“Purchase with joy”: The tech transition for total retail therapy

Transform Editor-in-Chief Gavin Allen talks with Huawei’s Kevin Kong about the future of smart retail.

Gavin Allen: Are AI and smart retail the killers or saviors of the physical store?

Kevin Kong, General Manager of Enterprise Logistics and Retail department, Huawei

Kevin Kong: It’s not killing or saving, but reshaping. It’s changing the logic of the three core retail elements of consumer, product, and shop – from experience-driven to data-driven.

Retail has changed from product-oriented – where you went to the shop and selected from what was on offer – to customer-centric, online. It’s gone from people finding products to products finding people.

We’re at a new level, where customers can chat to an AI agent and ask, “Tomorrow is my daughter’s birthday, what should I buy her from your store?” The system can ask the daughter’s age, explore what she likes, then provide a list of gift options.

Gavin Allen: What’s different inside the shop?

Kevin Kong: We have more connections and IoT sensors to assess customer flow. People can use virtual mirrors to test skin products and try on clothes, or in restaurants smart cameras and edge computing can ensure elements such as table cleaning standards are met, sending alerts to staff managers if there are failings. Huawei focuses on the infrastructure, being open to software vendors, interacting with them to offer customized feedback and retail solutions.

Gavin Allen: It’s about much more precise monitoring and insight?

Kevin Kong: Exactly. Take bento boxes in a Family Mart. If unsold, they get tossed in the garbage. By using data intelligence and AI forecasting, we can see that, say, tomato plus beef is the favorite – whereas there are very few sales of potato plus vegetables. That means we can tell the store boss how many to stock of each over the next five days, to maximize sales while reducing waste.

Gavin Allen: Just expand on the commercial impact that AI is already having?

Kevin Kong: AI applications are a smart brain triggering the transition from experience-driven to intelligent decision-making and accelerating industrial digitalization. First, the product iteration cycle has been compressed to a weekly level and relies on AI rather than gut decisions. Huawei collaborated with DEEPEXI to launch the Fast5000E training and inference machine, combined with DeepSeek's foundational model. This helps fashion designers at Belle, a fashion apparel group, generate high-quality reference patterns with one click, reducing new pattern design time by 70%, achieving high efficiency from weekly to daily levels.

Some retail companies have relied on algorithmic order forecasting and inventory adjustment to build a flexible network of regional distribution centers and forward warehouses [where goods are stored in locations close to the end customer], enhancing supply chain response efficiency to the hourly level.

Last year, Huawei partnered with Tianhong/Rainbow, a shopping mall operator, to create the Bailing Bird Large Model. AI-generated receipts and points functions, AI customer feedback analysis systems, and AI shopping assistant systems greatly enhance the shopping experience. AI analysis helped increase the second-purchase rate by 40% compared to manual decisions, with work efficiency increasing by over 30%.

Finally, Huawei collaborated with iFlytek to develop the DeepSeek Spark integrated machine, designed for high-performance AI training and inference, particularly for large language models and other deep learning workloads. The machine’s capabilities enable customer services technology such as AI-powered chatbots and virtual assistants. This significantly enhanced user experience and perceptions, and increased customer acceptance of robot services by more than 60%.

Gavin Allen: So, AI can be embedded throughout the operator value chain?

Kevin Kong: Yes. We have the Pangu AI model, but we open source a lot of the specific models, allowing others to build their software on our Ascend computing foundation for particular industries. From an AI perspective, we focus about 60% on the hardware and then for another 20%, we recommend that our partners provide some level 1 or level 2 layered models serving as the capability layer. For the last 20%, we rely on our partners for deep customization and the total integrated solution to make software that’s unique for their company and requirements.

Gavin Allen: And what are the challenges for companies embarking on this digital journey?

Kevin Kong: Margins in retail are tight. So, total investment in digital technology has been limited. There’s a focus on short-term ROI, and that’s been a bottleneck. Second, they don’t have enough high-quality data, which is the source for AI and digitalization. So, data governance is vital.

The third issue is connection. That becomes more complicated as the retail chain expands – like Luckin Coffee for instance, going from a handful of stores in China to having tens of thousands of shops globally. It’s about how to manage the different layers of the network, and the security fragmentation. Also, to increase profits, a business has to optimize the whole workflow, not just one segment of it. An enterprise has to link upstream partners, from warehouse to shop to feedback from customers. Too often, as the range becomes more complicated, there’s a lack of flexibility in supply chain management, inability to respond quickly to market demand, and low overall supply chain efficiency. Finally, companies don’t have enough basic IT and AI talent. Technology evolves fast, but with a limited budget the challenge is to find the talent to maintain it. That is a common bottleneck for many companies and makes it hard for them to support a long-term digitalization project.

Gavin Allen: How does Huawei help overcome those challenges?

Kevin Kong: Our focus is on the ICT infrastructure and we have a 4C synergy concept. The first C is a Convergent network. So rather than separate networks for employees, consumers and suppliers – which become complex and overlapping as a business expands – we combine the three into one. That reduces cost. We can then simplify the architecture and offer cloud, network, digital intelligence and security as an end-to-end solution, all in one box. Second, there’s Convenient maintenance: branch network launch time drops from one day to 10 minutes and fault location falls from hours to minutes. Third, it’s a Customized experience with online and offline integration and omnichannel marketing. And the fourth and final C is Credible security, with six built-in capabilities and threat awareness measured in seconds. AI is detecting 24 hours a day, seven days a week. The whole process is secure. And the data is separate and lies encrypted with you the customer.

Gavin Allen: So, if as an entrepreneur I wanted to set up a Luckin Coffee franchise, what help could Huawei offer?

Kevin Kong: We promise “zero touch” to set up a new shop, connecting your devices and embedding AI within the platform. We will provide the cables, switches, and Access Point (AP). When your workmen are ready to install it, let us know and we’ll express deliver the gear to you. In fact, if you don’t have workmen, we can recommend someone for you, chosen from our network or service delivery partners.

Once we establish a Wi-Fi signal, you’ll want your customers to scan the app, because then the re-purchase rate is higher (about 20% to 30% at Luckin Coffee). And as shop owner, you can offer targeted discounts on the particular kinds of coffee the customer likes, using precision marketing and big data analysis supported by Huawei Cloud.

Gavin Allen: And explain what the “1+1+1+N” intelligent networking architecture is that underpins that?

Kevin Kong: This offers massive terminal device connections through one IoT integrated network, one control and analysis platform, and one IoT device. Relying on SD-WAN technology, it connects the headquarters, branches, factory, and warehouse into a unified network, across geographic regions. It also brings comprehensive application effects to retail companies. Integration architecture has been widely used in stores such as Wumart, Lianhua Supermarket, Yonghui Supermarket, etc.

Take electronic price tags as an example. According to CR Vanguard test store data, it takes about 20 minutes for a traditional ESL (electronic shelf label) base station to refresh 20,000 electronic price tags, but Huawei's IoT-integrated AP can complete that task in just 3 minutes, vastly improving efficiency and enabling more dynamic price promotion refreshes.

Gavin Allen: What’s the current shape of the retail industry in the digital era?

Kevin Kong: China's retail chains are expanding at scale, with leading companies accelerating the "ten-thousand-store strategy." For example, the leading convenience store brand Meiyijia has surpassed 38,000 stores, the new tea drink company, Mixue Bingcheng, has over 47,000 global stores, and Luckin Coffee has more than 25,000 global stores. Leading brands are also extending multi-format layouts and different business models. For example, supermarkets are developing a combination of large stores and forward warehouses, and the fashion sector is exploring multi-channel layouts such as flagship stores + pop-up stores + collection stores. Retailers and brands respond to the fast-changing market, while AI reconstructs the logic of full-chain smart operations.

Gavin Allen: Will AI be replacing people within the retail chain?

Kevin Kong: It won’t replace people. It reduces costs and warehouse waste and gives retailers more space to manage. It allows them to allocate tasks more efficiently, to reduce layers, to refine management and focus on grabbing the interests of customers to encourage them to re-purchase. Luckin Coffee, for example, once you order from the App you press a button and can see the video of your coffee being prepared. You can witness the making of your coffee. The Xiaomi EV brand can offer their VVIP customers special secured parking spots in luxury shopping centers. You can make your customers feel comfortable, to encourage re-purchasing.

Gavin Allen: Do you think shopping in future will be more passive – AI will predict what I need and push to me, without the need for stores?

Kevin Kong: In China, 75% of retail is still offline (in a store), and 25% online. In 5-10 years, online may be up to 30-35%, but still the majority will be offline. It’s growing, but very slowly. Younger people want purchasing combined with social interaction. They might pick up an orange and a video will tell them it arrived from Australia just two days ago. They want to purchase with joy, with interaction. They want to attract eyeballs. It’s social, joy, feel and touch. Online is convenient, but offline still does optimization – analyzing what each region purchases and optimizing stocks as a result. It allows people to easily get what they need, offering interactions at the coffee bar, etc. – it all helps to trigger re-purchases.

At the same time, online and offline iare combining. Rainbow has an App online but includes information which pushes you to visit your nearby store where there’s a promotion offline. The border between the two is becoming invisible.

We cannot trigger or generate revenue for brands but we can reduce costs and improve efficiency. We continually enlarge our ICT front with our retail partners. No one person can solve all the problems. Our cloud platform is open, embracing independent software vendors to build the ecosystem.

Gavin Allen: And targeted to retail?

Kevin Kong: Exactly. We can offer a product list of software capabilities bundled in with the hardware. You as the retailers can select what you need, open it and use it. And we offer to run that as a service too. We release your energy to serving more customers. The product list is becoming richer and richer. We invite customers to play with it, to optimize the hardware, to speed up computing and communications, more convergent to one platform and reducing costs.

Gavin Allen: It’s a rolling program and always upgrading?

Kevin Kong: It’s never finished. We have to be open, always refreshing and learning from partners, and humanizing the process. The consumer pays the money so we should be focused on them – caring not just about what they need but about their emotional value too. They should feel joy so you should be improving their satisfaction a lot. That requires insight and analysis from retailers – how to serve the target consumer better and better.

We all need to embrace the advances brought by the intelligent era. Working with partners, we want to create value with digital infrastructure as the root, AI platforms and models as the core, and full-scenario applications as the bridge and ecosystem.

Companies are already witnessing the improvements and making AI their top strategy. And the ROI is getting shorter and shorter.

A lot of leaders are changing their mind as they see how AI will change the world: reshaping the business model, reshaping business logic and reshaping how they should manage their company.