​Coles Supermarkets embraces AI, cloud applications in 500-plus stores with Azure Stack HCI​

​Technology-driven retail efficiency at the edge using Azure Stack HCI

​With tens of thousands of high-quality products helping Australians eat and live better every day, Coles Group, a leading name in Australia’s retail sector, operates an expansive network of 1,800 stores across multiple brands, including 850 Coles Supermarkets locations. Committed to seamless experiences for millions of customers in person and online, the retail giant is an early adopter and investor in innovative technologies, such as AI and computer vision.

​“Historically, our investments have been in making sure that we have the right talent to inspire forward-thinking business leaders,” explains Silvio Giorgio, General Manager of Data and Intelligence at Coles Group. “For more than nine years, we’ve had an artificial intelligence team that runs like a startup within Coles—a successful approach to get traction—but we’re at a point now where we need to shift our focus toward scaling, creating positive challenges.”

​As AI matures and more companies take advantage of prebuilt models and off-the-shelf products, Coles wanted to move fast and make AI technology and other digital innovations accessible to more people within the organization, with the ultimate goal of creating new hands-on customer experiences. “We set a strategy to build the solid foundations we need, but that also focuses on the future,” says Giorgio. “The way we are scaling requires us to embrace technologies to lift our capabilities, increase speed to value, and attract future talent while simultaneously exploring innovation in a highly competitive market.”

​Coles built an adaptive intelligent edge backbone on Microsoft Azure Stack HCI, making it easy, efficient, and cost-effective to connect and manage a growing technology landscape, including tens of thousands of cameras, team member devices, and point-of-sale system workloads that need real-time responses. The intelligent edge backbone was born out of necessity for a solution that could address aging hardware and firmware updates remotely in one location without putting unnecessary burden on team members. By unifying siloed teams, distributed sites, and sprawling systems into a single operations and application platform, Coles is building out new capabilities with Microsoft’s adaptive cloud approach, and it continues to expand and strengthen its solution using pivotal Microsoft resources and research.

​The decision to use Azure Stack HCI was influenced by the desire to tap into the existing Coles infrastructure without dedicating servers to each application, and Coles was able to improve its stores’ infrastructure and run models on shared infrastructure without interruption. “Some of our stores are heritage stores that don’t necessarily have big, sophisticated comms rooms where we can have five servers for each different vendor, so the Azure Stack HCI layer is really important for bringing some of that cloud engineering into our edge locations,” explains Roslyn Mackay, Head of Technology Innovation at Coles Group. “We need hyperconverged infrastructure where we can still run some of our existing workloads in applications that might be on Windows or Linux virtual machines, but we also need to have some newer edge AI applications that are containerized and managed concurrently at our sites.”

​Unlocking centralized management, containerization, and collaboration

​Within its new Azure Stack HCI environment, Coles wanted quick and efficient access to press-button deployments for application releases and updates rather than having to manage its infrastructure on a store-by-store basis. “When you multiply those deployments across hundreds of locations, it suddenly becomes a very large and costly manual task,” says Giorgio. “The infrastructure we’ve set up with Azure Stack HCI in stores has been a real enabler for the central management of our applications.”

​​To maintain the environment behind the scenes, Coles is operating more than 250 Azure Stack HCI clusters in Azure, and every Windows workload has been transferred to Azure Stack HCI. In collaboration with Microsoft, Coles also built a multi-store deployment control plane, and it has the added support of NVIDIA, which provided the GPUs used in the store infrastructure and the performance needed at scale. “We worked like an extended project team where we’d give updates back to Microsoft and NVIDIA, and they would give us thought leadership,” says Mackay.

​​Coles fully stands behind Microsoft innovations and is confident that any new technology it tests will be helpful in the market and benefit customers, suppliers, and team members alike. ”Our long-standing relationship with Microsoft is credit to the way that they’ve not only innovated in the market over the last couple of years but also to how they are willing to innovate with us and other forward-looking organizations,” says Giorgio. Adds Mackay, “Just using part of the Microsoft suite for Azure Stack HCI worked really well with the other Azure tools we use, including Azure Machine Learning.”

​​The company uses Azure Machine Learning to train and develop edge AI models. “We’re going to be working more with Microsoft over the next 12 months to build out Azure Machine Learning operations for the edge to be able to seamlessly test and deploy new models and ensure auditability of our models and the different versions over time,” says Mackay. “The Azure automated machine learning tool was really useful for us, and it speeds up our data annotation time for training models by 50 percent.”

Delighting customers with AI innovation

​With Azure Stack HCI as the foundation, the Coles intelligent edge backbone has been a key enabler for team members’ ability to respond to immediate demands on the business. “Prior to the intelligent edge backbone, we had dedicated servers that would have virtual machines sitting on top with isolated applications,” says Mackay. “It wasn’t until we uplifted those servers for horizontal scaling and rolled out Azure Stack HCI that we could start to more actively deploy new and innovative applications and services.”

​Select Coles locations in Melbourne serve as innovation stores for testing surprise and delight experiences, including fresh produce recognition. Customers can place their items on scales and edge AI models trained by Azure Machine Learning effectively identify the products to simplify the checkout process. “In these trial stores, you put your item down and bang, there’s the item on the screen for you to accept rather than actually navigate pressing buttons,” shares Giorgio. “We got to a point where we were achieving 94 percent accuracy across all of our product SKUs, and the item recommended from the AI model was exactly the right one that was surfaced.”

​Similar frictionless experiences have freed team members from having to intervene in lagging processes and instead given them the capacity to focus on the parts of their jobs that add value to the Coles experience. “We’re saving fractions of a second on a transaction or an experience, but it does make a difference when you multiply that by the millions of customers who might interact with us every day,” says Giorgio.

​“We want the experience for our customers to be not so much about the store looking like it’s from the future, but that we have the products they need where they should be located and that a team member is available to support them,” says Mackay. “We see so much heavy lifting being done in the background through our Azure edge computing solutions and AI models to bring that to life. If we did not have that in-store infrastructure and hyperconverged environment provided by Azure Stack HCI, we wouldn’t be able to respond at this pace or in the same way that we have.”

​Meeting project goals and deploying new solutions in record time

​Managing hundreds of supermarkets with millions of transactions occurring every day on top of its expanding technology stack, Coles needed speed and reliability at ambitious scale. Prior to the intelligent edge backbone, it would take IT staff upwards of 30 minutes to deploy an application to the store via the virtual machine or container on Azure Stack HCI. With the new environment, it takes minutes. Impressed by the results, Coles has also since expanded its Azure Stack HCI footprint from 2 stores to more than 500 and deploys computer vision applications to those stores in rapid time. “We’ve met our target of a six-times increase in the pace that we can deploy applications to stores, so that speed, and the reliability at that scale, have been amazing,” says Giorgio.

​With Azure Stack HCI now in most of its retail stores, Coles not only has the infrastructure to support these new applications and experiences but does so using fewer resources within budget. “If every solution needed to have its own server, its own dedicated cameras, its own CPU and GPU, and all of the other necessary resources, it would be very hard for our business to justify putting up business case after business case for each use case,” says Mackay. “The fact that we’re able to use some of the infrastructure footprint that’s already there and enabled by Azure Stack HCI is what has really made this strategy affordable for our business, particularly at a time in Australia where there are a lot of cost-of-living pressures, and we can deliver value for our customers through our pricing.”

​Find out more about Coles on Facebook, LinkedIn, and YouTube.