Hey, this is Justine from Two Black Girls One Rose, and I'm here to tell you about Zola. How weird does it feel to be called someone's fiance? The first time you hear it, you do a double take. From there, it's let's enjoy this moment, turns into we're planning a fall wedding, and I am right there right now. That's where Zola comes in. From a venue and vendor discovery tool that matches you with your dream team to save the dates, website, and an easy-to-use registry, Zola has everything
you need to plan your wedding in one place. Start planning at Zola.com. That's Z-O-L-A.com. 88% of the workweek is spent communicating, typing, talking, and going back and forth on topics until everyone is on the same page. It's time for a change. It's time for Grammarly. Grammarly's AI ensures your team gets the point across the first time, eliminating misunderstandings and streamlining
collaboration. It goes beyond basic grammar to help tailor writing to specific audiences. Whether that means adding an executive summary, fine-tuning tone, or cutting out jargon in just one click. Four out of five professionals say Grammarly's AI boosts spy-in and moves work forward. It integrates seamlessly with over 500,000 apps and websites, is implemented in days, and is IT approved. Join over 70,000 teams and 30 million people who trust Grammarly to elevate
their communication. Visit Grammarly.com slash Enterprise to learn more. Grammarly, Enterprise Ready AI. Welcome to the Mr. Beacon podcast. We've talked a lot about ambient IoT over the years, this new generation of Internet of Things where everything gets connected. And yeah, battery-free Bluetooth tags are one of the key elements, but actually designing the project, designing the infrastructure, choosing the scope. These are all other very important elements, and there's no one that knows
more about this from a practical point of view than Mattan Epstein. He is a polymath. He knows lots of different things, and he's a squad leader within Williott that has worked with an enormous range of extremely large companies putting this technology to use. And I got a chance to talk with him. For about an hour about his career, about his music taste, but also about all the lessons that he's learned using this technology over the years. I hope you find it as interesting and useful as I
did. Check it out. The Mr. Beacon Ambient IoT podcast is sponsored by Williott, bringing intelligence to every single thing. Patan, welcome to the Mr. Beacon podcast. Thank you for hosting me.
Oh, it's a real pleasure. I think you're doing interesting work. There's a lot of people doing interesting work at Williott, and I really try and make the show open, balanced, we cover competitors or at least alternative technologies, but I really felt like you are one of the people that has had the most amazing journey because you have worked with so many of our customers. And I really wanted to have a conversation with you about the sorts of things those customers
are doing. So the use case is the applications. And as you're a squad leader, I'm doing the hand waving part of it, trying to get people to come and get engaged, but you're actually the person that's responsible for making it work. And so you've learned a lot about something that's never been done before. These tiny, posted, stamp-sized compute devices and they're constantly streaming data and how do we turn that into insights and how do we turn the insights into business
value. So I think you've got a lot in your head that a lot of other people want to know. So let's unpack it a little bit and let's start off. Just give us a sense of the breadth of all the different sorts of use cases that you've worked on because you've worked on lots of different things. And then we'll zoom in on a few of the key applications that are really bearing fruit now,
but give us a sense of the breadth of what you've worked on. Wow. So we have been the company almost five years and transitioned like after the first year, probably from engineering to working with customers and a thin, from use cases from knowing what's on your shelf, using the tag to sense the shelf itself. Yes. So almost like a proximity sensor. Yep. From medicine, we did
vaccines, a bucket COVID, if you remember. That was amazing. So a tiny vaccine vial, tiny with with our tag, with a Bluetooth sticker on and it could tell not only where it was, then that it was authentic, but the temperature and the dilution. And the dilution. The dilution level of the vial, which was very important with COVID. Yeah. Yeah.
Because you didn't want to get injected by undiluted vaccine. But most important, to in my opinion, the most important thing was continuous temperature sensing because you certainly, some of those vaccines, you keep them at the wrong temperature of half an hour and it does you zero good. Yep. So that was amazing. And I think that's actually the first time I saw you was on this video of where you showed this thing working and being taken out of cryogenic freezing.
And that was a fun, fun couple of weeks when we turned around the team. I was just maybe the face, but we turned around from a design of a new antenna to a use case to the cloud features and functionality. And to the time that there was no flights, no one was flying. It was COVID. And I flew to Boston to a company that made the vaccines and they were at any day for the amazing. And I can tell you they were impressed. I'll tell you why I can tell you because I met the guy
that was our customer yesterday. Really? In Atlanta at a conference. And I'm like, looking at him, we're having drinks the evening before the meeting and I'm like, you look really familiar. And it's like, oh my gosh, we spoke with each other. In those days I was heading up sales for really hot. And I can tell you that they were just blown away by the work that you did. So well done. I was just the face. It was the whole team that did their job. It was the whole
company. The whole company focused on that. So I would vaccines. We did tracking cars in parking for new cars that are dealerships. So they passed through a couple of stations and the dealership wanted to know where the cars are. We did tracking of potatoes. You remember the potatoes? I do. From field. So from field to storage turns out, and again, I have a data background with engineering, but I learned so much on supply chains and storage.
So when you buy fries or chips actually, they are pretty fresh. They have like three months expiration date, which is not a lot for chips. But the potatoes also are fresh because there is six months season of potatoes. So you get fresh potatoes. And when it's not the season of potatoes, they store them for six months. So there is a cycle of a year almost. And the storage is very, very sensitive because potatoes are alive. So they can, if there is too much
humidity, they get rotten. If there is too much heat, they start to grow again. And you don't want it. You want the right test. So the humidity is important and the temperature is important. And what we went, that was really back then before the squad didn't even, I went and I was like a meagiver. And I tagged the whole warehouse. So I took a fork lift and it took me up. And I just tagged all the pallets. It was like five stores, high of those big pallets of potatoes.
And what we created, and that was, I think, one of the coolest data insights we started to generate in the beginning was a 3D structured of these potato pallets with temperature. So that's three dimensional heat man in real time. In real time. Yeah. Of course, you take another, you see, if you have issues, but that I enabled continuous improvement of the surrounding of the potato, the environment, the temperature, the humidity, eventually to get the product and to lose, not to throw too many
potatoes in the process of making us fries. This is the potato chip company. That's a potato chip company. Okay. The biggest you have. And then you did literally the grocery farm to store. So all those, all those, all those projects were around the same time. And that was the time I moved from core engineering, building models, data science, building the infrastructure. Recently we started to change as a maturing company. And we didn't know how to introduce the
data to customers. So Tal or CEO asked me to join him and to go to customers in Israel and understand their problem and see how can we solve them using our tech. So all of those stories are in almost the same time. We just went to customers and say, hey, what's your, what challenge do you have? Let's try to solve it. Matam is going to come and he's going to do everything. And from a data scientist, I became a solution engineer. I do solutions now. And we went to one of the biggest retailers in
Israel. And they had an issue with their supply chain, with their first supply chain. And they wanted to know everything is automated in the supply chain. And this is unique to this retailer in Israel. It's a very technology, protect technology. So his whole distribution center is automated. And he claims, claims everything is fresh. Everything doesn't stand in the DC more than two days. And the suppliers are the problem. It takes too much time from the moment
they pick the vegetables until it gets to the DC. After that, it's perfect. We said, okay, no problem. Let's try to show you two edges. We take one supplier. We take a few stores. And we see the whole flow. So we found the zucchini supplier that supplied the whole supply of this retailer for the kidneys. So one supplier, whole zucchini is for 400 stores. And we tab all the zucchini crates that were grown from this supplier to the DC, to the stores.
And we met them in the store. And we saw a lot of gaps. So we saw that the time it goes from the supplier until it arrives to the store, could be 10 days, 12 days. But the claim was it doesn't stay in the DC more than two days. So where does this time go? And they found that eventually it is in the DC. And it's inbound. It's spending more time in the inbound than what they thought. So only looking at the edges started to give a lot of information to the customer.
And to find those problems in his system that he could fix easily. I mean, you and the company has always been working with so many customers. It's amazing. I think, you know, by my count, there's probably 800 people that have bought kits, played with the technology. Overall, the feedback's been really good. We've surveyed them and we got amazing results. But what we found, in my opinion, is that this is a paradigm shift. It's not just like taking a
widget and putting something that goes faster or is cheaper. It's a paradigm shift to holistic, continuous visibility. It's not when you tap. It's not when you scan. So the volume of data is massive and more data is good, but actually sifting it and finding the things that you're really interested in making actionable is hard. So it seems like what we've figured out is in these early days, we need to be involved in order to make it work and you're leading one of the squads
that has seen a lot of action. And again, by my count, we've probably had like 50 pilots, proof of concepts, call them what you will, where we've had some great experiences. And many of them have been very, very successful technically and actually successful from a business point of view. But for good reasons or bad, they've just not gone into mass production. I mean, one was, I remember we were going to scale with one of the biggest logistics companies in the world.
And I spent a whole day training their entire director and above management in this new ambient IoT technology that was going to allow these parcels to be tracked in real time around the worldwide network. And then they basically got rid of all the senior management and
boom, you're starting from the beginning. So we've had a lot of those experiences. But the cool thing is you've been involved in some of the other projects, which are a handful of incredibly large projects for some of the biggest companies in the world that are scaling. I guess I'm just lucky, I don't know. Yeah, I, you're like the zealig of for Woody Allen fans, maybe they'll mean, Matt Anne keeps on showing up in these historic times,
historic situations. So there's one use case that hasn't scaled yet, which I think is really interesting that I want to talk about because what we find is you don't just put an enterprise online overnight. You don't give omniscient visibility of everything. You have to slice it, dice it, start somewhere and grow from there. And generally speaking, our customers have either taken a narrow use case or they've maybe taken part of the supply chain and just tracked from the
DC to the store. But there's one project that you worked on, which I think is super interesting where it literally went from the dock to the, to the store. So it really went into end from the point that the, the fish was caught to the store. Can you tell us a little about a bit and we've actually done a few of these. So you can maybe interchange a few of the pieces to protect the innocent. But what I remember, at least one of those projects where you came back and you kind of filed
this report and it was like, oh dear, everything's pristine. This looks like there's absolutely no problems at all. Because we're looking for problems that we can solve, right? Tell us a bit. We're looking for opportunity. I want to go to make things better. Yeah. So I think the end to end approach. And I said also on the retailers in Israel, we started with the edge. So we did, we started with supplier to store, which everything in the middle we just didn't see. And of course,
that's what was kind of giving him a bite and he wanted to see more. So we started to filling the gaps for this retailer. And we've done this also for a retailer in the US. It wasn't for fresh. But for not fresh from supplier to store, including the DC. And we've done it in different countries. And to end the project specifically you're referring to is, it's really interesting.
You have different stakeholders at each. And so if we're going from a supplier, you have a supplier, maybe he's actually producing the produce if it's cutting the zucchini or packing meat or container or oranges, right? We've seen also oranges. So you have the supplier, he's professional in this. And usually he's not really great on logistics. So he uses a third party to ship his supply to his customer, which is the retailer. And so that's usually the
primary shipment or the first shipment. So really two hands, right? We have the supplier shipment. It goes into the DC, the distribution center. Sometimes you have also more cross-ducts in the middle. But let's take a simple supply chain. You have the distribution center where you have, goes inbound, can stay at the inbound for a couple of hours, maybe a couple of days, going to storage, then we'd go to picking. So you have other automatically or people go and
pick pallets. And those pallets will go to the store. That was third hand at least. It will go to a store usually with transportation of the retailer. So that would be the secondary, freight, force hand, arrived to the store and then probably also stay a bit in the inbound of the store or to cooler and then go to the front of store for you to buy it. So really five, six, maybe seven hands where every hand is like an agent problem, right? Now you want to optimize
the supply chain, but you also want to optimize your benefits, your KPIs. And there is not enough information that is flowing end to end which could optimize the entire KPI. Or you have minimum time. If you're talking about fresh, the right temperature, the right timing, eventually for you're the customer to get the best product when you go in by the other shelf. You don't know if one package of products has been in a different journey. Which one should you choose?
You're innocent. You want to choose the best one. You choose one. So what we find, we found a lot of things. But those are all opportunities. Each of those hands or agents, they want to improve their operation, but the goal eventually and you see it from everyone. The customer, the customer is the goal. The person who is going to eat the meat, the meal, the fish, he's an God. He needs to be happy. And everyone is driving this. So every step we find when the opportunity is other than the
supplier, but sometimes he doesn't know how much what would be the order quantity. So he produces until he gets an order. And he may have extras. And those extras are going to be sent the next day with a new shipment. So we already have a mixtapment. But how is it managed when it goes to the DC? You need to know that you have a mixed palette and maybe how to pick the palette or where to put this palette with thin issues of FIFA, also in the distribution center. So first and first out,
theoretically, it's actually not first and first out. Yeah. Because it may be first and first out, because this is the order that it came in. But it wasn't the order that it was produced. So it's not first expiry first out. Because it's hard to manage in supply chain or in those systems. So if we have identifiers that are put on early, where the tracking is automatic, you're not relying on scanning, then you can start to see when the queuing goes wrong. And I think kind of love
the first stage is let's really have first and first out. Let's not have zucchini that is sitting at the back of the packing shed for several days when they should only be there for a few hours in the heat of the sun and the farm. And at every stage, you can get the order can go wrong. Right? And so your, the first level is let's try and avoid that happening. And then it seems like you can graduate to accumulating temperature over time. And then you say, no, we're actually going to,
but this, this unfortunately got left out in the, the, the heat. And so we're going to give it a fast track so that we can use it before it goes back. The best guess scenario that you have an expiration date that is given to the product. But as we see the journey, we can have a dynamic expiration date. Something that was kept in better environment may have longer expiration date.
And on the other hand, something that was in the heat was in a trailer and we've seen it. A trailer that the temperature goes up and down, up and down, something that talk about maybe strawberries. And they need to be in five degrees Celsius. And we go them, we see them freezing and going up to 10 degrees, freezing. That's logistics. We can improve it. We can, the company can go and improve the trucks. Maybe there is an issue. But the store is this specific batch probably should
be sold earlier. Maybe not discount even. Right? Because you, you don't want to waste. You want customers to know what they're buying. To use it and to pay the right price. Otherwise, it's going to sit in the store for two days. And it's going to start to be mushy. And no one will buy it. That's strawberry, you know, strawberry just case. So what is your sense of the opportunity to fix things? I think when you don't know what's happening, you know, I always assume this stuff
kind of works pretty well. But I think what we're seeing is that there are more problems than people know, simply because when you turn the lights on, then you start to see that. So one of the concern always with working with when we do end to end, five, six, seven, hands, is that not all parties are going to participate. Right? Because you may find out what we call opportunities, which for them is more work or change of process. And if it's not broken, why fix it?
But what we've seen time and again, remember this quote is like, if we don't know what's the problem, we cannot fix it. And this is the other manager that always wants to optimize. They always want to be more efficient. They always want to get their next customer in line happy. So if it's the DC, it's a store or trailer, right? If it's a store, it's the customer, if it's a supplier, it's the DC. They want to make everyone happy. And there is always an opportunity to fix. And
there is always happy to find it. And I think there are two, you can approach opportunities into you can second things into two things to opportunities. One is you can look at the data, do analysis of the data and improve operations. Because you see that there is an issue that happens time and again and time and again. So you go into the store and when a shipment arrives, it always
stays two hours in the back room. And this is something that can be improved. Why? Because probably you don't have enough staffing for the store or the store doesn't know the ETA, the arrival time of the trailer. So they're not prepared in time and we've seen it. And so this is just looking at data over time, seeing trends. And that could be in the entire supply chain and fixing it. And then you probably fix it once. But since it's people and it's supply chain and there is a lot
of unknowns, you have opportunities that you could fix in real time. For example, a palette was mistorted. It was loaded to trailer one that is going to store one. But it was supposed to be loaded to trailer two, which is going to store two. If you catch it on time and this is going to be a simple alert, maybe a light above the doctor, you solve, you save a lot of money. So I want to kind of turn the spotlight on that because we've been at a free-ranging conversation.
We've talked about a lot of things that go wrong. It can be fixed. But this very simple thing seems to be one of the areas which is low hanging fruit, which is essentially preempting mischipments for valuable things. And you know, you can solve this problem in lots of different ways. But it turns out that putting a Bluetooth sticker on something allows you to spot mischipments without relying on manual scanning and without relying on a lot
of expensive infrastructure. You've got low cost infrastructure, no manual intervention, and you're preempting something that goes wrong more often than any one would have thought. It seems like there's more mischipments than we expected to see. Is that fair to say? It really depends on the operation. What I've seen, that it always happens. Again, not because people are bad or doing it intentionally, it's because we're human and we make
mistakes. And we've seen probably 1% in terms of palette level and 1% is a lot. It means that probably every foreshipment you have a wrong palette on a truck. This can be eliminated. We've seen processes to eliminate it with scans. So every time a palette is going to be loaded to a trailer, you scan the palette and scan the trailer. This adds labor. And again, because it's an operational environment with stress, with noise, with people that are working on a foreclist
for 8 hours. Sometimes they forget to scan. This data is not always accurate. And what we come and do and we replace it, we put a really cheap infrastructure on the dog door. We put a cheap pixel on the palette itself. And then you get a misert alert when the palette is going on the trailer. But the nice thing, this palette is still tagged. So when it's going to arrive to its destination,
you're going to get an unload signal to know that the palette has arrived. And if it's a palette that needs to be in call chain and it's left on the dog door, and this is only with dog door infrastructure, you would know that the palette is left in the back room and get an alert to take it to called storage. So this is with one tag, we just did an end-to-end with one tag and with a really, really simple use case. Yeah. There's a, I think it's a difficult question
to answer. So maybe there isn't an answer, maybe there are just things to consider. The question of do I do end-to-end or do I try and optimize one step in the supply chain? Do I just focus on DC to store, if we take a grocery example? What do you think the pros and cons are of your customer who started tagging at the straight off of the boat when the fish was caught? And then they see it all the way into the store and see that because versus just starting at the
DC and going to the store, what are the, we've seen both. We've seen both. Both can work. Both can work. It's usually a number of games. So it's a balance between how much time do you want to invest in seeing enough product arriving to your destination. So if we take one supplier, like we did with the bikinis and it supplies for the whole country and in the edge you're only at five stores. It means that you see 1% in order to see enough of the kidneys coming into that store and seeing
the whole journey, even if you're in the DC. So you want to see the whole journey. Then you need to do a lot of the kidneys or a lot of time, a couple of months. So this is one approach. So you see the end to end. It takes you a lot of time. Another approach is let's start with something limited. We're taking one store or two stores, but we're tagging everything that is going to those stores. So in a week you probably get the same
amount of data. Both approaches are right to start with and I like starting with both approaches. I prefer to start with a approach that you see end to end, at least for a couple of weeks. And then you optimize just specific use case. You start to scale it. So like me sort, for example. And you start to build around it. You add the layer and add the layer. You can always add more layers, right? But it's also always need to hear what's the customer
pain point. Just today I talk to, I'm working with one of the biggest online retailers. And it has different business verticals. And with two different verticals, one is taking the, I want to see everything fast approach. So we're going to do a quick one week with the everything and then decide what's the most important thing to start with. And another customer, he knows what's its problem. Another vertical, he wants to do only from what he calls inbound. So knowing that product has arrived.
It's the only thing that he wants to know today. He will expand it later on. Ryan Reynolds here for I guess my hundreds mid commercial. No, no, no, no, no, no, no, no, no, no, no, no, no, no, no, honestly, when I started this, I thought I only had to do like four of these. I mean, it's unlimited premium wireless for $15 a month. How are there still people paying two or three times that much? I'm sorry, I shouldn't be victim blaming here. Give it a try at midmobile.com slash switch,
whatever you're ready. $45 up from payment equivalent to $15 per month. New customers on first three month plan only taxes and fees extra, speed slower above 40 gigabytes, seat details. And now a next level moment from AT&T business. Say you've sent out a gigantic shipment of pillows and they need to be there in time for International Sleep Day. You've got AT&T 5G so you're fully confident. But the vendor isn't responding. And International Sleep Day is tomorrow. Luckily,
AT&T 5G lets you deal with any issues with ease. So the pillows will get delivered and everyone can sleep soundly, especially you. AT&T 5G requires a compatible plan and device. 5G is not available everywhere. C-A-T-T dot com slash 5G for you for details. Very interesting. Yeah, and I think, yeah, you do the kind of the sample. You see end to end. You see where the opportunities are and you figure out where's the ROI that I can
start to justify that. And I think, you know, one of the few projects that you haven't worked on are postal customer. They started off by not lighting up the dock doors, but lighting up the trucks. And now they've got like 5,000 trucks online and they can see all of the rolling cages,
hundreds of thousands of rolling cages in the trucks. And I think they were super smart because they realized if I can see the load factor, if I can see the utilization of the trucks, then I can prove the ROI of tagging hundreds of thousands of these rolling cages and I can start to see where they were left, where they didn't come back. And, you know, they didn't light up everything.
They didn't light up all of the sorting offices. They just started with the trucks. And there was enough of an ROI that they were, you know, this project has got huge momentum now and now they can go and start to turn the lights on in the sorting centers. And I think that seems to be the way it's going to go. Well, the most skilled customers and it is an art. It's a skill to figure out how am I going to make the business case? How am I going to show the return and then how I
going to build? I think they'll identify that pain point, limit the scope to something where they can get a result relatively quickly and then expand from there. I'll add to it one thing is that I think William to the company has matured a lot as well over over the years. When we went four years ago and we installed in what you see the edges, part of it was us learning. And today we can come with a lot of confidence and say, no, we can give you those three solutions. We can
connect them and see end to end. But we have a mature solution that we can do for you, which is only roll cages and trucks or only dog dog compliance. So those solutions, which we have matured over time, are also way to start. Because it's less of an exploration, you come with a built-in proven solution and you can expand from there. And this is also helping customers to understand where they want to start. And that's also double edged sword for us. The thing with William
is you want to be everywhere, right? You tag once in the supplier, you see everywhere. But where do you start? You cannot start. You cannot blow up the entire system and change it over day. You have to start gradually. You have to, it's like the frog in hot water, right? Slowly. And what are the things that you've seen work in terms of the softer skills, the, you know, what are the customers done that you like that seem to be a recipe for success? We've been talking
about the scope and where you start. But are there other things that you've observed best practices? Well, eventually we provide customers data. And the stronger, strongest customers that that I've seen are the ones that understand the value of the data, what it gives. And how to use it, how to implement it. Because if you know what data you want, we'll make it happen. And sometimes the customer doesn't know exactly what data that he want or how they wanted to look
and what value that would let him give him. So we do a lot of exploration in the physical world. We install out of infrastructure and those iterations are long and tedious. But when a customer comes and say, Hey, I want to know one thing. I want to know me sort. Give it to me fast. Don't give me too much too many false positives. Focus on this. So this is the use case. We know the use case. We know what data does the customer want. Then we can build a solution, optimize a solution
that will answer his needs. And in my mind, that moves the fastest and shows the value to the customer and not only shows the value goes into operation and production. Have you seen any patterns in terms of the structure of the project kind of executive sponsors? What sort of disciplines do you need on a project to make it work? It's a balance between being open to innovation. But at the same time, as a customer being with their foot on the ground and or exactly what they want. Because
it's not an innovation game. We want to provide value. So what I usually found that works well is executives who come from operations that they know the problem that they feel the problem. But somewhere in their life, professional journey, they've done something that is tech-related. Maybe they have data background and they want to do operations or maybe they have RF background and they want to frequency. Yeah. Yeah. Right. They're frequency and they went
into operations. But what always helps are people who feel the pain. They know what they want to solve. And they have enough experience also with the operational teams that they can go and sell this internally. Yeah. I think it's a positive, but it seems like what I've observed is there's a pretty high level executive sponsor. I mean, there are some technology. It's always good to have
an executive sponsor. But I think with ambient internet of things technology, when you're reconfiguring a supply chain, you're almost always going across functions and you're changing the way people work. And so if you don't have a pretty senior executive that has real power and gets it, then it's very hard to make it work. We don't change only how people work. We change also how people think. So if we take a simple doctor example, today most logistic change. You want
100% accuracy, right? You want to know that 100% of your pallets are loaded to the trailer. And with greater frequency, you guys close to it, but you don't get 100%. But you know what also with scans? Scans, you don't get 100%. You get 70% usually, right? When people scan. So it's a change of paradigm. And you need this flexibility and you need this executive that could help and and convince that this is important. I think that is actually a really important point. It's sort of
it reminds me of the self-driving car thing. Self-driving cars will kill people. It's just that they will kill a lot less people than human beings driving. So do we stick with human beings driving or do we move to a system where the and obviously we're not there yet. But in the future, self-driving cars will be two, three, four times as safe. But they will kill people occasionally. And I think when
you've got something new, it's very tempting for people to say, oh, I'm exposed here. I want to see 100% accuracy. And if you you can give people almost 100% accuracy, but it becomes very expensive. And then the whole benefit starts to fall apart. You over configure the infrastructure. You make the process more complicated. But I think what I see is if people are willing to open up their minds to the fact that we can give you a continuous view of inventory and assets moving around.
Not a dot that align. And when we do that, we can afford to have a few mistakes because we'll fix them because we lose sight of something. There's some water that gets in the way. A metal roll, a roll-in cage gets in the way. And then you see it again. And we've got the cloud and we can string it together. So people need to sort of, basically, it's like learning to swim. You have to just eventually let go and go for it. How to trust it. You have to have talented people that know
how to work with the data. Yeah. Yeah. So tell me what does the squad look like? So we talked a bit about this, but let's come up at it from a different angle. What are the, if I get a bill by own squad, and most of the squads are willing, we come in, we have all this multi-disciplinary team, and things happen pretty fast. But what does it take to take these Bluetooth stickers and get them to talk to Bluetooth radios and integrate them into enterprise systems?
So how do build squads are based on proficiency? And the squad can be bigger or smaller, but the proficiency is important. You need a later-erless squad that could work with customer that understands their pain points and bring it back to the squads and also to the R&D unit. You have data proficiency and that could be also squad-erless squad can be one person.
I've got a squad leader, I want a one squad leader for a while, right? So that can work with the data that can develop new functionalities or what we call business events that will be sent to the customer. Usually on the squad we have field engineers as well, so we'll go into a site survey, install tests, iterate on the installations, and those are field engineers
and also technicians. I want to go back, take a step back. So I think the field technicians, it's pretty clear, you need people that can put up the readers in the right place and develop a process for doing that that can scan. In one case we've got a customer that's lighting up up two big box stores a day and they've figured out, oh, we need a mobile app that will allow those skilled people, not familiar to do these things and make it easy to really have lots and
lots of people deploying this infrastructure. But I want to take this data person. So we have a tag, it's broadcasting maybe hundreds or thousands of these packets. And what does this data analyst, data scientist do to take all of these low-level broadcasts, which is basically a tag saying, I'm here, I'm here, I'm here, and then that signal gets read by potentially not one, but many readers. What, how do you turn that into a business event and what is a business event?
So, it's a great question. It all comes from the business need, right? So the squad leader wants to work with, like after we're to understand where the pain points, what use case are trying to solve and what data, what's the structure of the data that was told this is this use case. For example, when we talked about the, we use this example a lot, but the
mistorting. So we want to know that pallet was loaded to the right trainer. So the use case, if we go higher level, when we talk to the customer, he had a problem that pallets are loaded to the wrong trainer. The data that he needs is telling me to which doc door, because I don't know the trainer or read or other knowledge trainer, tell me to which doc door is this pallet loaded. And we put KP out of this. Tell me it up to five minutes after the load with 95 or 98%.
What, why do I not want to know within five seconds? Maybe you want to know within five seconds, but how much how many to pay for it? So it's always a balance between the motor infrastructure and also the cloud cost of actually processing it that fast or maybe doing a divin on your edge. And what's the customer in it? And also I guess this is the case where something gets loaded or a truck that has really
been loaded on the truck. It's just been someone carried in and then they carry it out and they put it on another truck. But you can give a couple of signals and that's fun. And if you want to be the confidence and would help we can add those or confidence level in the data that we give. So once we have those, the use case and the data event and requirements, then we have a solution engineer which is part of the proficiency of the squad which really
understands how our tag works and what's the network. When you said reader, I don't like to call them readers, they don't only read the the energized and they also give more data when they receives the tag and upload the data. So we have a solution engineer which is which really know
how the tag works, how the network works and how the cloud works. So together with the data, data scientists of the squad, they sit together in the design solution and where the engineer, the net, more of the solution or the network will come with different considerations of what that's a doctor, what infrastructure would be installed on the doctor maybe and this is actually like a better maybe know infrastructure on the doctor. What is a forklift because
a pilot doesn't go on its own to a truck, it will go on a forklift. And this forklift can be our network of saying I'm a forklift and I loaded it into a trader or in through a doctor and and that's actually much cheaper and even faster. So the forklift is energizing the tags, reading the tag and somehow you're tracking where the forklift is. So you know where the forklift is and you know that it's seen a trailer and you start track the movement of it and you can start
to get pretty smart about what's happening. And more of an infrastructure is the future and this is how we go. So going back to how we build the solution and what's the business event? The data, the solution, the solution engineer is going to organize the solution, simulate it as well on a simulation platform that we have and the altitude of the simulation
would go to the data proficiency of the squad. Again, it may be the same person but the data proficiency will ingest this data and start to work on algorithms that will give you the cleanest signal because the customer doesn't want to know all the small movements and all the multiple hundreds of signals. He only wants to know when defined by the event by by the business event requirements.
I want to know that the pilot was loaded to a dock door. So the data scientist will be the model to extract this in the highest accuracy with the lowest source positives. And this is all based on simulation. The next iteration would be for exactly the same team to go to the field and test it and optimize it and understand why did the simulation didn't simulate
all the other edge cases. And at the end of this process, this small team is going to deploy a business event to the cloud and this is what the customer is going to consume. So the the customer is just interested about a missalt or a misslaid, a match for their business event. But there's some parameters around how quickly they need to see it and what level of accuracy is going to give them the return on investment that justifies the justifies all this. So that's
what a business event is. And then you just cry. It seems like there's a recipe here for where you have the readers. Are they going to go? Are they going to be in people's pockets? Are they going to be on a fork with truck? Are they going to be bridges, gateways, readers? The bleachers devices. Are they going to be behind the dock tools? What do we call that? So eventually the solution is going to be into a blueprint. And then we have a great team or
customer success team which kind of build this recipe of how to build a blueprint. What process do we do eventually to give the customer this blueprint which explains exactly how should you build your deployment in order to get to the KPIs that you regret it that we're really about. And this blueprint is not only worth to install. You think of a blueprint and you think of a
blueprint. I think also father or architect. So we're familiar with the blueprint. But the blueprint actually gives you the way to validate the installation and the KPIs and the monitoring tools that will monitor that the KPIs are actually always met. And when we deviate from those KPIs it will alert and we'll have some tools as part of the blueprint to know why it did we deviate. Maybe it was a changing process. Maybe after two years the customer changed the foreclist.
So twice as faster. Maybe it has drones taking pallets into I don't know what will have in the future right. So we always need to monitor and then we have a data quality to the thousand. Very good. We've covered a lot and I kind of want to quit whilst we're ahead because I think we've covered a range of use cases. We've talked a bit about the shape of a project and whether you should. What are the considerations going into end? What are the considerations of just kind of
doing a stage? We've talked about the structure of the team. Now we take business requirements. We get a we define business events that are keep forms indicators. We contract for that. And we design a blueprint that basically specifies how we're going to replicate this at scale. I didn't say any other key things that we need to cover. It's fun. It is fun. It's
fun. We do it all but it's every day it's never a boring day here. We're always learning and we're always improving our product and everything that we just talked about is it came from doing the thousand mistakes and learning. I learned a great great team that is able to learn and smart enough to create solutions and think of how to solve problems that you didn't know that existed a week ago. I think that's true. A part of fun is what's the
difference between where I can play? One is kind of you got clear goals when you play a game and the goals are very clear. Reality tends to be a bit fuzzy but I think there's a lot that you notice there's sort of this excitement about all the problems that can be solved if you take all of that intelligence in the cloud ALI and powering, choosing and apply it to the physical world. I think we've seen enough here vaccine vials or fish or potato fish
problems that can be solved and we're going to make life better. We're going to reduce waste. We're going to make food taste better. We're going to reduce carbon footprint and it's not often that you get to do something where you can really make life better. I'm hoping that other people get to join in as we scale and solve some really important problems whether they're to do in the environment or the cost of food or the efficiency of how we tackle health care.
The opportunity is huge and I think we all see that and it's tremendously exciting to see real progress being made in solving these real practical problems at scale. You're the one of the people there is in the frontline that's making it happen. How did you find your way to William? I won't go into all my background but... Well I mean I have and like you were a commander in the idea. You were heavy stuff. Lots of people, lots of responsibility. Things that a lot of people in
my part of the world never experienced. This is something that everyone does in Israel. Army like everyone, I studied in university. I went into industrial engineering and this is kind of the engineering that you go if you don't know what you want to do. It can't touch everything and I really liked working with data and writing code. So I explored our masters and did more data science master and I worked in curious algorithms which today we called AI.
So it's deep learning and reinforcement learning. Really cool AI, really interesting laboratory and parallel also did my MBA. So I always were on kind of between coding, data, tech and business. So maybe a bit product. I worked a bit at Intel, I had a data scientist so optimizing CPUs with data and then tried to do my own startup. So we worked a couple of months and weren't able to fund and get to raise money. But it was an interesting idea. Tell us what the idea was.
So it was one of my ideas when I came back from my MBA. I did my MBA partially here in Michigan. I was in business school. So I came back from my MBA. I finished my MBA master's degree in engineering so I can do everything. Okay, let's open a startup, join a friend from school and we
called the AI industry. That was the name of a startup. We tried to create a professional social network for professional that are not tech or finance but for the talent industry, for TV, for filmmakers, for actors which today they work out with agents which cut a lot of the talent because they want to work with whoever is going to bring them most money. So new talents are hard to get in the industry and they're kind of redundant. They get a lot of money for
there are good ones but there are a lot especially in it where a lot of the just blocks. So we tried to kind of fix this problem in the market. We had good support and we had a lot of actors but we weren't able to get finance and also some personal stuff. So we decided not to, sometimes you need to decide where to stop. Super important. And I started to look for a job again. My wife,
back then my girlfriend told me that it's about time. So I had a couple of job opportunities and a good friend who worked with me at Intel told me that he's working in this amazing startup and he's building the cloud team. There are only like five people and he wants me to join and join the data part of the cloud team. That was Tricca. Tricca didn't nothing the company anymore but that's how I joined the company. So he had a lot of achievements and one of them was bringing
you. You've done some amazing work at Willio. Okay, well in the other part of the show we explore a lot of that work. So let's get to the really personal stuff which is the three songs that I have most meaningful. Yeah, so that's I think the hardest part of today. So everyone is no, we'll know already if they haven't already that I'm from Israel and my accent is Hebrew. So I'll have one song in Hebrew. So the first song I'm gonna do it from the light
to the heavy. Okay. So the first song is Lovely Day. Which is a fun song? Fill with us, right? Yeah. You're going to wake up in the morning. So I have a daughter. She's year in two months. She just started to walk this week. Oh, congratulations. That's crazy. And I almost like every I just thought about it. Almost every morning I put this song for us like in the morning. Yeah, just dance a bit and you know, see her face and I'm happy and that's
gonna be a wonderful day. Does it for a song? Fantastic. Good start. Yeah. Second song that I'm meaningful for me in Hebrew. The name is Sweetwater but the accent translation is probably fresh water. And it's telling a story about finding a fresh fresh water in the desert and how it really resembles finding love in life. And you know, you look a lot for it and you may you know,
move some bushes and get hurt on the way and not find water. But then you find a fresh water and it's something that me and my wife today, like in the beginning of the relationship, when we had some struggle, we listened to this song by artists that I appreciate Meir Arreal. So that's the second song. All right. Lovely. And the third song which I chose, it's relevant to our Israelis today. And the name of the song is I'm coming home. Other either American version,
but there is an Israeli version. Today we have 349 days that we have 101 hostages still in Gaza. So that's a meaningful song to me. If I had to bring it. Yeah, that way is heavily on a lot of people. Yeah. Wonderful. Thank you, Mattan. I really appreciate you coming on the podcast and sharing those songs and all of your experience. Thanks. So that is Mattan Epstein. It was a great conversation from my perspective. Thank you very much
for listening. Thank you very much to Aaron Hallock for doing editing and creating the media that you're listening to to Sierra Walden for publishing it. I really do appreciate your loyalty, your tenacity in listening to some quite challenging content. Be safe. And until next time, take care. While we're in the darkest months of the year, your headlights are working overtime and dimming over time and can lose up to 50 feet of visibility before burnout. That's roughly
the length of a tractor trailer. Before they burn out, switch to easy to install Sylvania's Silver Star Ultra headlights designed to improve your driving visibility by directing brighter light where needed. If you're wanting better visibility, look to Sylvania, the experts in lighting with 100 plus years of expertise. Switch today at Sylvania.com and see better tonight. 88% of the work week is spent communicating, so it's important your team does it well. Enter
Grammarly. Grammarly's AI helps teams communicate clearly the first time. It shows you how to make words resonate with your audience, helps with brainstorming, and lets you instantly create and revise drafts in just one click. Join over 70,000 teams and 30 million people who use Grammarly to move work forward. Go to Grammarly.com slash Enterprise to learn more. Grammarly, Enterprise Ready AI.