Grocery retail has an inventory accuracy problem—and Focal Systems is fixing it with computer vision and real-time data streaming.
The average grocery store has 65 to 80% inventory accuracy. One in 10 products is out of stock at any moment. For an industry operating on razor-thin margins and competing against digital-native challengers, that data gap is existential.
In this episode, Kevin Johnson, CEO of Focal Systems, sits down with Joseph to explore how his team is using computer vision, data streaming, and stateful stream processing to close that gap at scale.
Kevin walks through how Focal moved from brittle, homegrown pipelines to a fully managed streaming platform—unlocking new use cases including virtual store walk, planogram compliance, and real-time theft detection—all without the operational overhead of managing infrastructure.
You'll Learn:
About the Guest:
Kevin Johnson, CEO of Focal Systems, has 30 years of experience successfully starting, growing, and managing businesses domestically and internationally. Particular emphasis on business transformation — changing companies from slow to high growth, money-losing to profitable, industry-lagging to industry-leading, and from founder leadership to professional management.
Guest Highlight:
"If you built this rickety data pipeline—congratulations. That's your job for the rest of your life is to keep that thing operating. Or would you like to work on this new thing and, you know, maybe something else and value added? And so I think that's where you get full buy-in. That's where you hit the tipping point. It's like, oh wait, I could outsource this and I can go focus on the new. If you can outsource the things that somebody else has done and done really well, and it scales and it's robust, it's an easier decision. If you've got interesting problems to work on, they'll take it, you know?"
Chapters:
[01:07] Guest Introduction + Focal Systems Overview
[10:19] Segment 1: Data Streaming Goodness
[16:08] Segment 2: Beyond the Stream
[25:56] Segment 3: Quick Bytes
[31:38] Segment 4: Joseph’s Top Takeaways
Dive Deeper into Data Streaming:
Get Connected:
Resources:
0:00:00.2 Kevin Johnson: Today we're processing between 5 million and 6 million images per day and that will be 12 million images by the end of the year. And that's a massive amount of data, it's just massive. We needed a system that was not brittle. We needed something robust and resilient. We needed something that would scale rapidly with us and we needed something that wouldn't be reliant on... We wouldn't have the hit by the bus problem.
0:00:27.4 Joseph Morais: Welcome to Life is But a Stream. Today we're going to be talking to the CEO of Focal Systems, a company that is making unbelievable evolutions and revolutions in the retail category. And when we started the show, the idea was to spread these relatable stories to you about how data streaming and stream processing is changing your life. And honestly, what's more relatable than grocery shopping? And you may be wondering how grocery shopping can be enhanced by stream processing. Well, you're just gonna have to tune in for that. I'm your host, Joseph Morais. Let's get started. So thanks for coming on the show, Kevin. Let's jump right into it. Tell me about yourself and what Focal Systems does.
0:01:13.2 Kevin Johnson: Sure thing. And hello, my name is Kevin Johnson. I am CEO of Focal Systems. We are an AI company, an artificial intelligence company using computer vision to revolutionize and improve the way traditional retailers operate. We're really trying to solve some of the oldest and stickiest problems of operating a large dynamic retailer.
0:01:36.7 Joseph Morais: Where you see a lot of dense products. That's what you meant by that as opposed to say like a clothing store.
0:01:42.1 Kevin Johnson: That's right. To give a little perspective on the problems we're trying to solve and why it's a real challenge, I'll take just a typical grocery store, for example. A medium-sized grocery store that most people are familiar with is 30,000 square feet, 20,000 to 40,000 square feet might have 30,000 to 40,000 unique products and everything from a packaged good, from a bottle of ketchup, obviously to fresh produce, to hot food, to bakery items, to cold food, to frozen food, so a huge variety of products. Again, 30,000 unique products, some very obviously different and some with minute differences in packaging or flavors or size. And if you are running that kind of store, you've gotta order the right amount of products at the right time, get them onto the shelf at the right price and with the right promotion. And that is hard to do. It's a complicated process of logistics and of labor on the shop floor to get those products in the right spot. And if you get it wrong, even one product wrong, it can ruin the shopping trip for a person. And at the very least, they might not buy an item, they might not find a substitute, so they skip buying something.
0:02:54.8 Kevin Johnson: High likelihood they might walk out of the store. It's the number one reason somebody is dissatisfied with a shopping trip is if they don't find the product that they want. And worst-case scenario, you actually lose a customer as a retailer. So there's pretty high stakes to get the right products again in the right spot, the right time, right price, all those things. To make matters worse, if you sort of think, well, we'll just order a bunch of everything and keep it in the back. Well, now you've parked cash in the back of the store that's not out for sale because you built this big inventory buffer. And since plenty of those items are perishable, you risk, not only have you parked money back there, but you risk that money sort of expiring, going to waste. If you get this wrong, on either side, you order too little, don't have enough product on shelf, you lose a customer, order too much, have too much in the back, you're losing money. And so all this depends on good data at the beginning and end of the day.
0:03:47.3 Joseph Morais: Retail grocery, that's a tough business. Like you said, you have a number of different items that look the same. Like how do you differentiate between this can of Campbell's soup versus that can of Campbell's soup when it essentially looks the same? Products are perishable, demand is volatile, competition is as tight as it could be. Is real-time data and insights necessary for kind of future-leaning retailers to be able to defend that edge?
0:04:10.6 Kevin Johnson: That's right. And not even future-leaning, just to survive today. So we talked about sort of the dynamism of the products in the store. Now add to that you're operating at razor-thin margins. And if you're a traditional retailer, you're competing with all the online newcomers and the hybrid newcomers. So you're competing with delivery services that might pick up the order somewhere else. You're competing with Amazon and their sort of online-only version. And again, that problem of getting the right product on the right shelf at the right time is really a data problem, figuring out when to order, how much to order. And the better data you have, the better you can do at that task. That may seem relatively easy given what people know about general modern technology. People know there are barcodes, there's barcode scanning at the checkout, computer ordering systems. Retailers have modernized over the past 40 years. Shockingly enough, the task of just figuring out what products are on the shelves, what is there in the store, is still a pretty manual task. It's the same as it has been for the entire history of retail.
0:05:18.1 Kevin Johnson: It relies on a person walking around the store, in the old days with a clipboard, and that's still true in some stores, clipboard and a pen or paper. The more modern version is to walk around with a hand scanner scanning barcodes. In a typical store, just that process, forget bringing it out onto the shelf, but just the process of walking around and figuring that out could be 8 to 12 hours a day. And that's just to walk around once and maybe cover part of the store. And as you would guess, that's a process that's highly prone to error. The people who are doing that task are getting interrupted. They're having to help somebody find the diet cherry cola that was supposed to be there and it got moved. So they're getting interrupted, they're getting distracted. And so believe it or not, with that process, the state of the art inventory accuracy in a grocery store is between 65 and 80% accuracy. So if you were world class under the old system, you'd have an 80% accurate view of what's on your floor, which is really not very good in this day and age.
0:06:26.0 Joseph Morais: No, 20% is a big margin.
0:06:27.9 Kevin Johnson: Big, big difference and you've got to spend money to order those products and again get them on the shelf. So that's why most stores typically have 8 to 12% out of stock at any point in time. And so think about that, 10% of customers that walk through that door might have a bad experience, might not find the product they want, might walk back out. 10% doesn't seem that much unless you think about the fact that they're operating on margins in those stores. So it's a huge amount. And given that you could lose that customer forever, it really is a high impact problem to solve for retailers. And it is solved by having better data, by closing the gap between that 65% and the ideal of 100% accuracy in terms of what's on shelves.
0:07:10.9 Joseph Morais: You described kind of the businesses that are customers of Focal Systems, but I'm curious, my note here is like what does a typical Focal Systems customer look like? And you talked about the businesses, but I'm curious, what is your first point of contact? Like who is reaching out to you from one of these grocery retailers and making that initial contact to onboard or express interest in Focal Systems?
0:07:34.2 Kevin Johnson: Yeah, it depends. It depends on the retailer. We're typically targeting relatively large retailers, so we're talking about 400, 500 plus stores within the chain and going up to the thousands of stores. And depending on the company, they might have a head of innovation who is tasked with going out and looking at the latest technologies to solve problems. But to successfully deploy a system like Focal Systems, it ends up taking everybody. It's got to get to the supply chain, and the store operations really has to buy in and be ready to operate differently because we radically improve the way they operate the store. We take away what was that sort of old-fashioned walk around the store and look once per day, and we replace that with an image every hour of every product on every shelf, if not more often. Radically different. You go from having an inaccurate 70% accurate one time a day picture of your store to having an always-on 95% accurate picture, every product, every shelf.
0:08:07.9 Joseph Morais: Well, I was gonna say my follow-up is gonna be what does a successful customer look like? And it sounds like you kind of... That's that number. That's that target, 95% accuracy.
0:08:41.6 Kevin Johnson: 95% accuracy. A successful customer looks like someone who's embraced the technology and is working with us to learn all they can do with it. Because again, it's cliche to say a paradigm shift, but it's a radical. The 95% is the starting point. So now, okay, you've got more accurate data on what's on the shelves. What can you do with that? Well, next, you can place more accurate orders, more just-in-time orders, so you don't have money trapped sitting in the back because you've ordered too much, too little, too early, too late. You're ordering the right amount at the right time. The next thing you can do is get it out of the back and get it onto the shelf more efficiently. So we alert the store to an item that's out of stock on the shelf. We only assign it as a task if we know there's inventory in the back, and we show them a map exactly where the item is and we show them where it goes on the shelf. And on top of that, we group those tasks together so we don't send a store team to go pick one item and put it in aisle one, pick another item and put it in aisle two. We group it together so they're taking the right items and bringing them to the same geographic area of the store.
0:09:47.5 Joseph Morais: You got my brain working now 'cause this is gonna set up our next segment. But I was just thinking about just the trends that you could give to your customers. Normally, you could say, "Yeah, well, coke is selling really well. We know that. We know that we sell 10% of it or whatever it is every day." But you could start to give those answers on an hourly basis. "We know that we sell the most Coke between these two hours." And suddenly that gives insight that they've never had before.
0:10:12.3 Kevin Johnson: That's right.
0:10:19.9 Joseph Morais: So we set the stage, let's dive deeper into the heart of your data streaming journey in our next segment. So again, you really talked about what Focal Systems does. But now let's get into a little bit more of how that occurs. So what has Focal Systems built or currently building with real-time data streaming?
0:10:35.7 Kevin Johnson: We talked about the fact that we're a computer vision system for retailers. What that means is we have cameras that we deploy in stores. So a typical grocery store or a home improvement store might have anywhere between 400 to 1,000 cameras throughout the store. So we can again literally see every product on every shelf, including the produce that might be in flat bins or things on end caps. These stores are highly dynamic, highly irregular, so we have cameras all over the place. They're taking images every hour, sometimes more often, and we are uploading those images to the cloud where we can process them and turn them into intelligence and then turn them into action back for the store. So we're interpreting them and then we're turning those into tasks and actions back into the stores. Now, some of those tasks, some of those actions might not be real-time. It might be an order that goes out once a day or even every other day. But some of those, replenish this shelf, is much more real-time. We also have a couple great different features and we can get into those but one of those that's sort of most viscerally satisfying to our customers is the ability to look at any section of the store throughout the day and we call this virtual store walk. You could be the CEO of one of the top five retailers in the U.K., for example, that happens to be one of our clients. They have a store in Gibraltar, Gibraltar near Spain so thousands and thousands of miles away. The CEO at any point during the day can pull up our virtual store walk and say, "I want to see what the dairy aisle looked like at 9:00 this morning in Gibraltar." So obviously we're far beyond the batch processing world. This is not a run-a-batch-job. This is a real-time data stream. So it is streaming data that's necessary to interpret and generate those tasks.
0:12:29.3 Joseph Morais: So what were the underlying data or technology challenges that led you and Focal Systems to adopt data streaming? Were you using something before and you realized this is just not gonna work?
0:12:40.7 Kevin Johnson: I think we realized early on we needed to stream data, if you will. There's just too much coming. There's a constant stream of data coming at us. So it made sense to think about this as a data streaming company from the start. We found as I think most people would if they were processing the kind of data we do, is that the homegrown system can be brittle. And the obvious thing is it doesn't always scale. You're building for one situation as you get your first client and your first store, and then that could turn into 10 stores and 50 and then 100, then 500. To put it in perspective, today we're processing between 5 million and 6 million images per day, and that will be 12 million images by the end of the year. And that's a massive amount of data, it's just massive. Making that pipeline scale on our own is tough. And I'd say another thing that some startups don't think about 'cause you're... Again, you're just... You're in the heat of it.
0:13:39.0 Kevin Johnson: You're trying to just solve problems and as quickly as you can. And if you've got someone who's smart and they can start wiring it together, that's great. But typically you have one person doing that, maybe two. Well, now you've only got one person who understands that system. What if you've got the hit by the bus problem? What happens if they quit or they get sick or they're not there or they go on vacation? So we realized that we needed to solve multiple problems. We needed a system that was not brittle. We needed something robust and resilient. We needed something that would scale rapidly with us, and we needed something that wouldn't be reliant on... We wouldn't have the hit by the bus problem. And for us then it became, all right, who can help us with this? Who can provide the best solution for us?
0:14:23.9 Joseph Morais: You mentioned computer vision and the cameras. So the cameras are taking pictures of everything and then some microservice with computer vision or something like that is interpreting what it sees and then it's outputting those as events into a Kafka topic. I'm curious what is then taking those events and doing that stream processing for you?
0:14:41.1 Kevin Johnson: Yeah, it's Confluent's Apache Flink that is a stateful data streaming for us via Flink, is managing that process for us.
0:14:51.1 Joseph Morais: Okay, excellent. So you're using fully managed Flink in Confluent Cloud. That will be very exciting to the product team here, so...
0:14:57.5 Kevin Johnson: Oh yeah.
0:14:58.7 Joseph Morais: It's very... We'll call that out.
0:15:00.9 Kevin Johnson: That's how you win Data Streaming Startup of the Year is by using Flink with Confluent.
0:15:05.8 Joseph Morais: I was gonna bring that up, but you found a way to put it in there that was a lot more organic than the way I was gonna do it.
0:15:10.9 Kevin Johnson: It flowed right in naturally.
0:15:14.1 Joseph Morais: So you're trying to win another award next year. It won't be a startup award. We'll have to figure out something else.
0:15:18.4 Kevin Johnson: It was startup of the year is what it was.
0:15:21.1 Joseph Morais: Have started successfully. So next we're gonna dive into how your partnership with Confluent solved your data challenge. But first, a quick word from our sponsor.
0:15:34.9 Speaker 3: Your data shouldn't be a problem to manage. It should be your superpower. The Confluent data streaming platform transforms organizations with trustworthy real-time data that seamlessly spans your entire environment and powers innovation across every use case. Create smarter, deploy faster, and maximize efficiency with the true data streaming platform from the pioneers in data streaming.
0:16:08.7 Joseph Morais: Now we'll go beyond the stream on why Confluent was the right fit. We know that data streaming was the answer. But having run open source Kafka myself, I know that there's a lot to wrangle with that. And I also know with that whole hit by the bus scenario, you could have just went and found people that know Apache Kafka. It's a very well-established open source distributed technology. There's plenty of people that know Kafka. What ultimately made Focal Systems choose to work with Confluent specifically?
0:16:34.9 Kevin Johnson: To tell the truth, one is that we were already working with Confluent for other pieces of what we do. And so when we discovered we could use you for Flink as well, it was sort of like, well, that's handy. And on top of that, we've found Confluent just great to work with.
0:16:49.4 Joseph Morais: With our partnership, again, we talked about what storewalk is and we know that it uses Flink, but I'm curious how the team at Focal Systems tackled it once you kind of realized that Confluent was gonna be able to help you with that as well. Did it start as a minimum viable product? Did this come back from feedback from your potential customers? Was the original iteration of Focal Systems technologies always gonna have this storewalk? I'm really kind of understand how it came from idea to actual execution.
0:17:16.7 Kevin Johnson: It's really... We try to really listen to our partners. And so the most interesting things we develop are when we, A, stick to our core value proposition and don't get too far off track. These are gonna sound like contradictory things, but we've been asked to do a lot of things that sort of don't involve computer vision. And so we try to bring it back to, "No, we're gonna use computer vision to solve a problem." But at the same time, think as the customer asks, "Can we use computer vision to solve that problem?" And so storewalk came out of that. I think It was originally a store manager saying, "Well, we've got the cameras. Can I see how... I manage two or three stores, can I see how my other store is doing when I'm not there?" "Yeah, of course. We're a computer vision company. We've got cameras, we've got pictures. We can do that." And it started in the more batch world. I think it started with, "Yeah, you can look yesterday." Then it became, "Well, that's no good. That's not fun. I want to look now. I'm in store three now, I want to look at store two." And so it very quickly became real-time. The use of Flink with Confluent wasn't strictly driven by storewalk. It was more driven by the realization, again, we can build our own data pipes, we can build our own stream, if you will, but it's quickly... It's not gonna scale, it's gonna be brittle, it's gonna be a single point of failure. We need help. Who's best to help us with that?
0:18:47.5 Joseph Morais: Right. You went to your customers to figure out like what to build, but then you used best-in-breed practices on how to build it and that ultimately led you to Confluent, and I'm thrilled that the partnership has happened. So when I was having this conversation with the Flink team, they mentioned that you guys are also using Flink for something called the Action Tool. Can you tell us a little bit more about that?
0:19:07.6 Kevin Johnson: Sure. So the Action Tool, that is where we turn that data into tasks, into actions within the store. So we talked about storewalk. That actually is surfaced to the store management via a tool we call Impact. And that's where store management can see, they can do storewalk, they can look at adjusting their ordering, they could look at how their store employees are performing, whether they're doing the tasks on time, whether they're doing them correctly. They can look at planogram compliance within Impact. The people in the store putting the products on the shelves interact with something called Action Tool. So that is the tool where we are sending the, "Restock the Cholula hot sauce on aisle nine, bay three, shelf five. Here's the map, here's the picture. By the way, if it's also on aisle seven on the endcap, so you can replenish it from there if you don't have time to go to the warehouse." So that's Action Tool. But again, back to the intensivity of the data, we're also identifying the products. So not only are we looking to see whether it's in or out of stock or low, we're also checking to see if it's... And we didn't touch on this part of the store complexity. Stores all operate according to a plan. It's not by accident that Coca-Cola flavors are lined up the way they are and that they get so many shelf spaces versus the Pepsi. That's all part of a plan, what's called a planogram in store parlance. And so we're also checking for the store to see if they're in compliance with their own planogram. So we identify the product and also tell the store and the workers in the store, "No, you need two more facings of Cherry Coke. You're not operating according to the plan," or, "It's on the wrong shelf," or, on the wrong section of the aisle. So we're doing that as well. All that comes back to the store employees, the store staff, via the Action Tool.
0:21:02.3 Joseph Morais: I'm curious, this is an interesting one. Was there anyone at Focal that resisted embracing a fully managed platform? So there's sometimes when you have a SaaS platform, I wouldn't call it a black box, but it pulls away some controls. Though, again, having been an operator myself for many years, I'm fine. Like if you want to run something for me, as long as you're not gonna break it, I'm happy to have you do that. But I'm curious, was there anyone on the team that said, "No, we need to own it all?"
0:21:24.5 Kevin Johnson: Well, probably the guy who built the original pipes. "That's my job. That's my baby. Why are we doing this?" But I'll bring it back to our conversation about the workers in the store. I think, okay, so if you built this rickety data pipeline, congratulations, that's your job for the rest of your life is to keep that thing operating. Or would you like to work on this new thing and maybe something else and value-added? That's where you get full buy-in and that's where you hit the tipping point. It's like, "Oh, wait, I could outsource this and I can go focus on the new." So I think that was, I don't think there was a ton of resistance. I think there was a little bit of not-invented-here syndrome that quickly went away because we have so many technical challenges to focus on and we're trying to do things that nobody else has done. So if you can outsource the things that somebody else has done and done really well and it scales and it's robust, it's an easier decision.
0:22:27.8 Joseph Morais: Now, what is the vision for data streaming at Focal Systems? And that pun was absolutely intended.
0:22:33.9 Kevin Johnson: Yeah, absolutely. We are a vision-driven system company. I mean, for us, I think it's we're obviously expanding, we're adding stores and categories. So as I mentioned, we're about to pilot within supermarkets. We're already in the health and beauty aisle, obviously supermarkets have that themselves. We're about to do a proof of concept with a purely health and beauty pharmacy-focused company, which will be exciting. We're expanding into home improvement, which is exciting. We're expanding geographically. So for a startup company, we're shockingly international. But I still come back to, for me, the most exciting thing is to listen to the customers and work with them to solve their problems. And like I said, we stay focused on computer vision but then we get asked the question and we have those aha moments like, "Wait a second, we can solve that." You mentioned shrink earlier. And so there are two forms of shrink. One is... The biggest one is actually product spoilage, which people think shrink, they think theft.
0:23:34.5 Joseph Morais: Right. It's not just that.
0:23:36.0 Kevin Johnson: Slightly edging it out is product spoilage, which we do solve and have solved for a couple of years. But the theft piece we hadn't really been focused on until last year. And a client said to us, "We're having increasing theft in our stores. Can you..." And it was just one of those like, "Hey, could you solve that for us? Could you help us with that?" And it was one of those like, "Wait, why haven't we been solving this? Of course we can. We have cameras on every product in every aisle and we're integrated with your systems." That might not sound like a big deal because stores you know they do have CCTV footage and they have a loss prevention team. But that team has to look at 17 hours of footage throughout the store every day after the fact. And again, back to real-time, they look at yesterday or the day before and they're trying to figure out what may have been stolen, may not. Now we send them a file. At the end of the day, they can, or at any point of the day, they can look at a report. They can see what we have identified as theft, likely theft. We give them the aisle, the shelf, the bay, the amount stolen, and the 30-minute on average window in which it was stolen.
0:24:43.4 Kevin Johnson: So they say, "All right, I'm looking for somebody who took six lipsticks from this store, this aisle, this shelf, between 2:00 and 3:00 this afternoon. They go to their CCTV footage for that store, for that aisle, for that spot, for that time. They know what they're looking for. They spot that person and we literally are making arrests the next day. So they're then putting it in their facial recognition technology, especially organized retail theft they do it over and over again. So they just put it into the facial recognition in their stores. The next day the person comes in and they arrest them. We're literally catching people the very next day, which is phenomenal.
0:25:19.7 Joseph Morais: Yeah. Once again, your system is just pointing these customers in the right direction. Again, instead of combing through 17 hours of maybe 30 cameras, you can go right to the right timeframe and see, was this in fact a loss or not? And again... And it comes back to some conversation we had earlier. Once you have that real-time data for your customers, now they're coming to you with asks you never thought of. And you're like, "You know what? We have the data, we can do that." I love that. So that's the byline for this episode. We have the data, we can do it. Before we let you go, we're gonna do a lightning round. Byte-sized questions, byte-sized answers, that's B-Y-T-E, like hot takes, but schema-backed and serialized. I want you to give me the first thing that comes to your mind. Are you ready?
0:26:09.1 Kevin Johnson: Sip of coffee.
0:26:10.3 Joseph Morais: All right. So Focal Systems was Confluent's very first Data Streaming Startup of the Year. How has winning that award helped the company?
0:26:44.1 Kevin Johnson: Well, it feels good for one thing. And it's brought us attention. And I don't want to underplay the feel good. You have engineers who work hard every day, long hours. In some ways, the client teams get all the credit. We're often visiting stores and talking to clients, inventing new stuff, and the engineers are toiling behind the scenes. So it felt great for them to be recognized 'cause it really is their award. They're the ones who figured it out. They're the ones working with you guys, they're the ones who have made this happen. And so it was a great bit of recognition for the technical team. So that felt very good for them and for us to recognize them. And then of course, it helped raise our profile as well with investors and customers. So it's a great profile boost.
0:27:04.3 Joseph Morais: You hear that? Any startups out there, data streaming, you better be talking to Confluent.
0:27:07.7 Kevin Johnson: Data streaming, yeah.
0:27:09.2 Joseph Morais: What's something you dislike about IT?
0:27:11.1 Kevin Johnson: Kind of touched on this earlier. So I think we're a technology company, but technology doesn't solve all your problems. We have the tech, we have the data, we have the tasks, but you still have to think. You still have to work with us. We're showing you a whole different set of data that you've never seen before, a new level of insight into your stores. Let's work together to figure out how you're gonna change the way you operate. And if it's just... I mean, granted, half my company is former retailers, myself included. But if you just leave it to us to say, "Here you go," you're gonna get half the value. If you work with us, if you think, if you don't treat us as a commodity, you'll get closer to 100% of the value.
0:27:52.5 Joseph Morais: I like it. What is your hot take on the future of AI?
0:27:55.3 Kevin Johnson: My hot take among the millions of hot takes out there. I think one of the interesting... First of all, AI technology and capability is moving faster than most people think, other than people really following the space, of course. However, the deployment of it is moving slower for traditional companies than most people think.
0:28:13.8 Joseph Morais: Absolutely.
0:28:14.7 Kevin Johnson: When did you last go into a bank and go talk to a teller to manage and deposit checks? It's probably been a long time. Electronic banking is here. You couldn't imagine going back. Electronic banking took 20 or 30 years to roll out 'cause I was in that space where we were talking about, "It's here," and then the next year it still wasn't here, and the next year it still wasn't. It took forever, and now we can't imagine life without it. Various aspects of AI are gonna be the same, including what we are doing in Focal. We're talking to stores, some of them are thinking, "Well, this is amazing. We'll do it next year or the year after." It will be slow adoption, and then suddenly everyone will be using it and we won't be able to imagine life before it.
0:28:59.9 Joseph Morais: The tech always outweighs the adoption. That's a good one. What is your non-tech activity or hobby that's impacted the way you think about data?
0:29:11.1 Kevin Johnson: My biggest passion outside of working for an AI company, tech company, is coaching. I coach distance running, track and field. You'd say outside of data, but of course I use the data. And so it's figuring out... I think when you're coaching a kid, trying to help them realize that data is important. So teaching your athlete to gather the data and make use of it. You can run as hard as you want, but unless you're using data to influence your training, you're gonna miss out on being as good as you can be. So I try to help my athletes think about, "Okay, how did you feel today going into this? How did you perform? Even in practice, "How did we perform?" All right, next week or next day when we come back, we're gonna take that data and use that in the workout, use that in the goals. And so data is everything.
0:30:02.4 Joseph Morais: Where are you getting outside inspiration, Kevin? Is it maybe from a book or a thought leader that you can share?
0:30:08.4 Kevin Johnson: Well, I'll come back to coaching and the kids and seeing... Yeah, where I really get inspiration is seeing a kid at any level dedicate themselves to something and get better. There's nothing more satisfying than seeing somebody have that unlock, like, "If I put the work in in the right way, something can happen that I didn't even expect." And I think I get inspiration from that because if I'm doing it right, I try to bring that back into work. And it's easy to see in a kid, and in some ways it's the most satisfying. We're not done, we're not dead yet. We can bring that same approach back into our adult lives and still have that same kind of improvement. And I think it's easy as adults to sometimes get sort of stuck or complacent. We should take that same approach ourselves.
0:30:57.8 Joseph Morais: I like that. Being inspired by the betterment of the youth is definitely a unique answer on the show, but a very strong one and it's a very strong one. Any final thoughts or anything to plug, Kevin?
0:31:08.1 Kevin Johnson: Well, if you're a retailer out there, I'll plug using Focal. We will absolutely revolutionize the way you do business. And if you are a startup like us working with AI, working with data, streaming is the way and you should be talking to Confluent.
0:31:23.1 Joseph Morais: I love it. Thank you so much for joining me today, Kevin. And for the audience, stick around because after this, I'm giving you my top three takeaways in two minutes.
0:31:39.2 Joseph Morais: All right, let's get to those takeaways. The first one is some statistics around the way grocery stores work today. 10% of items could be out of stock at any time and that their current systems only have 65 to 80% accuracy, that's just mind-blowing to me. But that's ultimately which led Focal Systems to build that storewalk functionality so that anyone anywhere in part of their leadership can go and walk that store. And the example that Kevin gave us about a CEO in the UK walking a store in Gibraltar is just kind of amazing 'cause this gives them that true insight, and it's not just a dashboard. The type of insight that you can only really experience in a virtual type of scenario like that. So my kudos to them for building something so game-changing. My next takeaway is this idea that Focal Systems turns data into tasks. Again, I was a utility clerk, and I still to this day do not know what aisle the peanut butter is on.
0:32:28.8 Joseph Morais: Eliminating those mindless tasks of just constantly walking around and scanning everything without any thought into it and turning that into actionable things that these clerks can act on in the beginning of the day and then gamifying it by showing them how much money was saved. Really impressive way to make Focal Systems useful for not just those leaders, but the people that are there on the floor making those experiences for those customers. And then the last takeaway, and Kevin said this more than once, so I have to talk about it. Real-time is what makes all of these functions that Focal Systems has built truly magic. And then once they had the real-time data, they realized that their outcomes were limitless. And the example that we had of that is when the one customer asked Focal Systems could they help them with loss prevention, and they realized, "Well, we have the data. Sure." And they built a brand new piece of functionality using those real-time signals, stream processing, and data streaming on Confluent Cloud. Once you get your data to real-time, what possibilities can you provide to your customers?
0:33:27.7 Joseph Morais: That's it for this episode of Life is But a Stream. Thanks again to Kevin for joining us, and thanks to you for tuning in. As always, we're brought to you by Confluent. The Confluent data stream platform is the data advantage every organization needs to innovate today and win tomorrow. Your unified platform to stream, connect, process and govern your data starts at confluent.io. If you'd like to connect, find me on LinkedIn. Tell a friend or coworker about us and subscribe to the show so you never miss an episode. We'll see you next time.