intelligent Retail report planning to increase the percentage of global budget spent on AI in 2026, Minbiole expects to see more examples of how “intelligence on tap is redefining how we create competitive value – to scale rapidly, operate with agility and generate value faster.”
Discover more insights like this in the Winter 2025 issue of Technology Record. Don’t miss out – subscribe for free today and get future issues delivered straight to your inbox.