Building a Frictionless Shopping Experience with Mindaugas Eglinskas of Pixevia | Soar Payments LLC

Building a Frictionless Shopping Experience with Mindaugas Eglinskas of Pixevia

From their early days as competitors to Instagram, to their collaborations with defense organizations. In this episode of PayPod, host Jacob Hollabaugh sits down with Mindaugas Eglinskas of PIXEVIA to discuss the fascinating story behind the creation of autonomous stores. PIXEVIA is an AI platform for retail, which eliminates tills, provides best customer experience for store clients, and optimizes operations for store managers. Watch this episode for a fun and interesting dive into this latest development in payments and fintech.

Payments & Fintech Insights In This Episode

  • Pixevia was able to collaborate with the Minister of Defense and companies in the Netherlands and France to engineer a system capable of defining vehicles, number plates, and drones.
  • Using the Autonomous Store is simple and comparable to using a subway or underground system.
  • The store accepts a variety of payment methods including Visa, Mastercard, and Amex.
  • The advantages of an Autonomous Store include extended hours, reduced need for employees, better security, and reduced waste and shrinkage.
  • It is more challenging to steal from an AI-driven store than a regular store due to the capability to track every movement.
  • And SO much more!

Today’s Guest

Mindaugas Eglinskas : Pixevia

PIXEVIA is an AI platform for retail, which eliminates tills, provides best customer experience for store clients, and optimizes operations for store managers. All is done by combining information from different security cameras. With 90-95% of grocery transactions taking place in physical stores, it is a game changer in a 12 trillion dollar grocery retail market. Their proprietary computer vision technology brings the physical retail spaces into the digital age of seamless checkouts, real time 3D operating platform for store managers and a completely new standard of consumer data for brands. They were the first to launch an AI-driven store in Europe and now deploying stores with top European retailers.

Featured on the Show

About PayPod

PayPod is the leading voice in the payments and fintech industry, covering payments, risk management and new technology. Host Jacob Hollabaugh interviews leaders who are shaping the payments and fintech world, as they discuss the latest developments in the payments and fintech industry.

Episode Transcript

Jacob: Welcome to PayPod. The Payments Industry Podcast. Each week, we’ll bring you in-depth conversations with leaders who are shaping the payments and fintech world from payment processing to risk management and from new technology to entirely new payment types. If you want to know what’s happening in the world of fintech and payments, you’re in the right place. Hello, everyone. Welcome to PayPod. I’m your host, Jacob Hollabaugh. And today on the show, we are going to be talking about the new gold standard of shopping experiences, smart stores. The future has become the present. The tech has caught up with kind of the wild dreams and ambitions that I know I maybe had as a kid. And now we as consumers are going to have just about the most frictionless, in-person shopping experience imaginable. I’m absolutely fascinated with this concept, but obviously I have no idea how any of it works. So lucky for me and for you all listening, I am joined by an expert on these topics and one of the men leading the way on bringing these ideas to market. Mindaugas Eglinskas joins us, CEO of Pixevia, the company building AI powered autonomous smart Stores for the best customer experience. Mindaugas, welcome to the show. Thank you so much for being here.

Mindauges: Thank you. Thank you for having me.

Jacob: Absolutely. I am pretty blown away. Most of the guests we have on the show, I get into my research looking into the background of them and the company and it’s always fascinating. But I have to say this was one I was especially looking forward to because I myself have not got to experience an autonomous store quite yet. There are some in the city I live in, but I haven’t got to frequent them. But I’m fascinated by the concept. I’m really looking forward to talking about it. So let’s start back where the company started. Pixevia is about ten years old at this point. I think if LinkedIn is accurate, how did the idea for building an autonomous store even get on your radar back in 2013? Where did the original idea come from?

Mindauges: Yes, a bit crazy story. The path to creating autonomous stores was anything but straightforward. Initially, we began as competitors to Instagram. Later we shifted to defense and ultimately our final product was Autonomous Store. At the beginning, I started with my research at Vilnius University, where I worked as assistant professor with my students, and we worked with convolutional networks way before We were cool from 2005 to 2015. And as I mentioned, like our first application was real time processing for phones. Like a competitor to Instagram. It can sound weird generally, but like the idea was to extract filters like our Photoshop was already here, utilize our knowledge and put everything and present a better app for the world. Then we moved to video processing for drones in need. I’m a little bit different case. Drones use processors similar to phones and also like I have very strict energy constraints. We collaborated with Minister of Defense and companies in the Netherlands and France, and we engineered a system capable to define vehicles, number plates, people from drones. However, between 2015, 2017, nobody really was interested. Like now it’s different case, but like back then, not a big deal. I would say this way. And we also implemented quite big project in parking space. We used the same power system, but just we put cameras on high rise buildings, in this case stationary cameras, and we were about 2000 parking spots. And we send this information in real time displays which park was empty, which are occupied, and just a step by step by improving the system like our final product from 2019 is autonomous stores. And we are proud that we built very first autonomous stores in Europe. Amazon was first in us. We were first in Europe. Most likely the question is why autonomous stores?

Jacob: Yeah, what was that pivot? Which by the way, all of those sound like great use cases and especially the one I’m like the military contracts and stuff. I’m sure there’s some people out there who would still want that version of the technology maybe now. But yeah. What was the what in 2019? Was it just that you saw this technology? The best use case for it was autonomous stores. The biggest market was to put it into autonomous stores. What made that last pivot.

Mindauges: Generally is like a shopping experience. The traditional shopping experience in grocery stores with old data compared to advancements in other fields. We have seen improvements in cars, airplanes, TVs, phones, laptops, music streaming everywhere. But when you go in the grocery store, you still stand in the same line, like you put the same items on a table as that most likely some people did 50 years ago, just nothing changed. So we decided, okay, this enormous field should be disrupted. And like we have already the super powerful system for that by improving them step by step from like a phones defense parking and then like a real time image processing and information processing in grocery stores.

Jacob: Yeah. So our grocery store is obviously where you started. Is that the. Best use case that you found or which are there any other types of stores that you’re on the early list of places you would want to go and use this technology in.

Mindauges: Food and grocery retailers to this enormous local business? Now we are in this part. Most likely we get requests from other fields, but like for next two, three years, most likely we will stick with grocery retail.

Jacob: Yeah, because I’m guessing it’s quite a large undertaking to map out the specific types of products and the flow of the store and everything else to be able to work with. And then you said you were the first in Europe and Amazon was the first over here where I am in the States. Are there a lot of competitors in this space or are you not just the first, but one of the only ones working on trying to build this sort of system?

Mindauges: Yes, we’re competing with some of them are based in the US, in Silicon Valley, some in, in Europe. We don’t see much competition from European based companies. So there are very few players which are based in Europe. Main competition comes from the United States and Israeli based companies.

Jacob: Yeah, makes sense. Walk me through then. So how it works from what’s it look like from the customer perspective from when they enter the store? What is the full process look like?

Mindauges: It’s quite simple to use. It’s even comparable to users of London Underground or New York Subway. You just tap a card when you’re entering the store. You pick items, you return items back to shelves, and the system automatically tracks your actions and automatically creates a virtual shopping basket. When you’re done, you approach the exit, you tap a cart again, you tap out and you exit the store. That’s it. Like you tap in. Everything in the middle happens magically, automatically. And the part you can download app later as your regular customer. But we made the experience as seamless as possible to anybody. Anybody like can enter, anybody can exit and no big deal to explain 30 minutes how to enter the store or how to create an account. We started in the same way as Amazon did with an app that like in 2020, 2019, we learned it’s very, very difficult to convince people to do the app when we just want to buy a sandwich or coke or water, like when we go, okay, thank you and go to the next store. It’s easier with regular customers, but still most likely back when losing 90% of possible our customers and learn that even Amazon created this way. Most likely it’s not the best way items in grocery stores.

Jacob: What is the alternative? That what exactly are they tapping in with now? Is it a specific card or app that’s like a Pixevia card or app? Or are they tapping in with whatever payment method they want to eventually use? What are they swiping in with?

Mindauges: You can tap like any like Visa, Mastercard, Amex, your phone, your watch, any payment method which can be used later for payment. It’s you don’t have to download special app or to have special card. You use the same way to enter as you pay in Starbucks or any store like any place.

Jacob: Yeah. And I liked what you said earlier about the for kind of a visual for those who haven’t been watching all the videos and everything online like I have equate it to the train system here in Chicago, in the city I live in. As far as the little kind of turnstile stall, when you walk in, you just swipe and you’re allowed in, you swipe in, you’re allowed out on the other side. So this is obviously makes a very, as we said, as frictionless of a experience for the consumer is I can really imagine. I don’t really know what the next layer would be or what friction is really left at this point. But on the retailer side themselves, what are the benefits that this is driving to the retailer?

Mindauges: Well, a number of them I just mentioned a few because they can take like our full presentation.

Jacob: Yeah, the biggest ones.

Mindauges: The main ones were store can operate for like extended hours, like including late into the night. And some countries it can be also open on Sundays because in some European countries Sundays are closed and only autonomous stores operate. So you have the first shift. You can serve your customers way longer. The second thing, you don’t require employees working early hours or late hours. You can have just one shift to replenish the shelves and that’s it. The checkout can work completely autonomously. You also have extra security features like management features. So like you can have one person which can fully work like in the store and like enough one person to serve quite big store. Other things, better shelf monitoring. They know when items are not available on the shelves in better inventory management data analytics, especially like this part, you can rotate stores in like a three dimensional ways and see what’s happening. Like in real time or not in real time was like an enormous media attention. That’s super important and a lot of momentum. Things like reduced waste, reduced shrinkage and at the end, like also reduced theft.

Jacob: Yeah, I’m sure there’s even more like you said. But yeah, that’s removing a lot of overhead, removing a lot of cost, finding a lot of efficiencies within the supply chain of it all, especially having worked in a lot of retail, the our inventory was always a joke before. Like it’s a running joke in almost any retail. You don’t need to actually look at the inventory because it’s going to be wrong and that if you could actually have accurate inventory, how much that would help. So how accurate is all of this, I guess would be one of the first questions I would ask. So when thinking about like it tracking the inventory or it tracking the sales, what is the kind of accuracy level of the system? Is it 100% across the board? It gets everything right. Is it dependent on certain store setups or something that one employee is doing or not doing? What’s the accuracy levels and have there been any hurdles in trying to work to improve that accuracy level accuracy?

Mindauges: Super important topic. It’s really accurate Right now. Current systems are really accurate and reliable. We talk about 98% accuracy, something like 99 that’s a bit lower when you start a new store. It requires some or the system to gather statistical data to learn a little bit. But when you have a really accurate system, it depends on many factors. Also how the store is operated, how accurately items are placed in correct locations, because sometimes items are placed randomly. So in this case it’s more difficult to work. So it depends on operational model. But like were are right also corner cases some. Things which considered difficult is less difficult. For example, when you put. Item back or even you put item back at the wrong location, but were different things. For example, I know you have a dog which moves items and system tries. Okay. Like I. Don’t see any person nearby like and something is still moving. What’s happening here? You have a child which parents faults like system recommends. Okay. I see a person with four. Hands. What’s happening here? I’m not trained like a person with four hands. Like. Whereas. Some like a very long tail of unusual cases, which are super rare. Like a number of. Cases on which we day to day improve the system.

Jacob: Yeah. And the more knowledge you get, the more data. You can bring in about learning about those. I’m sure you can just make them better and better. But you nailed the couple examples I was thinking of. Yeah, if I pick something up. But then I’ve definitely been that person who by I get up to the register, I’m like, I don’t actually want this and maybe I’m a bad person and I don’t go all the way back to put it back where it actually went. I just leave it sitting somewhere. And so good to know that, yeah, the more you get to experience those unique situations, the more the system learns to what to do about them, how to handle them. The other big area within retail, and especially has been here in the States, has become a big topic of discussion lately. Bigger than it always is. I don’t know, maybe in Europe it is similar, but retail theft has been on the rise in the pandemic and post pandemic world a little bit. So how does an autonomous system like yours handle theft in any way? Is it actually part of me would think without any knowledge that it might be better at handling it, because if someone’s trying to hide something in a code or whatever, the system would maybe pick up that they picked that item up and took it out of the store with them. But how does your system kind of address and handle theft?

Mindauges: Yes, it’s.

Mindauges: Much more challenging to steal from autonomous tour. Like every movement of item is tracked, every movement of person is tracked. From our experience, the ways in which people steal items are quite predictable. Yes, we know which items are high risk items and we have specific alerts when pageants. When you see the system sees like a number of high risk items, they can super quick. So we instantly trigger alerts and security companies. I start to monitor what’s happening in Westworld. So like, we learned a lot of things and the system can even predict some events which can happen in very quick time. So like advantage way more challenging to deal with AI driven store. And when you have a hundred of cameras compared to a regular store.

Jacob: Yeah, for certain. And that’s I mean, again, back to my own retail experience, it was mostly that even if we knew someone might be doing something, it was that we didn’t have cameras or anything. And cameras are the backbone of your system, so it would make sense that it actually improves upon that and makes it If I was someone wanting to steal something from a retail store, I probably wouldn’t pick the one that is operated by a bunch of cameras looking at everything and tracking everything. With all of that tracking, though, with all the cameras, the natural other topic that would come up, I would think would be privacy concerns or potentially some sort of pushback from customers or maybe even specific countries, maybe with different rules around what can be monitored, what types of consents have to be given, what are, if any, of privacy concerns? Have you experienced any pushback from consumer bases on? We don’t like the idea of walking in a store and being tracked by all these cameras at all times, or is part of using a store like this that there is some sort of consent that is given. But swiping in is part of that is giving your consent to be in this ecosystem and be monitored.

Mindauges: Yeah, like.

Mindauges: It’s an important topic. And as you know, the European Union has the strictest law and regulation regarding privacy and personal data, and these requirements are strictly monitored. But in reality, we know very few things about each customer. We know a lot when he is in the store or she is in the store. But when you walk in the second time, for you, for us, this person is completely new idea, completely new identity. So like the customer journey is completely anonymized. Videos are stored for very short period of time. So from data privacy perspective, the store is not much different from any other stores where security cameras are installed. But yes, people sometimes are curious. These cameras are not. So I would say visible. All stores have cameras with cameras and we explain a lot like it’s different store like it’s written. It’s like a different store. And I would say the new people realize 90% of cases when you go to the exit and we see, okay, like everything is great. Wow. Okay, what happened? And only when we try to see around, okay, like not like resources, some some magic happening here. Okay. Like I see more cameras, but like, normally when you walk in the store, you just stop and you now have to tap into different things. Different sanitation system is normally no questions here, just the questions are wondering starts when we approach to exit and see what everything is complete. We don’t have to do anything. Just press one button tap and that’s it.

Jacob: Do you see it being able to stay that way? Because my kind of business brain goes to that. This current version sounds great and sounds like as a consumer, I wouldn’t be concerned about privacy at all. But I can think of some like marketing teams out there that might be dying to come to you and say, Well, if you do, let us identify the person or match your system up with this data, we could give offer them a notification that’s this is on sale and we know you like this or whatever. I could just see a future version of this. They’re being marketing people out there that are like, We would really like to join forces and bring our data and actually identify these people. So have you been approached ever at all yet by of anyone wanting to work with you or possibly go further into break out of your current model? That does sound like privacy is on lock and are trying to get you maybe to take it a step further and start identifying people or doing anything, because I think the negative part of my brain naturally goes well. At some point in the future, there’s going to be people that want to do that version of it.

Mindauges: In this case, also rather simple, the explicit consent is required. So like for most people, even also for me, it’s I would like to have better advertising, like a better suggestions. And in this case, if you agree, yes, in this case, retailers can join some information on online purchases that like will show the value written like this, information collected and also consent given like a join and operate on this information.

Jacob: Yeah certainly. And I’m actually someone who like you or what you just said of there’s many places in times I’m actually fully wanting that, especially like in a grocery store or something. I would be totally fine if I got to give consent and say, You know me, you can remember me and track me and tell me when that thing I buy once a month and is on sale and maybe I should buy more of it. So I think, yeah, as long as there’s consent given, there’s going to be some a lot of times where a lot of consumers are like, Please tell me if the things I buy are on sale or in any way that you can help me or save me money, anything like that. So the last couple things I want to ask you here then at the end is you’re obviously an AI powered company and this year, 2023 has been the year of AI and machine learning and a lot of other new technologies. The big first wave of mass adoption and not maybe even adoption, but just of consumers and regular folks knowing about and hearing about and thinking about all these new technologies. Where do you think we’re at in the adoption curve of actually learning how to properly utilize and harness all of these new tools? Do you think we’re still very early? Do you think we’re somewhere middle along the way? Where do you think we’re at and actually learning how to utilize all of these new tools and that kind of adoption curve?

Mindauges: Yes, like.

Mindauges: What we learned. But the biggest challenge is not technology itself. The biggest challenge is to extract the value from the new product, and it’s often necessary to make subtile changes or bigger changes in processes. How the store operates, how digital enterprise is, how security managed and so on and so on. It’s like similar when you buy a car, but you use as a horse like you apply same habits, things to use as you tool you had, you know, centuries or like decades ago. In some cases it’s required like a change organization to change processes or get the full value from AI driven tools. I produce data and make enterprise really digital. Here I see the main challenge not about the AI or product itself or ability to implement changes in organizations and also ability to deploy and operate way more in way more digital way.

Jacob: Yeah, fascinating. Last question then for you. When you kind of look maybe the next five years or so of the company and where you might want to take this technology and expanding its use, are there any types of stores or types of products or specific markets that you’re hoping to get into? What’s next for Pixevia? What’s the scaling plans look like with this technology?

Mindauges: Right now we are already scaling into huge markets, Europe and the United States. Now we have plans to open stores in Middle East and Asia Pacific. We have plans already in deployment this year alone. We will open more with ten stores. We joke that we open most stores when Amazon closed through last two years to keep industry in balance and for next year, our plans goes from like a dozen stores to hundreds of stores. Most likely Agnes Mention mentioned markets Europe, the United States. Middle East. Asia. Asia Pacific. For us, the preferable formats are to gas stations. That’s a bit smaller stores and slightly bigger so-called neuro precise stores in city centers, which are a little bit smaller than supermarkets, but like a little bit bigger than gas station stores and have like a proximity stores close to your home, which can serve most of your items. But it’s not a huge supermarkets like here our plans and like now we deploying step by step in major world markets.

Jacob: Yeah makes perfect sense and yeah all the places that the stores that I don’t imagine us moving to too much e-commerce the places we’re still going to actually go to and visit in person, which is one of the reasons why the grocery store has the kind of first base concept makes so much sense. Well, Mindaugas, this has been a real pleasure for those listening who may want to follow you or learn more about Pixevia, keep up with everything you and the company have going on. Where would be the best place for them to go to do so?


Jacob: Mindaugas. Thank you so much for your time and knowledge today. I’ve greatly enjoyed it and hope to speak to you again sometime soon. If you enjoyed this episode and want to hear more, head on over to to subscribe on your podcast listening platform of choice. That’s s o a r p a y .com/podcast.