Mister Beacon Episode #161

Transforming the Supply Chain with Ambient IoT

February 14, 2023

This week on Mr. Beacon we have Thaddeus Segura, who is the VP of the Data Business Unit at Wiliot. Thaddeus has a background in retail, having worked at Walmart for ten years and talks about the application of Ambient IoT in retail and the problems that it solves.

The data analytics team at Wiliot is involved in every major project, turning raw data into insights and actions. Thaddeus shares his perspective on the problems in retail that are ripe for solving with Ambient IoT and how his team is approaching this. He explains that the granularity of better data has shifted in retail and how technology is used to manage inventory and sales. We hope you enjoy this conversation.


  • Steve Statler 00:00

    Welcome to the Mr. Beacon podcast. This week, we're going to be talking to Thaddeus Segura, who is our VP of the data Business Unit here at Wiliot. We try and balance normally odd stuff with without stuff. But with Thaddeus joining us He has an amazing background in retail 10 years at Walmart, super smart guy. And I think you're gonna enjoy this conversation. It's a real eye opener in terms of the application of Ambient IoT to retail, we talk about some of the problems that exist that can be solved the approach to solving it. And I certainly learn something every time I speak to Thaddeus I, we've been off the air for a while I've been taking a bit of a break, we try and make sure that we're not just putting out episodes because we have to, we've got good guests. And that is certainly one of those. And I thought it'd be fun to do this interview in front of something we call the wall of websites. We which is something that our tech marketing team has been working on. So I wanted to show it off. And basically what you're seeing is 1000 Ambient IoT sensors behind me, each one, when it broadcasts, you will see a little wavelet and that those colors are indicative of temperature. So over my left shoulder, then you'll see some patches of of orange because there's a space heater that's blasting hot air at that. And so you can't see radio waves. But if you could, this is kind of something that that you might see. So hopefully it's not too distracting. And I hope you enjoy my conversation with Thaddeus. And Mr. Beacon, Ambient IoT podcast is sponsored by Wiliot, bringing intelligence to every single thing. Thaddeus, welcome to the Mr. Beacon podcast, it's great to have you on the show.

    Thaddeus Segura 02:17

    Thank you. It's a pleasure to be here. I'm excited.

    Steve Statler 02:20

    Well, you have a really interesting job we don't often interview people that work at without, we've certainly we've done it. And we'll probably do it a little more often as the stuff we're doing gets more and more interesting. But I'm really keen to talk with you. Because you have seen us from the customer side, you used to work at Walmart. And now you're doing this job, which is probably a job that many people looking at whether you would think didn't exist, you're the data guy, your VP of our group that's doing analytics that is involved in every major project that we do and is taking raw data and turning it into insights and actions. And a lot of people probably look at Willie out and think that we're this Ambien IoT silicon company, and we are. But we decided pretty early on to focus on trying to add value to essentially commoditize that lower layer and make our living in the area that you operate in. So there's lots that I want to talk to you about, I want to talk to you a bit about the you know, what you're seeing in terms of the use cases, the how what you're doing now differs from the kind of data analysis that was possible with RFID. I want to pick your brains about how your team is approaching this. And you know, what kind of things they're doing what they're not doing. But before we get into that, you came from Walmart, you saw a bunch of problems that are very, you know, they're immensely capable, world class leader. What's your perspective? On the problems that are right for solving with ambient IoT? What are the business problems?

    Thaddeus Segura 04:11

    Yeah, I think I can talk generally about some of the problems that are facing retail, and you expand a little bit beyond that. But one of the things I saw over the my decade with Walmart is this shift in the granularity of data, you need to be able to operate and be competitive. So it used to be good enough to know whether or not you had, you know, something in the building. And I always use Serato as an example, because I eat a lot of Serato. And it can be hard to find these days. But in the past, it was good enough just to know if you had syrups are there and you have a shelf and you could fit a whole bunch of Serato on it and you just had to keep it there and keep it full so that when someone showed up, they could buy some you could have a margin of error of you know, 12 or 24 bottles and it really didn't matter. Today that's not the case. Today, the exact number you believe you have is plugged into a system and that's your available They'll sell inventory. And people can see it on the website before they drive to the store and Instacart is coming by and predicting whether or not they can fill the order. And it's such a data driven ecosystem. Now with omni channel retail, that it's no longer a world in which you can have a margin of error of plus or minus 12, you really have to have an exact match. And retailers are really bad at that. And it has been such a rapid shift from the world in which we operate into the world trend now that everyone's looking for technologies to try to solve that problem. And I think that that's not just true in retail. But that's true everywhere, like the amount of granularity you needed before. And the margin of errors you could operate out before were just completely different than what they're at now. Because everything is a computer and automated. It's true data. It's not these these fuzzy things and humans incorrect. So you don't have someone in the store, like making a substitution in the moment. Serato is out of stock slug of chili, you're picking something and then it's being driven by data into another system. And so things really have to be perfect these days. And people are looking for technologies to solve those problems.

    Steve Statler 06:01

    So it's like the COVID crisis, just accelerated the buy online, pick up in store, buy online, have it to delivered. And that's not what the original systems are built for right

    Thaddeus Segura 06:13

    now. And they're complex legacy systems. These are not simple machines. These have been built over 50 years. And were never intended for this kind of scale or this kind of use. And so that's why some retailers like Amazon have like an inherent advantage as they came up. And they built these things are scratch from scratch for scale, or other people will have like a fundamentally different approach. And something they really have to like get through and solve on top of old legacy systems that are in many cases to integrate and to completely rebuild from scratch.

    Steve Statler 06:45

    So why can't everyone just look at what Amazon is doing and do the same thing.

    Thaddeus Segura 06:52

    It's tough. They're also different pieces of competitive advantage. They're like, you know, one thing that I always people ask, why was it still at Walmart for 10 years. And the truth is, when you get into the middle of the country, they have a great foothold. You know, something like Amazon makes a lot of sense when you're on the coast, when you drive up and down the street. And there's a house every 50 feet, or 20 feet or five feet in a cul de sac. But when you get out into rural Nebraska, or South Dakota, and the houses are 100 acres apart, suddenly, it doesn't make sense to have two hour delivery. And so you know, this hub and spoke model, this omni channel model, you're even having physical stores, it still makes a lot of sense and everywhere other than the coasts. And so I don't think it's just a one or the other model. It's not just pure online, which is why you see Amazon opening brick and mortar stores. And it's not just brick and mortar stores, which is why I truly believe in omni channel is the solution. If you want to scale to solve the problem for the whole world, and not just the metropolitan areas.

    Steve Statler 07:50

    Yeah, it seems like you can't just build a parallel infrastructure of distribution centers that are run by robots, because then you end up having all of that overhead. And you have the overhead of your brick and mortar stores and the people in them. And it's just, that's a road to ruin in terms of the economics and you got to capitalize on this. I mean, Walmart especially but so many retailers have that proximity don't they to, to the people that they're serving. So, you know, we've focusing more and more on Ambien, IoT on this podcast, everything being connected. And any one of the things I love about it is there's just so many things that you can do, if you take, you know, we've been, if we look at the connection between things, and the cloud, we've been drinking through one of those tiny cocktail straws where you really have to, there's a really small amount of throughput and only so much can get through. But now when everything can be connected, you can do anything. But the question is, where do you start? What are the use cases that make sense, given where the technology is today? And that's going to change but what's your view? You've been at Willie up for a while now? And so you're seeing both sides of it? What are you What do you think the use cases are that really makes sense to apply Ambien IoT technology to today?

    Thaddeus Segura 09:24

    Yeah, so I think I just wanna make a comment on what you said, because when I was interviewing for this role that stuck out is the single biggest risks. And the thing that scares me the most is the solution space is so unconstrained, like as a product manager, actually like thinking outside of the box, because my first job was defining the box. Yeah, let's add those constraints. Like it became easy. Here. The technology is so flexible, right? Like, we have temperature and location, but we can do humidity and light and gas sensing and all these things that we're experimenting with and it's like, where do you go, what do you try to solve? You don't fall into the innovators dilemma where you like pick And hold yourself in one place, but you can still keep going. And so I have to think about in terms of different horizons and like different periods of time. But right now, this might be my bias speaking, I think there are a lot of pieces of low hanging fruit in retail, that we can go after and, and supply chains. And so some things, we've been able to do it, we've gotten a ton of traction on three big buckets I'm looking at the first is inventory accuracy. The second is food safety. And the third is what I call it in the index array. And I can go into any of these in depth of a high level, like inventory accuracy is really like tracking and understanding where things are at in real time, we have an advantage, because we have this very low cost, low touch infrastructure, we can put it anywhere we can deploy it globally, we can run it on batteries. And so there's so much flexibility, that we can see things that other technologies can't with much lower touch with food safety, because we can inherently measure temperature throughout the whole lifecycle, we can check track provenance back to a supplier or a grower, we can isolate down to you know, single cases, whether or not something was at risk for a food outbreak. So you don't throw away all of your food if there was potentially, you know, a salmonella, and you have 10,000 pounds of spinach, but only one case was potentially at risk. And then finally, you know, this index array is an idea that came out because we have matured and how we're approaching data. But in the beginning, we kind of just shoved stuff out there and then dug into the data and saw what we saw, it's something that was really interesting is it's kind of like going to a doctor, when you have paid, you don't know what's wrong, or what's broken inside, you have no way to see. And now we can scan the whole thing and understand like what's going on inside and be able to diagnose this. And so yeah, there's a lot of different things. But those are the three buckets that I'm working on right now.

    Steve Statler 11:45

    That's awesome. I love any list that has three things in it. And those are just very powerful use cases. But I want to kind of challenge the the real time inventory piece because, you know, I love to think in metaphors. And I feel like, what we've got is this high definition color capability, we talk about our tags as pixels. And so maybe the metaphor is, is apt. But I remember, you know, back in the early days of personal computers, people were like, I don't need a graphics screen, I can be fine with black and white. Why do I need color? Why do I need this just to play games? And so, you know, if the CFO of a retailer is looking at that and saying, Oh, you want real time? Well, you know, I'm sure that's going to be very impressive. But do we really need real time? Why do I need real time inventory?

    Thaddeus Segura 12:45

    You need it? You need it? Very badly. Okay. So I'm sure you've ordered things online in the past, and you place for Instacart order, and then you brace for it, like how many texts am I going to get? How many substitutions how many things are going to be unfilled today. And it's a problem everyone's having the fact that you are able to place that order is indicative that some system somewhere had a critical threshold of confidence that that item was in stock, and in stock above and beyond some level of safety stock that you are going to get that item Otherwise, they wouldn't have showed it to you. So it passed all these logical tests, and you were able to order it, you gave them your money. And then still 10% of the time, they can't find it and they give you your money back on the order of 10s of billions of dollars a year across all retail, it is this massive problem. And the thing is that so often that items in the store, their data was right, but it wasn't in the one location where you thought it could have been. It's in a 200,000 square foot box or a 50,000 square foot box. And you're trying to send a person to another location, but they can't actually track it down because it's still on a pallet in the back room. Because three people called in last night and all the freight didn't get stocked. Or it's on a display. Because it's primetime. We're going into Valentine's Day, and you ordered how Valentine's candy for your partner and they went to the aisle but it's actually on 20 displays all across the front of the checkouts. And so that seems so intuitive when you're a shopper like you would have found that. But these people are being measured on efficiency. And they're trying to find things so quickly that it's not good enough just to know that it's in the box, you need to know where it's at within that box, and you need to know where it's at within that box in real time. And that is a 10 digit problem in terms of revenue for just American retailers right now. And so I don't think it's science fiction. I don't think we're trying to sell people a solution they don't need. I think there's a very tangible, measurable, quantified need that our technology solves today.

    Steve Statler 14:45

    I think you've done an excellent job of answering that question, but I think it's still hard for people to get their head around. Why is this so challenging? I mean, it's you know how Why can't you find something? You know, you you have all this technology already using RFID? Maybe you certainly scanning QR codes. And so why can't you find the product in there? And what are the implications of not being able to find it? I mean, certainly disappointing. Someone with an online order is one of them. But I think it's worth just dwelling a little bit more on what the reality is for the people that are running. I mean, it seems like such a simple thing, and the product comes in the back, you put it on the shelves, and you sell it. But what so what's the big deal? Why can't very, you know, world class companies, excellent, companies do better at doing that.

    Thaddeus Segura 15:46

    I think if you had, if everything was smooth, and everything ran within, like the expected bands, it wouldn't work that way. But I think the most tangible way to characterize the problem is thinking about like labor. And the truth is like, it's become really hard to fill these roles, not just the soccer roles, but also the truck driver roles. The the Longshoremen that unload the shipping containers off the boat. So if you're at 60%, staffed when you're unloading the boat at the dock in Oakland, and then the train doesn't stack, so it doesn't actually get on the train at the right time, then you have to call people in so everything that was blocked and you had this bottleneck, suddenly it's cleared. And all this freight flows through at once. And so the store doesn't just get the 8000 cases they were scheduled to get that night with consistent demand, they get 16,000 cases. And they also have their own colons. And so you run into these problems where you don't have these this consistency of flow throughout the supply chain. And everything's become so lean, because we were all about just in time for so many years. And then suddenly, something happened via COVID that completely broke the entire system. And I don't think we've recovered. And so you have these things where these things come through, like you'll get all this freight in a single night. And it may take us for a week to get back on process, because they were so used to operating with such a small margin of error, that a slight deviation causes this cascade of Miss chain effect that prevents the stuff from getting into the right place. And in some cases, there are automatic systems that will detect, let's say, a dip in sales and order more merchandise. So if you're a store with 300 pallets of unworked merchandise in the back, and the system says we're losing sales, let's say more, well, you're Compounding the problem. And it just becomes a cycle that you can't escape. And it's really more common than people when people think like, there's a lot of time the freight you're ordering is in the store. And it's just waiting to be stocked.

    Steve Statler 17:41

    No, I think any of us who've shopped, which is almost everyone have experienced the frustration of not being able to get what they want. And we're kind of we expect we expect better. So it makes sense. Let's just spend a bit more time on that second category, the food safety thing, because I think that's super interesting. And it's important to everyone. as well. You know, what, what is the current state of play? And what's the opportunity to? Because we've got regulations, right? There's we're supposed to be able to do traceability we've had other guests have talked about having to respond to the regulation, but what sort of where are we now? What's kind of best practice? And what do you think the opportunity is there to improve food safety with Ambien IoT?

    Thaddeus Segura 18:33

    Yeah, so I turn to food safety. And this whole category, I have three sub points, and they are precious keys tracking compliance, and what are called provenance. And so within a freshness, you know, this one isn't food safety from a compliance perspective. But freshness matters, like a good strawberry and a bad strawberries experience. It's not just like slightly worse, it's just fundamentally different. Like I can like, imagine a great grape, and like that's like worthwhile on a bad grave. It's just miserable if it's not related between, right. And so freshness really matters. And especially now that not everyone's selecting their own fruit, you so often, if I'm ordering through Instacart, or some fulfillment service, I'm not selecting. So it's no longer okay, that they just have this distribution, and I'm picking from the best. It has to be that the whole distribution is more uniform. And it's all the best because someone else is picking their fruit for me. And so I do think that there's an impetus and some stress on retailers to improve the quality of all their free because consumer selection behavior is changing and will continue to accelerate as people move to crowdsource with third party fulfillment systems. So there's so much we can do around freshness, even with just the sensors we have today. So just with dwell time and temperature tracking, you can do so much with existing TTI models where you can plug these in as inputs and understand like what quality your grapes may have, and not at the truckload not at the bush a little bit at the individual Will case or even if it mattered to you at the individual back, like we could track the freshness of a single banana if we wanted to, I don't know if it'd be worthwhile to put a tag on a banana, but we could if we wanted. So that's the first one. The second, you said TTI.

    Steve Statler 20:13

    What was TTI? Time temperature integration? Okay,

    Thaddeus Segura 20:18

    so there are existing models that take into account things like dwell time, since you harvested the fruit, the temperature that it was at, because if you're a banana, and you spend 10 minutes, 60 degrees, that may be okay, if you spend one minute at 120, that's probably pretty bad. And so these models take into account over time with how much exposure to different bands of temperature have an impact on accelerating the ripeness or even contaminating the fruit, if you think about like meat, temperature outside of a certain band can lead to major bacteria growth. So there are existing models, there are more complex models that take into account things like humidity, which is why some of the sensors we'll put up the end of this year, and in the next year will be so critical for these models, because we'll add another key dimension for these existing models that retailers will be able to use.

    Steve Statler 21:08

    Very cool. So I interrupted you, you were I think we're just wrapping up point one.

    Thaddeus Segura 21:14

    So the second is around case tracking. And as I understand that there's compliance coming out in both the states and in the EU around the ability to track cases. So we have FISMA compliance in the States, we have digital passport in the EU. And we need to be able to show where these things were at at a given time and show consistent chain of custody all the way through. And with a simple UPC or a simple SKU, their shortcomings, you have a whole pallet of bananas, it's all 4011. It's one UPC, and it all looks the same. But the compliance will dictate that you can differentiate between those 16 or 18 cases that are on that single pallet and understand if they were picked at a different date or came from a different place. And people will need technology to disambiguate those, and they'll be looking for for new technologies, because they'll have to upgrade. And I think we make a ton of sense. And just by meeting that food safety compliance requirements, you unlock all the other benefits of the technology.

    Steve Statler 22:11

    Very good. And you said FISMA what does that what's that?

    Thaddeus Segura 22:15

    I wish I knew the acronym. But

    Steve Statler 22:17

    I think I'll tell you what I think I think it's Food Safety Modernization Act. But anyway, it's the law, right?

    Thaddeus Segura 22:23

    Yes. Federal law in the United States? Yes.

    Steve Statler 22:27

    Very good, very good. And the third point,

    Thaddeus Segura 22:31

    third mills are in Providence, which is that a lot of people make claims around grass fed or organic. And I think that in five years would be table stakes, that you'll have to be able to prove those claims. You know, so many things are being audited, and finding out that this Coconut oil contains 0% Coconut, or all these claims that these manufacturers have made aren't actually true. Yeah, I think as soon as the first retailer exposes visibility directly to consumers, and is willing to show their chain of custody and prove it's going to be a race to implement that feature everywhere. And it's going to become table stakes, as soon as one person makes that leap.

    Steve Statler 23:10

    I think you're right. And actually we got a shout out in the Wall Street Journal. There was one of the excellent sustainability reporters based in London did a thing about in the EU, you have to there's a law that where you have to prove that your product, the provenance did not impact like clear cutting, you don't want to have beef that was raised on lamb that was basically a result of cutting down the Amazon even more than it's been cut down. So I see that one for sure. So I guess inevitably, these are challenging problems, big rewards and getting it right. Competitors competitive penalties, if you don't how is Ambien IoT different to the other sources? This is a question that I get a lot. You know, people are like, well, we've got QR codes. We've got RFID. We've got cameras. Why? Why does Why does a computer the size of a postage stamp? Why is it worth considering that as one of the the tools?

    Thaddeus Segura 24:25

    I love this question, because it just want to answer it. I was like walk away feeling like even more like I made the right choice, because I really truly think we have a competitive advantage here. And every time I even say it out loud, I believe in it a little bit more. So let me actually answer your question. There's two reasons. One is first that it's passive. And then two is the infrastructure. So I've already described the labor problem everyone's facing today. You just can't add another scan. Like I know that sounds so simple when people are like, just scan everything when it comes in and you'll have confirmation like, we've done that we did that math so many times within Walmart and the number works out to be so much bigger than you think that it actually is. And above and beyond that you're, you're inaccurate, like, anytime you introduce a human activity, there's air incorporated with it. So any solution that we actually put out there, because we're trying to just solve these problems on the margin, it has to be reliable. And it has to be passive. And we are passive and we are reliable. And so that's the first answer.

    Steve Statler 25:26

    Expand a bit more on what you mean by passive. Yeah,

    Thaddeus Segura 25:30

    passive, I mean that there's no required human interaction. Like we I this tag, if this is applied to bananas, anytime this thing gets energy and calibrates and transmits, I get the data point, and not just one, but I get like 80. And then I clean it into something usable for the customer. But I understand things about the environment and like what it's doing to the capacitance of the tag, and the temperature, and all these things that we can use to go build these complex models. But I don't need anyone to do anything, we just walk on and go.

    Steve Statler 26:03

    So you're not having people wandering around. You don't have to train employ and rely on staff with handheld scanners, which, you know, will work often but not always, is the thing that's I was sitting in a meeting with a world class retailer. And you know, the thing that that it's funny how all our problems are so similar that like we don't have time, employees don't have time, we just need something to just work with less intervention, less things that can go wrong. And I think that's something that we can all we can all relate to. That's good. And anything else you want to say in terms of this ambient versus the alternative question,

    Thaddeus Segura 26:48

    just the next level down is infrastructure. Because when you look at these passive technologies, like computer vision, and RFID, and all these other things like they are very infrastructure heavy. If you want to monitor in stock with cameras, you need cameras at a set angle, looking at every single thing where you need a robot driving around, and it leads to a whole bunch of promises. We've tried that. And so the question then becomes, okay, if I'm going to have a passive solution, how do I get these signals. And the truth is that most technologies that are out there have like big, heavy, expensive, fixed infrastructure requirements. And we don't, we can put up a little bridge that fits in the palm of my hand. We can energize it from a battery, if we need to, we can equip it to a cart or attach it to a forklift or pallet jack. Anything that has power we can automatically equip it to and then suddenly, you can light up all the places that merchandise would go all the places that people would go for a fraction of the cost without having to saturate, you know, a couple of million square feet in a warehouse.

    Steve Statler 27:52

    Makes sense. So why not use robots? Robots are really cool, especially the ones that have full legs and look like dogs.

    Thaddeus Segura 28:03

    Robots are cool. I think that they are the future. I don't think we're there yet. We have some stuff to figure out. And I'll leave it at that. We can go deeper if you want to.

    Steve Statler 28:16

    But well, I'm interested in one of the things that I heard you talk about, I don't want to kind of go anywhere that leaves you feeling uncomfortable. But the cost of machine vision, everyone sort of thinks that machine vision is free. It's like okay, yeah, I have to buy some cameras. But after that, it's free, isn't it?

    Thaddeus Segura 28:33

    Yeah. So I can answer that question for sure. So, Amazon, we all think about it as a retailer. But Amazon has made their money over the last couple of years on AWS, it's the compute. It's not the books in the 17 packages, we all get at our front door every day, they make their money on Compute. Google is there like all these places have found out that there's a ton of money in doing your computation in the cloud, and they charge you for it. And by definition, it's not free. You know, everyone's excited about Tete GPT. But if you go Google or SgPT itself, how much it cost to train it, you might be shocked. Because these big models use more energy in the pre training process and some small countries using a year. It's a massive undertaking. And so when you think about complex computer vision models, it's not just the simple thing you're solving, you are doing a ton of compute in the cloud, and there is still a marginal cost associated with it. So if I'm trying to predict cats versus dogs, I need a little model that only needs to do two things. If I'm trying to figure out which item I'm seeing in a Walmart, I need a big model that's looking at 100,000 things for that specific store, and I needed to work fast. So every time I make a prediction, there's a marginal cost associated with that. And the marginal cost of that one prediction is probably in the order of magnitude of what we charged for like a month of the data coming off of our tags. And so if you're scanning millions of items a week, it becomes very expensive very quickly. And so it's not just this free thing you Build once and you deploy your pain for every image that gets run through the machine.

    Steve Statler 30:04

    Okay, that makes sense. Let's talk about what your team is doing. Because, again, people look at Willie out and they say, Oh, you're a chip designer, or you're a tag company, just give me the tags. But it turns out my reckoning your team is close to a very soon will be the largest team within our company. And you don't have anything to do with I mean, you use the data from the tags, but you're not designing chips, you're not designing tags. And, you know, back in the early days, when I was the first person that joined the company, outside of who wasn't in r&d, and my most common request was just give me the tags. I'll take the data, leave it to me. What so why do we need your team? And why was why was it so large?

    Thaddeus Segura 30:54

    Now, I think that in the data community, there's this famous joke of, you know, someone in marketing will come and make your request and use all this data. And you just go to the SQL database and say, select all from this pristine table. And if it works, that way, it would be a whole lot easier, and our lives would all be a lot more enjoyable. But it doesn't work like that. I mean, these tags, you think about what they do, they're harnessing energy from the environment across two different wavelengths. And then they're calibrating and then they're transmitting anytime they can actually get the confidence that something's going to hear them. And so inherently, it's opportunistic, and it's noisy. And it's RF. So you don't know if this tag went directly to the receiver, you don't know if it bounced off of three things. First, you don't know if there was a person in the way. And so it's not just a simple thing, have a look at the data and see where it was at there is all sorts of cleaning and processing and grouping and D aggregation and D noise that we have to do to make the data from this raw thing into something that's actually intelligible, actionable. And so there are multiple steps we have to go through to get from point A to point Z when someone can actually derive business value from the data.

    Steve Statler 32:05

    So part of what you're doing is kind of cleaning up the noise. So it's kind of a blessing and a curse. Suddenly, we have real time data. But there's a lot of data. And so it's like sifting through it. So what is the you know, what's the approach that you're taking to turn that massive data into what customers want? And what is it that you're aiming to deliver?

    Thaddeus Segura 32:31

    Yeah, absolutely. So again, my team has three different pieces. And so I have a research arm, I have an analyst arm, and then in the middle, I have a product team. And so the research arm, these are the the algorithm developers, these are guys from physics backgrounds, who are brilliant, understand how to take this raw data and to D noise it and turn it into something that's actually intelligible. And so I call that layer when we transition from packets to events. And so there's logic in between there that takes all this raw information coming off the tag and the metadata from the bridge from other data from gateways anything else, and turns it into something small we can manage. Now, the analysts take that data, the event data, and sometimes the packet data, and they'll start looking for insights that satisfy customer needs today. So these are people on the front lines, these are people that will go to the store and stock a shelf to understand what it looks like and what we actually need to be able to see out of the data so that we can clean it further, and be able to actually take these raw packets, which still may be in the order of 50,000 per tag, and turn it into something that actually would trigger an action to extract business value. And then in the middle, we take those things, and we try to turn them into reusable tools. So we have standardized events, but we can also group those events together through playbooks and trigger some certain actions. So maybe if you had an invoice, and it told you what was expected, and you saw items that weren't on there, or went to a different store, you could trigger an action to automatically correct inventory. And that is how we have this, this tearing of raw packets to events to playbooks to actions, and the actions are where the business value actually happens. There's so much work that has to happen in the middle that I need this massive team to do it. And so when we give raw data to a customer, and they try to replicate that on their own, it's just a massive learning curve. And it's just too much and so that's what we're we're trying to do and make that reusable and repeatable for everyone else. So they can actually go from a tag to an actual ROI without having to figure it all out in the middle.

    Steve Statler 34:33

    I sometimes joke that using this ambient IoT technology is like turning on god mode. You know, if we're all in a sim simulation, which arguably we may be, and suddenly you go from just seeing the bit of the game where that's around you to sing everything and then you know, I think a lot of people think about well, would I really want to be that God would I gotta go crazy because I just got sensory overload. There's just so much and I think it's Historically, it has been about producing these post mortem reports and a very simple set of figures that some VP can look at. Or maybe someone operationally can do something in a very slow way. But I think what you're describing is much more interactive. It's about operational support for the people that are doing the work as well as visibility for executives. Is that fair? Absolutely. Yeah. Very good. So what what kind of, in a way, see, we flip the switch? In godmode, suddenly, we can see things that no one could see before no one, you know, before there was a temperature sensor at the end of the reef, refrigerated container truck. You know, there was a daily scan. You're trying to solve the problems that you described, but what are their kind of surprises? What are the insights that you're seeing now that you're seeing data that really has never been seen before?

    Thaddeus Segura 36:02

    I don't know if there's surprises as much as there.

    Steve Statler 36:07

    So it does.

    Thaddeus Segura 36:09

    I think we've all known this stuff was happening, but we couldn't see it. And I'm struggling to find an analogy. But you know, you're really trying to map a room with LIDAR versus a soccer ball. And we're kicking it against the wall and configure what was going on. But we couldn't see all like the ridges or details. And now we can because the data is so much more granular. And so we're seeing all the things we expected, or seeing things that should be kept at very narrow temperature band getting way too cold or way too hot. We're seeing stuff like 20 to 20 degree variations between the top of the palate and the bottom of the palate. And that matters a lot. We're seeing Halloween candy get left on the back of an ambient truck and hitting 120 degrees Fahrenheit while it's in transit, presumably all melted, right? You're seeing doors left open on coolers and freezers and, and all sorts of stuff that we always do it happening. And now we can see it. And we can't stop there. Because it's not good enough to your point to just report the news. Like we need to be able to action. And that's what's so nice about seeing this stuff in real time is it's not just like, hey, you had this problem a week ago. It's like you have this problem right now. And here's exactly where it's occurring. Just go fix it. Or if it's a data problem, we can fix it passively without having to interact with anyone, and just rectify it. And so you don't even see it occurring. It's just all below the surface. And suddenly, things are just flowing more smoothly.

    Steve Statler 37:34

    Pretty good. Yeah, I think that's great. And I want to point out that we have actually, quite a few customers. So I think you've been very good at not being specific about where you're seeing what problem and what opportunity and so forth. So we this is this is something that's broad and is impacting the industry rather than any one particular than the, you know, what are the things that you're really looking forward to? What are the problems that most excite you that can be solved with Ambien, IoT,

    Thaddeus Segura 38:10

    the things get me excited, like, I think that all this stuff around inventory management, food safety, in the end, the things that listed will be the next two years of business, I think it will be great and build a lot of business value for big enterprise customers. That's what gets me out of bed in the morning. One, I think we have an opportunity to really impact ESG, like actually measure it, and actually give people the ability to action on it, where it actually occurs. And it's not just going to be this, this broad swath are purchasing carbon offsets at the ton level. Like, you'll be able to differentiate the amount of carbon created by one apple versus another based on the methods taken by that farmer. And so that's exciting. Beyond that, I think that the ability to do some really advanced things like we have experimented with different like chemicals that could sense gases as they were emitted and measure the impact of capacitance on the actual tag. Some of that stuff gets me really excited as you think again, about permissions or or even detecting things like biome outbreaks. Like imagine if we had sensors everywhere that the capacitance changed, as COVID was in the area, that would have been a game changer. We could have put up a billion tags and actually seen in real time where the virus was at like this microscopic level. I know that sounds like science fiction, and today it is. But the technology exists that I think we can get there in 10 years. And that's the stuff that gets me really excited. So I see the next two years as being these supply chain these retail use cases. Beyond that, I think things will expand. I think we'll get into things like agriculture and actually be able to start measuring if we're over watering, or ESG in the end and really hopefully start showing some of those insights directly to consumers. Because I think when we empower more people to make more responsible choices. electorate start moving the needle and putting pressure on retailers to make better decisions. And then beyond that those those next five years I think will start to get really interesting. And that's, that's what I'm excited for.

    Steve Statler 40:14

    I'm really interested in glad to hear what excites you, because those are exactly the same things that excite me. I feel like we're, especially the climate stuff. I'm excited on two fronts. One is I see, whilst it's great that we're now getting to everyone having to do an annual report. If you and I got a cancer diagnosis and annual report would be not something we would settle for, if this is a serious problem, we need to manage our businesses. And I really feel like carbon equals cost. And this can be the next six sigma that the world's best companies managed to. And the thing that really makes me excited and optimistic is, this isn't just about people being virtuous, or taking playing the long game and helping to solve one of the world's biggest problems, it's actually ways that companies will differentiate, make more profit. And and also, I want to have those strawberries that tastes so good, and you know, something that will actually impact people and make people happier, as well as solving some of these terrible problems. So Thaddeus, were you expecting to be doing what you're doing? Now, when you were wrapping up your philosophy degree? Was it do you study philosophy? Right?

    Thaddeus Segura 41:31

    I did. I did. philosophy and economics. It's a long story, how I ended up there, I was actually supposed to do engineering physics, didn't get into my top school, and then out of spite, went to my safety school and went undeclared. And then found philosophy, you know, freshman year undergrad, and loved it, and started doing that, but realized probably wasn't a safe career path, and then added in economics for a little bit of safety. But still, you know, went down the philosophy path as much as I could and still got a degree in that as well. But not at all. Not one bit. I don't think it was a term back then. But I ended up here.

    Steve Statler 42:11

    Very good. So how did you get here? What was your what's your career path been? Yeah, we should say your VP of law, what is your title? What's your.

    Thaddeus Segura 42:22

    So I'm the VP of data products and algorithms hear Willie out. And I'm pretty much responsible for taking the data that comes off the tag and turning it into a product that's actually usable for the customers and drives business value. So that is my role. In terms of how I got here, it wasn't a traditional path, came out of undergrad in 2009, during the height of the recession, and wound up in retail doing project management, and climbed the ladder, I ended up at Walmart. And by 25, I was leading teams of 300 people. By 27, I was leading teams of 500. And by 29, I knew I didn't want to do that for the rest of my life. So I pivoted over to product management. Working on automation robotics, realized I had major gaps there. So I went back to school and studied machine learning at Berkeley, and pivoted over to work on emerging technology with startups. And that led me to a role in supply chain modernization where I met Willie out. And after having met 100 startups and peeking under the curtains, I got a different vibe, familiar immediately, like it was really defensible. It had like an unlimited top end. And so I wanted to be part of it. And that's the whole story of how I ended up here.

    Steve Statler 43:35

    Well, I'm really glad you did. You know, I don't know if you've watched any of our podcasts before. But we have this tradition of asking people about their musical tastes, not because it's got anything to do with IoT. But I'm just interested. Do you are there three songs that you can bring to mind that are like favorites that have some significance to you?

    Thaddeus Segura 43:59

    So I have watched the podcast and this question actually stressed me out the most. So when I first joined the company, I watched many of your podcasts. It's like, I hope I never have to do this because of the songs question. So in my defense, most of the music I listen to is only at the gym. But I went way back into my memory and tried to think of songs that do have meaning. And I grew up as a as a huge Eminem fan, and all of his career over the years and he's, you know, 50 years old and still putting out music now. But there was a song in 2002 called Tilly collapse, which is a favorite song about him. It's just like pushing through. And I think that resonated with me back then. Because I had a tendency to think I could just run through walls and learn at some point in my career that that wasn't possible and you can't just work 100 hours a week, forever. And then here's another song called going through changes, which I think at the time was like 2010. And that resonated with me a lot as well because I had to reframe my mindset And it took me some time, you know, after I pivoted out of like tactical project management into like technology to really find my groove. And so here's another song called I'm back. And I felt like that was like the whole culmination of the cycle. So those are my three songs for you. That got out of the way. So we can we can move on. Yeah, no,

    Steve Statler 45:21

    that's, that is definitely broken some new ground first time that anyone has chosen three songs from the same artist. And first time anyone's chosen Eminem. So congratulations, you're truly an innovator there. And I love the the story that goes with it. Well, it's been. It's been great talking to you, Thaddeus, thanks so much for coming on the show. Awesome. Thank you, Steve. So that's it for this week. Thanks so much for for listening in. Thanks so much for staying at the end. And if you find this useful, let us know please rate review us. Share it with your friends. And feel free to give us some feedback on whatever the platform is that you use, or you can go to the Mr. Beacon website all the way up website and let us know if there are things that you want us to drill into and help with in terms of shining a light and getting other guests on but if you have thanks for listening, see you next time.