What's Up with Tech?

From Surveillance To Smarter Stores With AI

• Evan Kirstel

Share episode

Interested in being a guest? Email us at admin@evankirstel.com

Cameras don’t just record anymore—they inform, alert, and help teams act before problems get expensive. We sit down with Jeff Corrall Chief Product Officer at March Networks, to unpack how AI-powered video is evolving from security footage to real operational intelligence you can search, measure, and scale without ripping out legacy systems.

Jeff explains why continuous video analysis is costly and how an image-first approach delivers the wins that matter. By capturing snapshots at set intervals or when simple edge analytics detect an event, their platform sends images to the cloud for deeper interpretation. That means retailers catch blocked fire exits before fines hit. Managers get texts when queues spike. Regional leaders track trends across hundreds of sites in real time. And teams can literally “Google” their stores for spills, cash left out, or outdated signage, fixing issues at the speed of a search.

We also dig into the AI3 camera, a 360-degree device that blends people counting, queue monitoring, and traffic flow with traditional video, replacing multiple sensors with one scalable unit. Beyond retail, transit agencies pair passenger counts with payment data to spot revenue leakage, while banks audit promotional compliance by uploading an example poster and finding any branch where it’s still displayed. The near-term horizon is especially exciting: fusing point-of-sale data with AI triggers to calculate conversion, surface anomalies with visual proof, and notify the right person to take action—before customer experience or revenue suffer.

If you care about operational excellence, loss prevention, and smarter use of your existing cameras, this conversation is packed with practical insights. Follow the show, share it with a colleague who owns store ops or security, and leave a quick review to tell us what you’d automate first.

Support the show

More at https://linktr.ee/EvanKirstel

SPEAKER_00:

Hey everybody. Really interesting topic today as we dive into the world of AI-powered video surveillance and how it's not just about security anymore. Really um doing business in a much more smart way with March Networks, Jeff, how are you? I'm very good. Thanks for having me. Well, thanks for being here. Really intrigued by what you guys are up to. Before that, maybe introduce yourself, your journey, and how do you describe March Networks these days?

SPEAKER_01:

Yeah, so uh well, March Networks has has transformed a lot over the years. I'm the chief product officer responsible for our product plan and our RD departments. Um, you know, we in the market have traditionally been a video surveillance company, but we're very focused on business intelligence. Um we, you know, sell a lot of video surveillance products, but really our customers are using those not so much uh for the traditional video surveillance purposes, but to actually improve their business and learn things about their business.

SPEAKER_00:

Brilliant. And of course, AI is changing everything, and it's also changing the way we think about surveillance in general. Give us context and how has it impacted um the products and your go-to-market?

SPEAKER_01:

Quite a bit. I mean, our um sort of jump into AI started with really understanding the cost profile. So, you know, if you look at the way you or I would use uh AI today as we're chatting uh with AI, um when you start to apply that to things like images, uh, it's very interesting what the cost profile is to do that. It's very, very expensive. Um, so while we uh can do amazing things by analyzing video and learning all of those things about video, uh, it's hard to do that economically. Uh so we've had to employ some really uh interesting techniques to analyze images at specific times to learn things about what's happening in the video. But eventually we're driving towards a future uh where all video is analyzed all the time. So we can't get there today, uh, but long term that's where we're gonna get. So today, what we can do is we can analyze uh important events and snapshots in time to tell you things that are happening in your business.

SPEAKER_00:

Wow, that's amazing. Very different from the old school security systems you might see in I don't know, TV and movies. Yep. Not exactly that. So maybe uh start with an example, anecdote story, maybe a retail or a bank using AI surveillance to actually improve their operations.

SPEAKER_01:

Well, one of my favorite examples is safety and compliance. So if you look at applying a traditional surveillance camera to you know detecting when, for example, one of the biggest problems a retailer or a QSR have is they have an inspector come in and there will be boxes stacked in front of an exit door and they'll get a fine. And this happens every day across the country.

SPEAKER_00:

Wow.

SPEAKER_01:

Um, the traditional video surveillance solution, you would really need like um a very uh specific configuration on the camera. So you have to go in, you know, configure the camera to look for that kind of thing. You need to put a new camera in, or you need to add some kind of processing unit in order to detect when that particular condition has occurred. With AI, you can just take a snapshot from an existing surveillance solution from all of your exit doors, and AI will tell you when one of them is blocked. Uh, so that not only um is much more economical in terms of leveraging AI to solve that problem, it's also very accessible to our customer base.

SPEAKER_00:

Very cool. And you talk a lot about moving from reactive to proactive. Um, what does that mean exactly in practice?

SPEAKER_01:

Well, um, and and this doesn't just apply to AI, it also applies to another area uh that we work in, which is business intelligence. Uh, but it's then taking those types of conditions and notifying somebody that they need to be be taken care of, right? So in if you use the same example, and there's lots of examples, but let's stick on this one. Um, you know, historically it would be okay, we're we as a uh as an organization, we're getting a lot of fines for safety and compliance. Let's go back and and let's let's look at uh video from many of our locations or a sample of our locations to determine where this is happening. Let's you know, train managers to be looking for this, uh, but still you don't really know what's happening on the ground. Uh, in this particular case, we could actually send a notification to say there's a regional manager that's responsible for 20 retail locations to say you have this condition occurring right now, right? You need to go fix it. Um, and then on top of that, we can take those an accumulation of those conditions and report them through a business intelligence product to the overall organization to let them know how often this issue is happening across their entire organization in real time.

SPEAKER_00:

That's fantastic. And um, how do you think about scale? I mean, if you have hundreds, thousands of locations, that's a lot of processing, that's a lot of networking, that's a lot of cloud. Um, how do you bring that all together?

SPEAKER_01:

Well, we the way we've done it is looking at images. Um, so uh I think in the in the long term, when somebody's thinking about how you would apply AI to video surveillance, it would be constantly watching a video stream, right? For the types of conditions like the one we've been talking about, you don't need to be constantly looking at the video. So we'll take a snapshot every 15 or 30 minutes, we'll analyze that snapshot, um, and then we'll notify somebody, or we'll even make that searchable. So, to for example, we have a 70 location QSR customer, and we've essentially given them a Google search into their operation, right? So we take an image from every camera every 30 minutes, and you can search for anything you want to. You could search for water on the floor, you could search for cash laying around. Like if there's cash laying around in a quick service restaurant, usually that means you're gonna have some losses, right? You go in, you say, find me some cash. It it gives the uh the user the ability to search for anything they want to. And what what I call those is sort of semi-permanent operational issues, they're not sort of real-time people moving around. I think we we've spent as an industry way too much time thinking about people and vehicle objects when it comes to video, but the majority of our customers' problems have to do with operational issues.

SPEAKER_00:

Really interesting. And where is the intelligence in your solution? I mean, you have endpoints, you know, cameras, uh, you have the cloud. How do you kind of bring that all together?

SPEAKER_01:

This is where it can get really confusing. So I'll I'll I'll try to give the the shortest possible answer I can. We can use a combination of analytics and the camera. So, like the analytics, for example, can detect a person object, right? And we can use some kind of event mechanism to say we've detected a person. That's not necessarily AI, that's traditional analytics running in a camera. Uh, but then what we do is we send that image up to the cloud for further analysis, right? So the majority of valuable analysis that's happening, the detailed analysis, is happening in the cloud today.

SPEAKER_00:

Interesting. Yeah. Um and you you know you know, there must be huge opportunities for you, but also challenges. I think about most retailers they're living with the best tech from the early 2000s uh in their network uh and their IT. Um, how do you overcome some of those challenges in deploying a modern AI-driven video surveillance system? And what other challenges exist to the customers you were working with?

SPEAKER_01:

You know, that's been an issue for so many years because um especially with traditional analytics and AI, like the deployment is so costly, right? And you know, I don't want to keep repeating the same same point, but that's why we focused on on images as a starting point. Um, because what it allows us to do is take that existing infrastructure and send those images up into the cloud and do processing on higher-end technology. And it's it's it's quite impressive, actually. I think we surprised ourselves when we figured out that we could walk into a customer with the cloud technologies that are available today and the way that you can deploy them and say, no, you don't need to replace all of your cameras. And in many cases, you don't even need to replace the recording equipment. All we need to do is enable the technology, right? Um, what that does a little bit is it it kind of limits the the capabilities of um the analytics that you can run on site, you're offloading that to the cloud. Uh, but those analytics that you're getting in the cloud are are doing a full description of what's happening in your operation. So it's uh that's how we've over that's how we've overcome the challenge. There was another world where, you know, and and you can find these solutions on the market. We have some uh elements of this as well, where you go to the customer and you say, okay, you know, we need to develop an ROI with you where you spend, you know, five times as much as you would for a video surveillance system in order to put this analytics technology on site. My experience with that has been that the customer gets really excited about it. They come up with all kinds of use cases, but then ultimately decide that they don't want to spend the money. So to me, especially in video surveillance where the cost of doing that analysis is so high, the the key for success is making it economically viable for our customers and helping them with the ROI.

SPEAKER_00:

Got it. Interesting balancing act. But you do have your own uh camera. Yep. Maybe describe that, what makes it stand out, and um what are some of the advanced features there that you're looking to bring to market?

SPEAKER_01:

Yeah, all of our cameras have great analytics. Um and that that's what I would would call one of the less costly, you know, rip and replace that a customer would need to do. So what you can do is say, okay, there's some advanced analytics that we can do on site. You don't need to replace all of your cameras, but you can replace key cameras, right? Um, we just launched a product called the the AI3. And the AI3 is essentially a 360-degree camera that can perform uh a ton of retail analytics. So it can do uh queue line monitoring, like how long is your queue? It could do a notification saying, hey, you've got 10 people in line, like a text message to a manager who can go deal with the situation where you know there are customers waiting. Um, it can analyze the flow of traffic, how many people are coming in and out of your locations. Um, but it's also a camera. That's the important thing. Um, because there have there are retail technologies, there are 3D sensors and things you've been able to use for quite some time to do this kind of analysis, but you would be asking the customer to use two devices, a camera in order to watch what's happening, and then another device that accurately sees people moving around. What we're starting to see with a combined solution, and this does come back to the economics, is more customers are choosing a product like the AI3 because it fits more of their needs and they're deploying it at scale rather than just looking at one or two locations to analyze their operation. Um and the other element of what makes a product like that successful is the fact that it can have that immediate impact, right? So, you from a corporate level, if you're looking at the the way people are moving around a retail operation, yeah, you can look at statistics and analyze which location has more people versus another. But when you can tell a customer that they're able to notify somebody in real time that the experience the customers are having right now are not good, they've been waiting for too long, that's when you really get people's attention. And when you get when you're able to provide a solution like that, they're uh the customer is more likely to consider on-site um investment.

SPEAKER_00:

Amazing. So you know, the applications, the utility, the value, and retail is is so obvious, but it sounds like there are many other environments that could be really useful uh for your technology, you know, healthcare, hospitals, uh transportation, airports. Tell us about what else you're looking at beyond the retail setting.

SPEAKER_01:

Well, so that what I just described applies to transit. Our transit customers are super excited about the the AI3 product, particularly. There's a mobile version of it coming very shortly because they want to track people coming in and out of uh of their uh vehicles and be able to look at that. Um look at that overall, compare it, and this is where data comes into it. Like if we know how many people are going into a bus and coming out of a bus, and we have a connection into the I'll call it point of sale, but the the uh payment system inside of the bus, we can tell you whether or not everybody who's on these routes are paying, right? So that's something uh that's really interesting. Uh the AI aspect also has um applications in things like financial. And one of my favorite examples of that is uh imagine that you know a bank is providing a short-term promotional interest rate on a mortgage. They've got signs up everywhere. Um and that ends, you know, what is it now? It's November 6th. Let's say that that ends on December 1st, right? When December 1st rolls around, how do I want to how can I check that all of my advertisements for that mortgage rate are aren't up at the banks anymore, right? Today you you call around, you just you you make you you make sure that everybody has the instructions, but you have no way to check. Um, with our AI smart search product, you could just upload an image of that particular advertisement and find out immediately where we still still see it on location. So that's like that to me to me, that's a really, really interesting application where you would want to make sure that something's not uh existent anymore across your install base, and you can do it with a single search.

SPEAKER_00:

Amazing. So many opportunities. Uh, where do you see this technology going next? Um, I usually head down to the big NRF event in New York City, the National Retail Federation event every January. You get an amazing perspective on new applications and analytics and retail. But what's on your radar for the next year or so? Uh where are things headed?

SPEAKER_01:

Well, for the next year or so, our focus is on how we can ensure that the events that are occurring at the customer site are driving the analysis to the AI. So we're trying to tie those two things together and tie them back to even the data that's coming. One thing we haven't talked too much about is our full integration to point of sale systems, right? So we can take information about uh the sales that are occurring, we can combine them with analytics events that are that are happening to do things like conversion rates, and then we can apply AI on top of that to learn about what's you know specifically is happening uh on site. So the big question that we're working through now is how can we combine the point of sale data to trigger the uh AI to do the analysis and describe something for our customers and then notify somebody that something needs to be taken care of.

SPEAKER_00:

Very cool. Well, fascinating uh journey, and it's great to see real value coming from AI uh today. So congratulations on all the success.

SPEAKER_01:

It's a lot of fun to develop in this area right now.

SPEAKER_00:

No, it's good stuff and really cutting edge as well. Thanks, Jeff. Thank you very much. And thanks everyone for listening, watching, and checking out our companion TV show, techimpact.tv on Fox Business and Bloomberg. Thanks, everyone. Thanks, Jeff. Thank you. Bye.