The Future of Retail AI & Edge Computing with Marc Del Vecchio (Supermicro)

Scott Mann

Welcome to PodMagic, real conversations about solving real IT challenges. I’m your host, Scott Mann, SVP Global Sales at StorMagic. And here on PodMagic, we’re always exploring how simple, reliable technology can benefit you and the people you serve, whether you’re running branch offices, retail stores, or supporting customers on the front lines. Our goal is always to bring interesting guests, deliver some value, and have some fun along the way. 

Today’s guest is all about what it actually takes to run AI where retail happens. So, not in the data center, three states, three countries away, but right there on the shop floor. Since joining StorMagic, I’ve had the privilege of working alongside today’s guest, and there is no person I can think of that would be more suited to talk about retail technology on a podcast than this man, Marc Del Vecchio. Marc, welcome to the podcast. Great to finally have you on.

Marc Del Vecchio

Thanks, Scott. Great to be here. Thanks for inviting me.

Scott Mann

Our pleasure, Marc. Really our pleasure. Marc, just for some background, is a Staff Solutions manager at Supermicro where he helps retailers figure out how to run AI directly in their stores. From computer vision to real-time inventory tracking, and the goal being without drowning their IT teams in complexity. It’s that hands-on experience across so many different retail environments, which is exactly why I’m so excited about today’s conversation. And without further ado, we can get into it. 

Let’s start with where everything stands today. So retail, we know it’s experiencing another major technology shift. What is driving the renewed focus on the store as a strategic technology platform rather than just a simple point of sales?

Marc Del Vecchio

What retailers are realizing is the wealth of data that they have at their stores. On the e-commerce platforms online, they get all that data from the purchase online, they get the cookie data off the pre people’s browsers, whole bunch of rich data. But now they’re starting to realize that, wait a minute, there’s a ton of data I am not collecting or doing anything with at the edge. So that realization coupled with the technology is now caught up to the idea, is driving that big push.

The thing too, is going to be budgets. Retailers want to invest in new tech, AI initiatives, and they find that that basic workload consolidation of taking a whole bunch of disparate hardware that runs their stores and consolidating on that, containerizing it, consolidating it onto a common platform has real value for them on the OpEx side. Much easier to maintain more uptime. Reducing that footprint on site, that IT footprint on site is really where they’re seeing the savings. When you have purpose-built edge servers, not just taking a random workstation with Windows XP and running your whole store off of it, which unfortunately some still do, putting purpose-built edge hardware, like hardware from Supermicro out there. Putting it in the back of the store, putting into high availability clusters, running virtualization platforms like StorMagic are really the golden ticket to taking the store, making it smart and making it operationally efficient for them at the edge.

Scott Mann

You mentioned e-commerce and the reality is a physical store competes against “why wouldn’t I just buy it online”. And margins are so thin. Operational efficiency is key and then that customer experience of why should I come to the store as well.

Marc Del Vecchio

Yeah, a hundred percent. You really hit the nail on the head that retailers operate historically in super thin margins. They need to be as lean and mean as they can. They’re looking what their competitors are doing, they’re trying to keep up with them, they’re trying to make themselves more efficient. It’s a constant game of finding what the next thing is, implementing it, but implementing it efficiently to hit that ROI target that initiated the project to begin with.

It’s a lot of new stuff for retailers who really in the past have said, Hey, it’s not broke, don’t fix it, kind of attitude towards technology. They’re realizing that if I keep with that, I’m going to be left behind. When you’re a retailer and you’re not keeping up with your competitors, you’re probably not going to be doing anything much longer because you’re going to fall behind and eventually it’s not going to be a good end for you. You have to keep up with what your competitors are doing. You have to keep up with the times and you have to modernize.

Even if you’re not looking for some crazy initiatives, like all this great AI use case I want to implement, it’s just keeping up with your OpEx expenses. The longer you run on old hardware, the more problems you have and the more likely your store can have an issue and the system goes down in a physical location, you may not be able to process payments or really service your customers, which is definitely not a good thing.

Scott Mann

That’s interesting. It’s like a different culture shift. We know the driver behind it is, as you mentioned, it’s AI. It’s become a boardroom conversation across not even retailers, across all industries. Where are you seeing that retailers are finding the fastest and most measurable returns today?

Marc Del Vecchio

That’s a key way you put it. What’s the return? Because again, those thin margins, they aren’t going to do projects, they’re not going to chase initiatives that don’t have a clear path to ROI. What we’ve seen over the past few years, and I always say that I think retail, the retail vertical has been the biggest adopter of edge AI out of any of the verticals I work with. And that’s because they’re seeing the dollars saved by it, or potentially more dollars earned by it. For what we’ve seen to date, loss prevention by far has been the most common implementation for AI at the edge. And again, it tells that story. Before I put in the system, my shrink at these locations was X. I installed a loss prevention system. Now it’s Y. That is just a clear, nice pencil path right to what they’re saving by implementing the technology. 

Now working with some of our partners, they’re talking about like a grocery store, we have self-checkout lanes, which are extremely popular now. You see them almost any big box store has self-checkout. Even it’s even trickled down to small or mid-sized stores as well. We have vendors that we work with saying basically, you can save almost six figures per checkout per year by implementing their solution. You’re a grocery store, you have 20 checkouts. That’s not a trivial amount of money. That some real savings just by implementing that technology. 

Another place that is starting to adopt AI within retail is the supply chain side of it as well. Starting to automate a lot of that process, the decision making is another place where, it’s not as big as loss prevention by any means, but we are seeing that it’s starting to become a more and more popular use case. We’re seeing ISVs pop up that are specializing in just that particular use case. It’s coming. I think it’s going to be one of the next bigger things for.

Scott Mann

I guess that goes everywhere from the actual retail floor all the way back to the warehousing and everything and order processing.

Marc Del Vecchio

Absolutely. Because to them, it’s a holistic picture. It’s not just, “hey, what am I doing at my retail location?” All the supply chain, the warehousing, distribution, all that that goes into to stocking that store is hugely important and has potential for real savings if you do it right.

Scott Mann

The shrinkage is probably more tangible of a number to direct the immediate savings there. And then these are other things that eventually will come. So that makes a lot of sense. You start with your biggest gap in your margin hole. Is that shrinkage? 

I know you’ve spent tons of years working with computer vision in retail environments, which use case is evolving now from experimental projects into business critical applications?

Marc Del Vecchio

I probably started doing edge AI workloads about 2016. Definitely I’d say we were ahead of the curve on it.

Scott Mann

Probably wasn’t called edge AI then, was it?

Marc Del Vecchio

No, there wasn’t really too much of a name for it. We were just doing AI work. It was really called computer vision AI, which it still is, but we were really focused on security use cases. We were focused on facial recognition, identifying people, repeat offenders for shoplifting. We were looking at object detection, whether it be for weapon or before the beginnings of loss prevention. Very, very basic. But a very common use case at the time, and again, it’s probably dictated by the hardware that was available, not a whole ton of processing power, it was demographics. Who’s in your store? Who’s walking by this digital signage, male, female, age ranges, that kind of thing. What’s my impression count? In 2016, telling an advertiser, hey, in the real world, digital at home. I can give you an impression count like you get from side banner ads on your browser. That was mind blowing to them. No one had ever done that before. We focused a lot on the demographics for digital advertising. We were starting to play around with programmatic advertising based on who was in front of the camera. So doing like a real time bidding system, which occurs on your browser every second you’re on a computer, but bringing it to the real world, which was some really, really cool cutting edge stuff. We were probably at the time too far ahead, which is a thing, you can be too soon with an idea. You know, we kind of outran the hardware. But one of the things we often say is we want to turn your dumb cameras into smart ones. Let’s take those feeds that are really forensic and let’s make them proactive. And that was kind of like our tagline. It definitely resonated with people.

Moving forward 10 years, now technology is advanced, I’d really say probably the most direct descendant of that is going to be store analytics. Again, your crowd counting demographical information. It’s going to be loss prevention. Once that got perfected with detecting objects accurately, understanding the difference between two objects, tying in with the POS system, which is very important, that’s really where it culminated and became like a very robust product. It’s probably the most widely used AI use case at the edge right now in retail.

Scott Mann

That makes sense, and the data exists, how do we use the data? We’re evolving and everything’s AI now. We went from human input data to now in environmentally generated data, which obviously generates significantly more data. 

What challenges does that now create for retail IT teams like you’re managing, not just at one location, but managing it at hundreds or thousands of locations. 

Marc Del Vecchio

Tens of thousands, depending on who you are. Some of our customers are tens of thousands of locations and you’re absolutely right. It’s the pro and the con. The pro of, working on the data where it’s generated is great, but the data is where it’s generated, which is all these disparate locations. Really, data security is top of mind.  If you’re collecting information, especially if it has PII in it, where you can identify the customers, say in a video feed or its analytic data, that all that data has to be secure, has to follow the same IT standards that company follows in the data center. In a larger enterprise organization, you’ve got to do that in 10,000 locations now. And you have to have a solid plan around that. Data security is a big one for sure and then simple data storage. 

Going back to what I said about retailers, they would live on a platform until it died. It’s really, if it’s not broke, don’t fix it kind of thing. The thing about systems that were put in seven years ago, eight years ago, none of this AI stuff was around. No one thought about it. They would put in a small four-core processor to run an entire store and it worked at the time. Minimal amount of storage. They tried to make the hardware as cheap as possible to show a good ROI. Well, fast forward, and that hardware’s not really well suited to the modern use cases we’re talking about today, like AI. So simple data storage even can be an issue for them. It really pushes a lot of these retailers where they may think, “hey, I’m going to try and do this all on the same hardware like I’ve been doing for years”, but a lot of them are going to be pushing the realization that it’s time for a refresh. It’s time to refresh my infrastructure, both the hardware and the software stack to accommodate my goals of what I’m trying to do now in my locations.

Scott Mann

You went from a world where there was data that was being collected that wasn’t mission critical. It was okay to be down and offline for hours.

Marc Del Vecchio

No one was looking for it. “Hey, where’s the data from store three twelve?” No one did anything with it. POS data is different, but really anything else, it rolled up to a port maybe somewhere in the back office. It wasn’t critical like it is today.

Scott Mann

The assumption is everybody goes in the direction of, if you had to store data locally there, you had a small low compute system there, and now the first thought process is, what about cloud? Many organizations across all verticals assume that, cloud is the answer to every AI initiative. 

We know that’s not necessarily the case and it’s not a one size fits all, but why aren’t more retailers considering where data should actually be processed? And I think that’s the key to it, where should it be processed?

Marc Del Vecchio

Yeah you’re seeing that shift for sure. Because as you look at how cloud infrastructure got built out, starting with AWS, it started that pitch of, “hey, don’t own your own equipment, come to the cloud, process everything up here. Don’t overbuy. It’s elastic, it can expand as you need it to.” Which is great and cloud definitely has a place. I’m not going to knock any of the hyperscalers. One, because they’re very large customers of ours, but two, they have a solid product and there’s a use for it. Absolutely.

But when you look at what you’re doing with AI inferencing, there’s some unique things that start to break that model a little bit to me. One is going to be privacy concerns. We talked about data security a little bit ago. Again, if you’re collecting information, especially video feeds where you can identify people in there, there’s a lot of laws and there’s a lot of reasons why you have to be careful in in doing that. You got GDPR in the EU. In California, you have CCPA. In fact, from when I used to do this, the United States is tough because basically you have fifty individual little countries and they all have their own rules around privacy. Some have none, some are very strict. Some states someone needs to give written authorization for them to be scanned by AI in a video. It runs the gamut. You have to be careful with that, but a lot of times that you can comply with that, 

is if it’s for a security use case. A lot of things get exempted when do that, or you process it locally. If you say, “hey, none of this video is going to leave my site, it’s all processed here, and maybe some metadata goes up that’s anonymized.” That’s fine, you’re staying on the right side of the law and you’re good. 

A lot of these use cases too, again we talk about computer vision or other data, it’s latency sensitive. It needs to be real time action. In the case of loss prevention, if someone’s hiding a steak under a magazine and swiping at the at the self-checkout, you don’t have time to delay in reacting to that. You need to notify the store staff immediately, “hey, there’s an issue at self-checkout six, someone needs to go over there right now.” So that latency can come into play with the use cases, which is another reason to infer on where the data is collected.

The third one is bandwidth. Bandwidth is out there, it’s cheap, but if you look at a larger store, I got sixty, eighty, hundred HD video streams to analyze. How are you going to send that up to the cloud without having a huge internet pipe? Which could be cost prohibitive. It may not be available in some of these locations, may not even be an option. You go into the more rural areas, that pipe might not be an option, or it may not be reliable.

We take internet access for granted in the US and we have a great infrastructure here, but even some places the connection can be spotty. You still see gas stations or some stores, look out back and you’ll see a satellite dish. They have backups like that because they can’t be down. They’ve got to process payments. So that tells you, even though they have like a hard wire connection to the to the retail location, they don’t trust it fully because sometimes it does go down. If you’re running critical AI workloads at the edge, they’ve got to be run at the edge. You can’t really have that cloud dependency.

Scott Mann

Yes, it’s more of a partnership and not a one size fits all, here or there, cloud or on prem.

Marc Del Vecchio

No. The other part of it too, probably the most important is cost. Processing the cloud, there’s always going to be an inflection point, where it is cheaper to run it local or run it in your own infrastructure. When you talk about AI workloads, what the charges are for running, that taking up GPUs in the cloud, that inflection point is a lot sooner than people think it is. Again, retail, cost conscious. If they can put out some CapEx to save on OpEx down the road, that’s something they’re willing to hear.

Scott Mann

Yeah, and that’s exactly where my head went around costs. It depends on the workload, where that workload needs to run, regulations, all of that. Also, where’s it going to be most performative but most efficient to them on the bottom line. All of the heavy lifting is really done in the plumbing of it all. And the big buzzwords in the market that we come across is a lot around GPUs and containers. But there’s a lot of hype versus reality. How are these technologies clashing or collaborating in production right now?

Marc Del Vecchio

Infrastructure is absolutely key for AI. You can have the best trained models in the world, you can have the best ideas, but if they’re not going to run properly, that’s a problem. Talk about model training. Model training happens in the big data centers, everything you hear on the news, a lot of the buzzwords coming out of that. But the inferencing is happening at the edge, so it isn’t really far-fetched to say that the models are only as good as the infrastructure they’re going to run on.

Again, if have this great model, but you have an unreliable infrastructure, it’s not powerful enough, or doesn’t have the right connectivity, or whatever it may be, what are you really doing? You’re not taking advantage of that investment you put into the large model training by not doing the inferencing optimally. You got to have the right equipment, combination of CPU, GPU, the software stack to really make that model effective for what you’re trying to do. 

Given the distributed nature, like we’re talking about thousands, tens of thousands of locations for these guys, remote management is a key concern as well. So, managing that infrastructure remotely is important because you’re not going to have an IT person at every location. They might be the other side of the country. You got to have hardware and software that’s remotely manageable, that can be imaged, that can be rebooted, can do all the basic troubleshooting steps, or maybe some more advanced ones remotely. Talking about cost, when they send a technician out, typically called a truck roll, that’s five, seven hundred dollars just to pull up and walk into the store. Not your diagnosing time, not your remediation, any of that. Those visits add up quick. If you’re looking at cheaper hardware, like it’s half the price of something that’s enterprise grade. Sure, but you’re going to recover that by not having to do that service. That service visit will make up for that difference. 

Never mind, what does that downtime on your business cost you? What’s the impact to your sales in that location? You add those two together and you’re going to quickly realize you need edge purpose designed hardware, and you need a software stack that is meant to live at the edge, be highly available, be largely hands-off, but have the ability to be remotely administered when needed.

Scott Mann

The answer’s not just another data center. There’s other considerations to take because of the nature of how distributed these can be.

Marc Del Vecchio

Yeah, absolutely. And we take a lot of that data center mentality, because data centers, you may have staff that have smart hands, whatever the data center provides. You still have that capability to remotely manage the hardware and you know perform those tasks. So, we’ve taken that mindset, that technology that we build into our data center systems, and we’ve used that in our edge system. At Supermicro, we don’t make

Consumer products. A lot of people haven’t heard of Supermicro because there’s no Supermicro laptop or desktop or anything like that. We don’t make monitors or anything like that, we only make servers. Even if it’s a smaller form factor, may look like a desktop, that’s a server, has remote management capabilities just like a big server does in a data center. Having that purpose built edge hardware, to us is very key, because if your infrastructure isn’t optimally configured, that model that you paid all that money to train and create, it’s not going to perform well.

Scott Mann

I imagine if you’re looking at the larger we go, the more we scale up, larger multi-site retailers, you’ve probably seen big mistakes that organizations make when they’re scaling some of these initiatives. A lot of it comes from this pilot environment where everything’s running swimmingly. All of these things that you’re talking about are all factors in distributed environments that might not be in consideration, but what are some of those challenges you see that they face taking it from pilot to actually execution in the field to the sometimes tens of thousands of locations?

Marc Del Vecchio

The first thing I’ll say, it really isn’t technology related at all, it’s project management. People think AI can do anything. Unfortunately, it’s not a movie. AI cannot do everything. Maybe not yet. Maybe it’s a good thing for humanity. You have to start to get grounded on what you’re trying to accomplish with the project. Have clear goals. This is project management 101. Have clear goals and set out to accomplish those goals. Scope creep is the enemy here. I’ve seen it so many times where, “hey, we’re putting this project in with AI, we’re going to do this.” Can it do that? Or can it do this? “Yeah, we can make it do that. We can make it do this.” And the project bloats. You get to the POC part, then they want to expand upon it and that’s where it really starts to fall apart. Because a curated POC where everyone’s got eyes on it, IT is watching it, the project owners are on it, the software vendors are in there, making sure it works. If it’s not truly scalable, it’s going to fall apart when you get to a hundred stores. Never mind, thousands or tens of thousands of stores. I’ve seen a lot of projects fail because they don’t really define what they want to do. It becomes the art of the possible. These fanciful ideas, which sound really cool, but in reality, just don’t work at scale. A POC and full deployment are two very different things.

I would say that’s first. Inherently, being able to scale is of course going to be required. You got to have that right software stack, hardware underneath it that can be remotely managed, that can operate without constant attention, because in the end, if twhat you implemented needs constant attention from your IT resources or vendors or whatever it may be, the project’s probably going to fail. Because it’s going to miss that ROI point it was initiated to solve to begin with.

Scott Mann

Scope creep, it’s the biggest hurdle we all run into. Defining and knowing exactly what you’re trying to accomplish at the end and not seeing all of these other nice-to-haves along the way, that all of a sudden make that end goal a lot more difficult to achieve.

And obviously there’s the constant pressure of reducing costs. You want to improve the experience, but you want to reduce costs at the same time and there’s sometimes a bit of a balance of that. But how can modern edge infrastructure help achieve both of those goals simultaneously?

Marc Del Vecchio

It’s going to come down to one, having the right infrastructure. I know I keep harping on that, and I know my company’s infrastructure, we make servers. Truly, you got to have the infrastructure right, the plumbing right, like we talked about. You have to have that right. If you have the CapEx, you put in the right hardware, you’re going to see the return on investment in your OpEx. Less downtime, less surface visits, better products being run, your initiatives running like you wanted them to. Having the right hardware, the software stack is going to be key to making that happen. That investment in new hardware, it’s good for today, but it should also be future-friendly. I never say future-proof because you don’t say it’s bulletproof, it’s bullet-resistant. Future friendly. 

We’ve had customers, especially when it started to kick off where people were going from, “hey, you know, I run my whole store off a little four core, eight core edge box.” Fine. Then we started seeing customers say, “I only need that much power now. But I know in the next two years, I’m watching the industry, I know what’s going to happen is we’re going to start doing some AI workloads at the edge. And I want a system where I can upgrade.” It has the headroom to do that. I don’t want to in two years rip and replace, you know, 10,000 boxes because they are not upgradable for AI. I want a system now that can do that.

We had one or two that were kind of like forward thinkers in that. And now it’s becoming the standard where we’re seeing people, even if they’re not using AI, which a lot are, they’re ready for it. They have the ability to pop in that GPU or whatever accelerator they want to use to make AI happen. That’s another way we’re seeing, CIO’s, leadership making the right decisions on the infrastructure side, laying down the foundations for what comes next.

Scott Mann

And not thinking of the foundation as just an extension of your data center. It is an extension of your data center, but you don’t just take this and put it there. It’s not necessarily built for the same use cases. I know some of the systems that we work with you guys, they’re custom built for these edge environments. Some things you might not be considering, and do I use the same processors at the edge versus what I have in my data center? All of that stuff comes into the combination.

Marc Del Vecchio

It’s a whole other bowl of wax when you think about data center. It’s something I talk about a lot. Data center, you got nice raised floors, air conditioning, humidity under control, secure areas, which is amazing for compute, of course. The edge is the polar opposite of that. You talk about systems hung on a piece of plywood in the back of a manager’s office. Some systems we find them, they put them in the kitchen by the fryers, like really environments that are not very friendly to electronics, never mind advanced electronics. That’s why you have to have that edge-built hardware. You take a workstation, and now it’s a server and put it on a wall, it’s probably only going to last a year or two before you’re going to start having some serious problems because it wasn’t meant for that environment.

Scott Mann

Absolutely. Let’s put the forward-thinking hats on. If you were advising a  CEO that’s planning the next three years of technology investments, what would be the one capability that they cannot afford to overlook today?

Marc Del Vecchio

Edge infrastructure for sure. Edge infrastructure is key. Invest in that foundation. Don’t try and just rip and replace what you have with something similar, just a little modernized. Think ahead. Your locations are becoming little data centers. We’re talking about high-performance compute. We’re talking about GPUs or other AI accelerators, data storage, power management, security. These really are becoming small distributed data centers. You have to treat them accordingly. You can’t just think, “hey, we’re going to throw this appliance on the wall and see in seven years.” That’s not the model anymore. You have to treat it like an enterprise location, which it is. Think of it as an enterprise location for IT as well.

Scott Mann

Agreed. That’s the CIO, the direction or the advice for them. Now from the crystal ball thought process from five years from now, what will define what a truly intelligent retail environment looks like and what should retail IT leaders be doing today to prepare for that? That five year, what’s it going to look like? Let’s look back at this question in five years and laugh at how off we were or right we were. We’ll see.

Marc Del Vecchio

Well, first of all, anyone who answered that question in 2015 was horribly wrong with whatever answer they gave. I just want to make that clear. What I see in five years, some of the things you can see is like omnichannel. That seamless transition between in-store and online purchases, I think that’s going to grow. It’s most certainly there now, but it’s going to get more functional, a little less clunky. For example, I got a couple items, I go up to the register and say to the cashier, “hey, I wanted that jacket, but they didn’t have it in my size.” They should be able to say, “okay, no problem. We’ll get your size, it’ll be in your house in four hours.” No logging in, no additional payments. I just swipe right there, pay. It’s like it was in my basket, but it’s going to be at my house in four hours. So that omnichannel experience of really blending brick and mortar, with online, with e-commerce, I think is going to be much more prevalent and much smoother than it is now, because really that doesn’t happen. You’re out of stock. Okay, well go on our website and order it. Well, no I’m here, I want to buy it now, I’m buying other things. Omnichannel, truly. And back and forth too. Hey I really want that jacket, but I leave in two hours for a trip. You can’t deliver it to me in time. Okay, where you’re going? Okay. An hour into your drive this store has it in stock. We’ll pull it aside for you. All those kind of things that make sense when you verbalize it. Those things can become reality.

So, I think omnichannel is one. Supply chain, like I mentioned, I think that’s another one that’s going to be huge. And it’s growing now because we talked about loss prevention and why it’s so prevalent is you have that easy path to ROI. X, Y. If I can say I can make my supply chain logistics three percent more efficient, that’s a number I can calculate. And for some retailers, that’s a huge number. Being able to realize that efficiency is huge. Never mind the world we live in of tariffs, supply chain issues, shortages, that kind of thing. Think about having an agentic AI system that takes all that into account seasonal demand, forecasting, suppliers, world events, and being able to dynamically issue POs to the right supplier, using the right method of transportation, having the inventory in the right region of the US for when you know it’s most likely to sell. Having that automated is going to be huge and it’s growing fast from where I’m sitting and seeing that. 

And the last one is going to be generative and physical AI. Going beyond the AI chatbot that we all love to hate sometimes with online platforms, looking at physical AI and agentic AI. Something that we’ve seen on trade show floors for years has been a digital avatar. You have a screen, it has a character on there, you can interact with it, it’s a little wonky, it’s a little delayed. Cool idea. The hardware has now caught up to that concept. We are seeing companies that we work with now start to deploy these at scale in retailers. They have a screen, you walk up to it, you can ask questions. Imagine you’re at a mall, you walk up to one of those directories, instead of just being like a static paper map, now it’s an AI avatar. Instead of trying to trying to find the sporting goods store, it’s B6, where’s B6? Where’s the little dot that says here you are? Talking to say, “Hey, I’m looking for a baseball bat for my son, he needs this size bat, where can I go?” They can say, “Hey, these two stores in the malls have it. This one is having a promotion on it. Do you need directions to it? Here’s a QR code that’ll walk you through the mall to get right to that store.” And that exchange can occur in any language. So now language barriers aren’t a problem. I don’t need to print something in 10 different languages. The avatar can sense what language you’re speaking and talk to you in your native language. So just making shopping easier, making commerce easier for retailers is one of the next big things. It has just such a cool wow factor, which is why you saw it at trade shows and labs and whatnot. The hardware has caught up to that idea, it’s truly amazing. It’s really, really cool stuff.

Scott Mann

Those are all pretty cool examples, not what I was expecting at all, but not surprised with your experience. How many times have I walked into a store where I’ve been like, they didn’t have my size? Forget about it, I don’t care. Then I’ll never look back again. For the company, you just lost a sale because you just didn’t have the availability of it and don’t have that availability on demand.

Marc Del Vecchio

In the e-commerce world, that’s abandonment. You abandoned the item in your cart for whatever reason. But it’s the same concept. This customer was interested in it, but the transaction didn’t finalize. Why? For whatever reason. But now, like you’re saying, in the store, you don’t have to abandon it. Sure, we don’t have your size, but it’ll be in your house tomorrow or in four hours, whatever you need, no problem. Just swipe it with the rest of the stuff you have in your basket.

Scott Mann

Yeah, forget about that out of stock stuff, now even online you get that. Then you’ve got your operational efficiency which is huge for the brand. We’ve talked about all of that. And then the customer experience in general. I have seen in some customers I’ve worked with, they’re even getting into robots on the floor, where you’d ask the robot a question, it would literally take you to what you needed. Pretty cool stuff like that, it’s usually at like one of their hallmark stores, as opposed to everywhere, but I could see that expanding and those are some cool ones. 

Marc Del Vecchio

Robotics in five years, I don’t know. Robotics absolutely are going to play a larger role. You’re seeing them more restaurants, airport lounges, you’re seeing the robots bringing the trays around. It’s starting. Five years, I don’t know. Maybe I’m wrong on that one, but robotics, absolutely. And if you ask Marc for a ten year outlook, absolutely robotics are going to be also a huge, huge thing.

Scott Mann

There’s probably a cost parity there that has to be met. 

Marc Del Vecchio

It’s got to mature a bit for sure.

Scott Mann

Awesome, Marc. I think we’re about at time here. I appreciate it. And this is fantastic. You have some really great insights. I was excited to have you on here, I just took a back seat and let you go and you did nothing but impress. I appreciate it. 

Marc Del Vecchio

Thanks for inviting me on.

Scott Mann

Our pleasure. And thank you all for tuning in to PodMagic, where simple, reliable IT meets real world impact. If you enjoyed the conversation, be sure to subscribe and share, and we will see you all next time.