Building the Retail Store of the Future with AWS, SUSE and Lenovo

The retail industry is highly competitive and companies are constantly looking for ways to differentiate themselves. They are adopting new technologies like AI and edge computing and modernizing their application environments with containers, Kubernetes and cloud providers. Some of the challenges they are facing are inventory management, customer service failures, data security, and ways to leverage omnichannel to integrate in-store and the online experience.

Inventory management is critical to eliminate overstocking or understocking. Overstocking can lead to stock obsolescence and increased holding costs. Understocking may cause customers to look elsewhere and can result in a loss of your once loyal customer base.

Customer service failures can be anything from slow service, long lines, or out-of-stock items.

All of these concerns are common across any company in the retail industry. Let’s now take a look at an example company and how we can remedy their issues using technologies from AWS, SUSE and Lenovo.

A Retail Store Example

Retail Challenges: Inventory & Customer Experience

Consider a typical retail chain, we will call them Johl’s for the purposes of this example. Johl’s is experiencing many of the common issues identified above.

• They are having difficulties moving inventory leading to stock obsolescence and increased holding costs.
• Their customer service experience has been less than satisfactory. With slow in-store service and long lines at checkout leading to customers looking to competitors.

Their goals are to:

• Improve sales
• Improve the customer experience
• Move product in a more timely manner
  • Reduce excess inventory
  • Feed demand planning systems
  • Forecast stock shortages
  • Manage product lifecycle, reducing expiration loss

Digital Solution: Mobile App & AI-Driven Coupons

In order to address the issues with stock obsolescence and poor customer service experience we can build a digital campaign that will include a mobile application for the customer to use in-store.

• We will use the mobile application to track customer activity in-store and use that along with previous customer sales to target coupons that they would have a high probability to use during their visit. This mobile application will provide an Omnichannel retail experience for the customer whether in-store or online. Omnichannel retail provides a consistent and unified experience for customers across all platforms.
• If we do not have a current coupon that matches, we can use GenAI at the edge to create a personalized coupon using in-store activity, previous purchases and even use customer loyalty data as parameters into the AI model.
• A token in the app will be used to send location data for the model in store.

In summary:

• A Mobile Application for a better in-store and online experience (Omnichannel)
• Track customer activity in-store to determine behaviors like:
  • Browsing, focused (dwell), ingress/egress
• Use in store activity plus sales history to:
  • Send targeted coupons in real-time to their phone
  • Target excess inventory
  • Reward with GenAI coupons tied to customer loyalty

The Flow – Pulling it all Together

Let’s now take a look at the process and data flow of this proposed solution. Our customer walks into a Johl’s branch with the mobile application active. This allows them to receive real-time coupons based on in-store activity, previous purchase history and customer loyalty incentives. They could be standing in front of merchandise in a pattern that is flagged as “focused” or “dwell”. this could potentially lead to a related coupon being pushed to their phone in real-time. They could also be determined to be roaming, which would send a notification to the sales team that a customer may need assistance along with their location.

The customer can also call for a customer service representative through the app to avoid having to search for one manually. How many times have you found yourself traversing departments looking for someone to help you in a completely different department? In addition, they can order products online that are not in-stock at the current store. This is where we combine customer experience with Omnichannel retail.

Below is a demo recording to illustrate what this would look like for both the customer in their mobile application and the customer sales representative with an in-store application showing traffic and alerts.

Technology Stack: AWS, SUSE Edge & Lenovo

How will we implement this solution using, as the title of this blog implies, technology from AWS, SUSE and Lenovo?

Here is a high-level architecture of the solution:

There is a lot going on here. Let’s break it down by solution, steps and data flow.

AWS

In AWS, we are training our AI model with customer sales history, promotional data and branch data. This model will also use previous in-store customer activity to determine trends for better predictions and determine coupon usage probabilities.

Once production ready, we will kick off a CI/CD pipeline which will wrap our model in a service and add our service to our Amazon ECR Image Repository.  We are now ready to deploy our service to the edge. That’s where SUSE and Lenovo come in.

SUSE Rancher Prime in AWS

With SUSE Rancher Prime running in AWS, we are able to manage our Kubernetes clusters at the Edge including application deployment and operating system patches and upgrades. We can also secure our workloads with deep packet inspection and policy management to prohibit unallowed activity per our known policies. We can also look for nefarious activity by bad actors trying to steal personal or financial information in our runtime environment. We can add SUSE Observability to help us monitor all of our edge clusters and applications. SUSE Observability helps us to remediate any issues that may occur limiting downtime.

SUSE at the Edge

SUSE Edge for Retail is a full Kubernetes stack to enable containers running in a secure, lightweight and performant environment across edge devices in retail stores across the world. It is composed of K3s, a powerful and lightweight Kubernetes distribution ideal for running containers at the edge and SUSE Linux Micro, a secure, immutable operating system built for containers with compliance included.

With SUSE Edge enabling our container workloads at the edge, we can have a message broker listening for MQTT messages on our edge device. We can also wrap our AWS services in SpringBoot applications to allow us to use AWS Step Functions for executing business rules at the edge. This allows us to determine if the customer is focused, browsing, or entering/exiting the store. We can also filter any “noise” from our IoT devices, in this case the mobile application.

We can then send this data to both the model at the edge for processing and determining or generating coupons for the customer to use in store. In addition, we’ll send this data to our AWS MSK (Managed Kafka) application to send back AMAZON Sagemaker for additional model training and refinement. Another valuable way to leverage the Kafka messages is to playback these messages to help identify high traffic areas or blind spots in our stores. We can use this to move high value items into areas with more food traffic or rearrange our store to eliminate blind spots.

Lenovo

So, where are we going to run these applications at the Edge? Lenovo has a powerful set of edge devices that are ideal for retail use cases. We have a couple of options:

Lenovo SE100 

The Lenovo SE100 is a robust, compact, and efficient server tailored for deploying AI and edge computing solutions across various industries, offering a balance of performance, security, and ease of management in distributed environments. The SE100 is ideal for Small Language Models (SLMs) and GenAI at the Edge.

LenovoSE360V2

The Lenovo ThinkEdge SE360 V2 provides a robust, high-performance, and versatile edge computing platform with strong connectivity, security, and manageability features, suitable for demanding environments and a wide array of edge-centric applications. It allows for up to two GPUs and can handle Large Language Models (LLMs) and GenAI or Predictive AI as well as video processing.

Both devices would be a good fit for Johl’s. The SE100 comes in under $2,000 USD while the SE360V2 is a beefier option coming in at a slightly higher price point.

Summary

In this blog we outlined how AWS, SUSE, and Lenovo can combine to build a modern retail experience addressing current industry challenges. We emphasized the need for personalized customer experiences, efficient inventory management, data security, and seamless omnichannel integration.

Using a fictional retail chain, “Johl’s,” as an example, we described issues like stock obsolescence, poor customer service, and difficulties in leveraging both in-store and online experiences. The proposed solution involved a mobile application for customers that tracks in-store activity. This data combined with purchase history, loyalty and promotional data is then used to deliver targeted, personalized coupons, and enhance overall customer engagement. Generative AI at the edge can also be leveraged for creating personalized coupons based on real-time activity and customer data.

The technical implementation involves:

• AWS: For training AI/ML models, managing data analytics, and providing a scalable cloud infrastructure.
• SUSE: For Kubernetes management, security and observability at the edge with SUSE Rancher Prime, and for providing a secure and resilient operating foundation through SUSE Linux Enterprise Micro and SUSE Edge for Retail.
• Lenovo: For providing robust edge servers and devices like the ThinkEdge SE100 and SE360 V2, capable of handling AI workloads, video processing, and edge computing.

The combined technologies aim to deliver real-time, personalized experiences, optimize store operations, improve inventory management, and enhance customer satisfaction. By embracing innovative technologies and collaboration, retailers can create the stores of the future and thrive in the digital age.

For more information, please check out the SUSE – AWS and SUSE – Lenovo Landing pages:

Fast and Secure Delivery of Enterprise Workloads on AWS | SUSE

Accelerate time to value with Open Solutions from Core to Cloud to Edge | SUSE