The Inside Track

this month we chatted with Yili Wu, CEO of SandStar, the AI company that provides industry-leading computer vision technology to asset owners and retailers:

Yili, thank you so much for giving your time to us today. Would you please introduce yourself to our readers by telling us a little bit about SandStar and your ambition for the company.

Of course, thank you so much Geoff. We founded SandStar in 2016. We’ve spent the past four years just focusing on one thing: using AI to empower the retail industry. From a technology perspective, we focus on two things: the first one is the eyes, which means that we want to use cameras to digitalize all the key behaviors related to the products. The second thing is the brain. We want big data, the brain, to help business owners to make better decisions. 

SandStar, the name of our company, comes from the fact that we want to focus on small things to create value.  Sand is such a small thing, but then we dream big, so star, it is the largest thing. From small to big, so SandStar! 

Thank you Yili. So, there are going to be some topics here that are going to be of great interest to our readers. Artificial Intelligence and Machine Learning are things they are consistently hearing about that might be helpful to their business. I know of particular interest to our readers will be your Smart Kiosk product, with which I know you have already gained quite a bit of business in our sector. Can you tell us, what is a Smart Kiosk and how does it benefit its customers? 

Sure, so the Smart Kiosk just requires a small change to a normal cabinet – a cooler or a freezer – to turn it into a vending machine. 

The solution is very simple; we take a normal cabinet and put two cameras, one at the top and one at the bottom, and then add a lock. Each time a consumer wants to buy something from the cabinet they just swipe their credit card. The credit card payment gateway tells us that we can unlock the door; the consumer then opens the door, selects whatever they want from the cabinet and closes the door. They are then charged automatically for whatever they have taken from the cabinet. This is the most basic scenario.

The benefits are very clear. Firstly, we help our vending customers to save more than 60% of the cost, because traditional, high quality vending machines have very complicated mechanical systems, which are extremely expensive. But just a simple cabinet, cooler, or freezer, with two cameras and a small computer on top of it is much, much cheaper. 

The second benefit… typically, based on statistics, more than 90% of consumers purchase only one product at a time from a traditional vending machine. You add your cash, select a product, the machine almost literally throws the product at you, and you take just that one product and your change. 

With the Smart Kiosk a consumer displays a different buying behavior; they will often open the door and take multiple products from the cabinet. More than 50%, closer to 60% of the revenue generated from a Smart Kiosk is from this augmented consumer behavior – selecting multiple products during a single visit.

The third benefit is the big data brain. It is really important to note that we don’t collect any private or  personal consumer information at all. However, we can help the business owner to know when the kiosk door is opened, what products are touched and then what products are sold. This kind of information can easily help to generate more revenue.

We conducted an experiment in China. We hired a highly experienced manager from a well-known chain of convenience stores. We asked him to make an inventory selection and create a planogram of the products. Then we used a purely computer-based system to make an alternative plan for comparison.


Once we completed the experiment, the big data brain was able to collate all the data and identify which products should be placed in the kiosk, and where they should be placed to maximize revenue gains. Typically, the big data brain is able to identify and increase sales by around 20% based purely on data collected from consumer buying behaviors. Again I want to stress that at no time is any private consumer information collected, but the system is still able to reduce costs and generate more revenue.

Ok, this is a really interesting point, because the way that I’ve often seen this type of solution positioned has always been just as a reduction in cost on traditional vending machines. Of course, that is attractive, but I think you are talking about a new opportunity altogether, this isn’t just saving money on traditional vending machines, this is an opportunity to make more money at the same time. 

So, from your experience working with your customers, how do you see them changing the way they run their businesses, operationally and commercially, once they start using your technology?

As an example, we have a customer who has a Smart Kiosk in a US airport. They only recently put our machine at the airport gate. During recent months, with COVID-19, traveler numbers have been much lower than normal. There has been no advertising campaign, they just the placed Smart Kiosk cooler, with a credit card reader – that is all. 

The kiosk is selling almost US$200 worth of food and beverages daily, even during this period of reduced traveler numbers. They have seen that having bought the kiosk, they can earn their money back in just six weeks, which is totally different to the traditional mechanical vending machines. Because of the way the door opens, consumers can just select several items, including sandwiches and salads, which are very difficult to sell from traditional vending machines. Everyone is very excited about the solution.

Ok, so far you have quoted some numbers which will really excite people, but I just want to address some of the more practical things, so our readers will have some idea of how easy it is to scale this kind of technology.


I’ll start by asking you this: We know that unmanned technology solutions are incredibly popular in China. You see more and more unmanned stores, there are lots of unmanned kiosks, cabinets, coolers, etc. But China is a unique market, with an incredible appetite for technological advancements. How do you see this technology growing in less technically advanced markets, where there isn’t a unified app like WeChat that brings so many opportunities together?

I think actually the markets outside of China, for example the UK, Europe and North America are even more advanced than the market in China. All of these markets have already adopted credit cards, much earlier than China, that’s why China has Alipay, WeChat pay and so on. The Western world already has credit cards. Credit cards work particularly well with our solution. The only difference for credit card purchases is that we attach a credit card reader on the door, otherwise the experience is the same as in China. I think that outside China the markets are already very advanced and open to this solution.

Thinking about cost, I’ve seen how much you have brought your costs down over the years, as happens with all new technology. What would you say to anybody who says to you that the cost is still too high, and it restricts the scaling out of this technology? How do you see the business case? 

The cost is separated out for the hardware and the software. 

For the hardware part, the actual costs are already 60% lower than traditional mechanical vending machines, so this is not a barrier at all. 

Then we go to the software part and the software part, the big data computer vision, generates much more revenue than the traditional vending solution. So, the software part is based on the business value, which means the cost is generally very acceptable to our customers. In the software part we provide the computer vision software license, we also provide a vending device management system and the big data prediction functions. So, you just pay once, and you get everything that you need for the business operations.

Yes, I wanted to ask you about that part and perhaps extend your answer here, because a lot of customers are worried that a vendor will provide them with technology and then won’t be able to support them on an ongoing basis. How do you work with your customers to ensure that they realise the value of your technology and your solution?

Primarily we are a software company, which means that we can easily provide our service remotely. This means that we can provide support to all our customers. 

We have a dedicated team providing top level support to our customers. For example, our project management leader worked for HP for 23 years before joining us, so he has many years of experience providing customers with top-level support.

With regards to hardware, we work in two ways. The first is that we fully leverage traditional brand ecosystems. Coolers are around everywhere already, so we just train these cooler brands how to manage the after-sales service for their customers and we want to leverage this network. Further, more and more we are hiring local teams, for example in South East Asia, North America, and Europe.

Ok, what would you describe as the hardest challenges to overcome to make this work? Every technology has its difficulties, what are yours, so people are aware of what they will need to deal with to succeed?


Actually, the thing is whether people want to adopt it. For example, we are in a relatively early stage of a solution, that has some limits for sure. So, I generally speak with my customers, stressing that the number one purpose is to generate business value for you, rather than test the limits of this technology. We don’t necessarily need to try to put very similar products into the kiosk to test whether accuracy is 98% vs 99%, we just need to make sure we can help you to make more money than before.

Also, during the COVID-19 period, we have been restricted to providing in-person support only to our local customers, which has slowed the roll-out of the solution, but we are now hiring more and more people in local markets.

I know that there are a handful of players that are looking at this market and have released solutions into this market. In your opinion, what differentiates a really good vendor from others in the market? What should the customers be looking for to say, ‘I want to work with this company above all others?’


I think the biggest differentiator for us, is that we have focused on using AI to empower the retail industry for the last five years, especially for the Smart Kiosks. We had some competitors back in 2018, I met them in New York, but right now I can’t really see anyone that exists that has invested in this solution at this level. We have invested in over 200 scientists and engineers, just to focus on this solution for the last five years, it has cost a lot of money, but I believe that we can make it happen. We have remained focused on this solution and nothing else and I believe we are the only survivors, no-one else is doing what we do.

I am sure that this is going to be really exciting for a lot of our readers, so if they wanted to trial working with you what should they do next?

I think the first thing is that we are very good friends of assets365, and we want to build on our partnership, so please contact assets365 in the first instance. If we all agree growth potential is there, we are very flexible and happy to support any kind of requirement.

Yili, thank you so much for your time, we really appreciate it, and we wish you and SandStar continued success!