Vested Capital
Vested Capital

Episode 13 · 1 year ago

(EP13): Marcel Herz Co-Founder And CEO Tiliter, Computer Vision Startup Disrupting Retail Checkout


Marcel Herz is co-founder and CEO of Tiliter, a company that is accelerating the democratization of computer vision.

Presently they are focused on the grocery store industry, offering a device used at checkout to scan and identify fruits, vegetables, baked goods - anything that is loose without a barcode.

This is just the tip of the iceberg however, as you can imagine being able to scan and identify real world objects will have countless applications, especially as the software gets better and quicker.

I made a small angel investment in Tiliter via the AngelList platform led by Brendan Hill, a previous guest on my podcast. I'm excited by the traction Tiliter already has - currently in hundreds of stores around the world - and also where this technology is heading.

You can listen in to this interview to learn how Marcel and his co-founders built the prototype of the Tiliter checkout, then began approaching large grocery store chains to test it out.

Enjoy the interview.



Hello, this is Yarrow and welcome to vested capital, episode number thirteen, featuring my guest marcel hearst from telit. bested capital is a podcast about how people make money and put their capital to work. I interview start up founders, Angel investors, venture capitalist, Crypto and Stock Traders, real estate investors and leaders in technology. Today I'm extra excited to have my guest on the show because he is the CEO and Co founder of a company that I actually did an angel investment in via the angel this platform, in particular Brandon Hill. He was the lead of that investment in this company. It's called telter. I totally butchered the pronunciation of this company. In fact, in my head I've been calling it tillater for many, many months, ever since I invested in it. I never heard it pronounced before, so this is the first time I get corrected. It's telter. Marcel is the CEO and Co founder, as I said, and not only am I excited because I'm an angel investor in this company, but I'm also excited because this is some very cool technology that I think has some massive ramifications for the future. So teleit is accelerating the democratization of computer vision. Actually, hear Marcel explain, their entry to market is actually in the groceries retail sector and they're using computer vision to scan products, in particular fruit and vege so fruit and vegetables which, as you might know, often doesn't have a bar code or in order to get a part God on at they have to wrap it in plastic, which, of course, you know, increases costs and it's not great for the environment. So their first product in market right now was actually a checkout as scanner for fruit and vegetables. So a person come up like you normally do when you go to the grocery store. You put your apple down, your bananas, your grapes, your watermelons, whatever it is. It Scans, it identifies it, prices it and a way you go. So that's already replacing, you know, the manual process you might have had to previously do. What you type in the number for the product, and obviously that slows things down. That's just scratching the surface of the potential here. But they are very much rolling out, initially using this hardware product with the software back into the market. There in hundreds of grocery stores around the world, as you'll hear Marcel talk about, and it's proving very, very popular and I suspect, certainly for the short term future, that's where their growth is going to come from. But Marcel is very cognizant of where this is going. He has a vision for the potential of this technology because, as you can imagine, with computer vision you can start identifying anything in the world. So the potential is huge for collecting data, for making things more efficient. He did talk about one of the next things they're ready working on, which is counting the number of human beings waiting in line so they can open up more checkouts as that line gets, you know, past a certain point, to make things, you know, more efficient, better customer service and so on. So again, this scratching the surface. This is going to be amazing where it could take us, but right now it's just about rolling it out to as many grocery stores as possible. He does actually ask a few questions about, you know, what the future might look like and of course it's I didn't think of this, but it's a great, great point. They're probably won't be a physical check out and this probably won't be a hardware solution at some point it will just be software sitting on your phone, and your phone obviously has a camera, so it can ready be used as a scanner to identify objects in the world. You're probably aware of the Amazon ghost stores that exist right now where you can just walk in the store, you take things off the shelves and you put it in your bag and you exit and automatically charges you. That system is a little bit more clunky as they have cameras all over the actual space to see what you're doing. I could imagine a world where you just have an APP, you open it up on your phone and you just scan each item by hand as you put it into your basket in the way you go. In fact, that's apparently happening already with some stores around the world, since obviously there's bar codes. But this is only going to become more effective more efficient as computer vision gets better and to leader is very much at the forefront of that and focusing on the software side, this is also well worth listening to as an interview because I do ask Marcel to explain the origin of this, because it's a hardware challenge with software as well, so they had to create that. You know, a very early Beta prototype version tested out. He talked about some of the mistakes it made in key demonstrations. They have to do in front of executive so that was a bit and even in front of television. Actually had alive crew where it didn't work very well. It's scand an item and didn't have the right identification. So some very cool stories from the start of journey. But it is actually well and truly in market now and you know, it's a functioning product and that's very much why I decide to invest a. They have traction. It's working. Huge chains around the world are using them, grossery stores, but also where this technology is going. I could see it really branch out into some amazing places. So that's what got me excited about that's why I decided to do an angel investment and that's why I'm so grateful that we could have a chat with Marcel to talk about this technology, this company and where computer vision is taking us. Okay, if you have not done so already, please subscribe to this podcast. It's called vested capital. You can grab all the latest interviews as I release them. Just look for vested capital... your phone APP where you listen to podcast. Maybe you have it open right now. Hit the subscribe button, the Plus Button, the follow button. I'd love to have you as a subscriber. And also the one sponsor I always mentioned in this show is my own company, inbox DONECOM. We provide email assistants who will step in and take over replying and managing your email and building a system to take you out of the loop so you don't have to complete processes and treat your email like it to do list each day. So designed to free you up, give you that hour or two or three or four hours that you currently spend in email. You get this person to work with directly who will manage it and reply to your messages for you. Just go to inbox donecom for more INFO about that. Okay, now here is the fantastic interview with Marcel enjoying. Hello, Marsel, thank you for joining me today. Hey, yeah, nice to meet you and thanks for having me really excited to get this going. Me Too, I very excited because I love to her and what you guys. I am not sure I'm saying that right. I actually I did never realize there might be a different way to say that, but I love what your company does. Especially because it's sort of cutting edge technology. So, for everyone who doesn't know what the background story is, do you want to tell us what exactly is your company? What is it do and and what you know? How far along are you with the development of it? Sure, so, till lead us, like our vision is really about, like the mocktizing computer vision. This is where we initially started out, and talk a little bit about the story, like how does it all came about and why we believe it is so important for society. So, you know, computer vision has been really transformed. It for all kinds of industry. So right, like you mentioned me, like, hell, retail manufacturing, right, everything from autonomous cars to even autonomous stores, for like what we're seeing with Amazon, for example, with you know, these Amazon goal stores popping up all over the world, but we realize very early on that they are no commercial product available that allow these kind of computer vision identification things to happen. Or like they don't allow this out of the box. And you know something that you just plug in in a wall and it starts working, and it is essentially what we wanted to do. Right, and if we want back now the time around like four years ago, the cattalyst for me and my cofounders was, you know, a uni. We had these subjects where we identify cancer cells on em skins and using computer vision, and that's the first time we really got into this whole idea of, you know, understanding how machine learning and computer vision could really change and transform society. Come back to that point. What we realize very early on that, unless you have this principle of making computer vision accessible to the world, you need to in order to do that, you need to create easy to use products and in a like these examples are just mentioned, such as autonomous cars and autonomous stores. All of them are not self explanatory. I'd like if you have to die in into this, unless you have like a computer science degree or you know, like you studied engineering, there's literally no way for you right now to get this off the ground, and that's exactly what we wanted to do. Right and all, first commercial use case for this has been fresh produce identification. I think this is what everyone knows us right now. Fresh produce is what we're doing, and it comes back to this whole idea of using computer vision really to change the way people shop right now. So, for example, with. You know, when we start with fresh produce, when you go to into supermarket, the first thing that you see is this huge variety of large items of you know, fresh body use. All of them look very different and and all they price very differently. Right. So it does. It depends if you're in the US or if you are in Europe. All these items, you know, look different, price different and you know, like when, once you start your shopping journey, you put them in your basket or in your trolley and you continue shopping. You know, eighty percent of all the items in supermarkets right now have bark cootes, so it's easy for us to identify those just by using the existing scanning technology that's been around for like over forty years. But where you see problems with are like items that don't have bar codes, such as fresh produce, right. And you know, when you go through the store, first things, as a said, like you're getting fresh produced but then you're getting other items that have bar codes. You know, like in once you continue your whole shopping journey, you go over to the check out and really you don't really as a as a customer. You know, all of us are are used to self service checkouts. Right, like everyone knows how to use them, you know, you just skin the items, you put them over the skin and put them in your back. That's easy. But as soon as you come to the point where you have to do this with items that don't have bar codes, you're and real problems. Right, you're struggling because you don't really know like what kind of apples you selected earlier. Right, you don't know if they were read red delicious, royal gala or anything like that. So you revert to what do you think is is most likely the cheapest. You know, this is at least the problem that we are seeing here a lot. And when people just say,...

...well, I don't know, like the kind of Bravo Apple, you know it's definitely not that apple, because this is eight dollars ninety nine to kill, and then you got like the red delicious, which may be in special, for one dollar ninety nine. So there's this massive discrepancy between those different but similar looking items. And then obviously is a massive problem for retailers. You know, like this whole inventory management is something that they're really struggling with and you know, like you can even take this to an extreme sometimes you go, like to your local courts, or who worse, or and us to your Croagus, some warmut you see, you know, ginger, for example, costing forty nine dollars a pound, or like a Keylo right, and then you have those brown onions or red onions that cost maybe forty nine cents a pound. Right. So there's this massive discrepancy, right, even though they will completely different. Right, but there's no way of like telling this a part right now from the supermarket point of view. Right, like you can't just I mean unless you have someone standing behind every transaction, and anyone is doing which is, you know, not really feasible. Then this is really a personal thing. You know. That hits home every time plastic, right, like what we're seeing a lot of fresh produces wrapped in plastic. And why is that? Right? Yes, it's to extend shelf life to a certain extent. But what we're really seeing problems with is when retail is doing this really just to put back code on it, and that reason for that bar code is the same way of like eighty percent of all items have ba cold. It's really to speed that up. And you know, we think this is unfeasible. You know, like we shouldn't be suffering just because, you know, like it is this speed and also experience factor for the customer. So this is, you know, like the essentially that journey that just took you on. That was four years ago the first time. You know, like me, my personally, I'm from Germany, you know, if you don't really have self service checkouts there, but you in Australia you got probably sixty seventy percent of all checkouts our self service. And you know, like figuring all of this out like on the goal was such a nightmare. Recite like yes, let's you know, let's put something in you. If you think about computer vision transforming the world, may as well use it for a product that anyone will be using at any given time. Right, and you know, like we developed essentially a machine learning computer vision algorithms and we can dive in later into the technology and how it works and wide is so different to the normal approaches that we see out in the field. But you know, that was for us, all right, let's put this in it so self explaining right, like it's no brain now for anyone. Right, it's maybe not the thing when you are someone, Hey, what, how could you change my shopping journey. Majority of people wouldn't say just like identify fresh produms, but majority of people wouldn't even know. They're like yeah, it's annoying, but it's not like, you know, hey, this is like my number one pain point. But you know, it really speeds up the process and like, from the customer experience point of view it is great. Yeah, I have to admit when I first heard about you guys, it was it was Brendon Hill actually, and when he was one of your investors and I had an opportunity to be part of his syndicate lead and make a small investment in you guys, it got me excited because I thought, okay, like you just describe, it's a very obvious efficiency problem, simple. It's a great first go to market solution to a common problem that I can see you guys and you are getting traction with. But then my mind went, you know, where does this go? Where does identifying object with computers a lead to? And obviously I thought about cars or you know, autonos. Driving now there is a sense of identifying what's around the car using computer vision to figure out but I feel like you, you in particular, Marcelle as a CEO, you must be thinking of the vision of this company. You know, where's computer vision going to take you? Without maybe giving way all your secrets, I would love to know what what what is? I mean? I feel like you, you don't feel limited by just going after identifying fruit and veggies. You see this becoming a much bigger potential market. So where do you see you going? Yeah, on the person get you. I mean, you know, like the first four years of our journey. Now people refer to us as the fresh veggie guys, you know, like rather than actually technology company that we are, which is fine. I think I'd like to like, you know, like a page out of Peter tils play book, where you saying basically, you know, go after niche market, dominate that niche market right, and then expand, expand from there, right, and that's exactly like all playbook rights. Like we want to dominate first produce. We believe this is a winner takes at all market, but it's just the beginning. And do reason why we pick fresh produce? I think maybe, in hindsight, wouldn't have picked it, just because, you know, like the other struggles that you're facing along the way. But you know, like now we really embedded. Took a long time to get us off the ground, but the vision is, as I said, democktizing computer vision. You know, I it's going to take forty, fifty years to get there. And you know, our mission statement is to accelerate computer vision and then, you know, like by creating easy to use products. So essentially, what we did with fresh bodies was product...

...identification, very specific product identification, right, but this is not necessarily where I see you know, like I think that two ways of of going about this is either you going really deep into certain areas, in niche areas, and you can do this any time. Right, you can do this with fresh bodies. The next one, maybe bakery items, the next one. Maybe, you know, like even going out of this maybe you know like in a manufacturing identifying hardheads or you know like these kind of things, and you can do this, or you can go, you know, broad, and that's essentially where I see tell leader going, really facilitating computer vision rather than, you know, doing all the heavy lifting every time ourselves. Say, for example, with fresh bodies, you know, the heavy lifting was four years of you know, Patterner and and unique algorithms that we develop for that particular use case, which I would say so far set us apart from the competition. But this is not the best use of demoketizing computer vision, because if you're doing this for each one of those, you know, it's going to take you a long, long time. It's going to probably take us four hundred years to do that rather than forty years. So, you know, like and for examples, this fresh produce has been not all first product, but you know, we already have like a second product where we are essentially identifying people queuing up in stores and then feeding that back to the cashiers to open more checkouts. You know. So this is like another you know, like where this is going, where thislike development of, you know, identifying people in the store, because you can use existing infrastructure and open source software on certain things. You know, is much easier than diving deep. And for us as a company, I really see us, you know, like going they broad offering this, facilitating this, you know, by a platforms or really by a broad products where people can just plug them in and potentially even after chance, like to develop their own algorithms. Okay, interesting, so very much in the I guess retail space at the moment is where you guys are playing in, which which makes a lot of sense. Would you mind US updating us as we record this? where? How to say this right, it's to later, not till liter to lead. How far along are you, like what like level of presence do you currently have in the marketplace? Yeah, so we're life right now, I think you know. But fifty eighth sixty stores over hundreds, hundreds of systems deployed, and this really all across the globe. And this comes back to this winner takes at all market we started in Europe and in Australia. Australia is that perfect testing ground. It really is hard to correct, though, because you got basically three retailers controlling Australia, like you know, with sixty sixty five percent market share. So where you know the Europe and us, you know, like you've got those big players, such as Walmut Kroger, but small is relative in terms of Australia. But you got those like small champions right that I have like three hundred one stores as a footprints, you know, tens of billions of dollars in revenue. But it's very easy once you have convinced. Not Very easy. But once you convince like those large, you know, supermarket chains, and they become advocates, this is then when you like have to expand, like out of Australia and and that's what we did right. So, like we have got our presence now in Australia, in the US, and were going into South America, Central America Europe as well, and you will see this these massive role or it's happening now, like over the next six to twelve months, where we really see our technology and we'll see our technology in hundreds and thousands of stores. But you know, like it took that long, you know, like to really get that confidence. Yeah, I can imagine. I'd love to know the early days. Actually, before I ask you about the the starting point. I'm curious what the what's the business model with this, and I'm assuming it will evolve over time. But you know, do you do you rent a unit, like is it like a box that you basically put inside a grocery store and they pay a rental? Feel like, how does it work? Yeah, it's a bit more complicated than this, as we have quite a large product offering, but this is, you know, like if you. If you look at supermarkets, not many supermarkets look the same as as they like a Colt so worse, they're all look very they have like small slide changes, which makes it hardest to scale. But nevertheless, like we found, you know, like our five six product variations and that really scale globally. But essentially, you know, like this that two things. One is a hardware box, which is essentially I can etch device, you know, like which does all the predictions on the device itself rather than sending it to the cloud and back. This is, you know, like you retell us just pay for this on off fee rights, like the pay that hot with fee and then yearly software subscriptional license. That's the business model. And then we've got also like fully contained systems. They're basically like they function as checkouts, and those systems, you know, like they have a scale system in there. They got a poll the point of cell screen, but also, like our underlying technology power in the whole thing. And this is the same business models. So you pay for the system and then for the early subscriptionalizenes. Okay,... it. So it's sort of like a stand alone scanner or a completely check out. Do you might have to go back in time, Marcel, and I think about this and I can imagine entrepreneurs listening to this. Building something physical and technically advanced is daunting to begin with. And it sounds like your background is more medical, if that's right. You were studying, you more in that space, a biomedical maybe early on. So take us back to when you first have this idea and you decided to actually go for it. Did you think, okay, I'm got to raise ten million dollars in enture capital just to build a prototype and then we'll go from there? Like how do you start something as big as this? Yeah, I mean my background, my initial backgrounds actually mechanical engineering so on. This is where the whole like the designing and you know, like of boxes and and you know, like Michael, controlling, really comes from. I mean initially, you know, like if I take it back now, like say four years yeah, four and a half years, we didn't really know anything about like raising funding or you know, like how much do you actually need it? Essentially, when you start out, I mean this is what you start out. I'd like. You have no idea, you don't know if you need like ten million dollars. You think about like hey, maybe I need one hundredzero dollars, right, and I'm going to scale this up all around the world. This is your that a naive right, like when you first stay at out and this is what we did, you know, like raist a little bit of money from, you know, really just friends and family, right. We didn't know anyone and was super hard like to that was again four years ago, I would say. Like the startup seemed look very different here in Sydney, very different, right, like even like angel investors. You know, as soon as you mentioned the word hard where, everyone was just running for the hills. You know, no one, no one really wanted to invest anything like hard. We related. Then you have that next point where you say it's not just hard where, it's also going to be in retail, and everyone's like retail, oh no, why would we ever invest in retail? You know, like if you go to supermarket, nothing has changed in the last ten years. Right. Some of them even say, you know, certain retails are still using the same technology if they use forty years ago, but it is time. It's the same with every industry at one stage, you know, like the need to overhaul their systems in order you know, especially, I think covid showed this really where you see in all of the sudden, retailers need to step up the game if they don't want to lose two Amazon or like even to you know, like the gorillas of the world, you know, do these online deliveries. If they don't want to lose, they have to, you know, step up the game. I mean we never thought about even raising venture capital. We thought, you know, like hey, maybe, you know, we just need a little bit of money here and there, and then scaled this up globally. But, as you mentioned, you know, like hardware, there is reason why hardware is called hardware. It's hard and it's not as easy to scale as software. And Yeah, like, I mean along the way we learned all these lessons and we're like, all right, I think we need some venture capital here, right. But again, you know, like coming back to hardware, it is this competitive mode, you know, that that you have really well, where people look at you and you're like shit, they're bringing their heart resolution, it's their own you know, software system on their their own os and and it's very hard to replace. And then you go, like to retell us that have those long sales cycles, you know, like of one two years until they're buy something, but then they don't change it for the next five to ten years. Right. So once you're in, it's very hard to get out, and you see this with all these legacy systems everywhere in supermarkets all around the world. But yeah, venture capital was until I would say, like two years ago, a venture capital wasn't really on the cards as much. Okay, well, you have to have my question. The other half is, what did you build first, and how much did it end up costing to know, the create something? I mean, we all when first started to lead up, we all still used to work our day day jobs, you know, obviously not as efficient anymore as we used to, but this is how we created it, I like, really with our heart, earned money everything that we had, like Upwood, you know, everyone put all the savings in it, and then, you know, we got money from from our parents and also like the first, you know, friends that believed in what we could be doing. And this is how we build it, you know, like from a hot aspect. And obviously, you know, like we need more money. The more people we hire, the more problems. Is the thing. If you hire more people, you're also like causing more problems because it will have it. So we need to raise more and more money, and we did this like in a rolling session rather than, you know, saying, all right, we're raising our five million dollars, please all invest me. You know, raise a lot of like safe agreements, via safe agreements, you know, from institutional investors such as universities, but not really from venture capital until last year, in October, is them on the first time we really raised, you know, five, five and a half million dollars, and as part of a venture around. But before everything was rolling. So we always just raised money when we needed it, you know, and the frequency this fact...

...became like, you know, it became closer and closer. It's go back to the the prototype. Like was it regardless of how much it cost? I'm imagining it's you being a hardware guy. You have that skill set. Were you literally in a garage putting together components and then marrying that with software? Like how did the how did you physically create this? And in the first place? Yeah, so I would say that the first prototype basically looked like an Ikea lamp and we had a Webcam mounted on top of it, you know, like and kits, you not? Like I was collecting data for three, four weeks, you know, like we bought so many food and vegetables. We made upple pie for I think four weeks straight, you know. And this is how we collected the Datta. This is how we had all initial product type. Then the next one. I actually welded this out of steel. I don't think we have that prototype anymore, but I think the whole weight of the prototype was like twenty five kilods. Like kids. It was like made out of steel, like all physical steel, and we had like a haver we component that touched it. But really just like anything to get by, right. We didn't have fancy D printing technology or anything like that, so we just try to somehow make it look more commercial, which and then we invested in the like d printing technology, you know, like to really I mean this is the thing, right. It's like with start up costs. Why they're saying this new generation doesn't need as much up front money as they used to write, because you've got three D printing technology available, you got software packages, you know, like back then we were able to use, you know, like open source software, and you know that that's that's how we survived back then, you know, like without that, you know, initial capital outlet of like want two million dollars. Okay, so, just to I understand it in my head, you could get a Webcam, a light, so some open source software that does the basically the interpretation of what the Webcam is seeing and attempts to identify the object, and then you try and turn that into something that is commercially viable, like it looks like something people will be willing to put into their grocery store. The software functions without bugs. You know, you don't have to go in there and fix it, because when you're rolling it out to a woolworth or coals or kroger or whatever, and I'll the you, I can imagine the first time, you're probably sitting there watching it carefully. Please don't break, please don't break. But you know, to be in as many stores as you guys are in now, you have to trust that this thing is this going to be working, day after day, minute after a minute, you know, scanning, translating, everything works. So how long did it take to make it to the point where you felt it was commercially viable? Has So many great stories just just do you know, like thinking about that timeline. You know, it took US really two years to get I would say like one the half years, probably like to get it to a scalable point. So the initial problems that many companies are facing with computer vision is as soon as the lighting now use changes or like the environment changes, the whole system just predicts nonsense, right. And we had to move in to all first store. I remember also collecting data for a couple of weeks and, you know, like then be at channel seven, which is this TV show, you know, like in dead I think it was sixty minutes some I can't really remember that the program what it was called, but they're rocked up with, you know, like a whole film crew and everyone was expecting the system to work. But for some reason, you know, like in this is like what you discover along the way, for some reason our system just wouldn't perform at certain times, you know, and I was just like there was too much sun in the store and all data was collected at nighttime. These are the initial struggles that we had. So we go there and, you know, like the whole film crew, everything is going well and I'm just like I'm literally, like I'm sweating, right it's like I put down apple, comes up as a watermelon, these kind of things, right, and but but it was good, right like. It's this initial learning. All right, how honorable are these computer vision algorithms, and what do we need to do in order to overcome those initial problems that we faced? And that's that's what we did, right like, and we slowly developed it to that, like that software that you see now, and we worthe you know, like now, like I'm not very you know, like you can go in there right now and I know eye. Ninety five percent of the Times that system will be right. I know that, right and you know, we're doing hundreds of thousands of transactions every month and you know this. This is good now. But, like, I would say, like two and a half, three years ago, we still had like this initial problems. And this is just computer vision, right like, and it has develoved a lot, you know, like these people net algorithms, for example, that you see that identify people. Those things were really shaky a couple of years ago, and now, like they have finally arrived at the point where you can deploy them without without too many false predictions. Right. Yeah, and you you tricking a thought when you were talking about the counting of people...

...with with computer vision, and then you're moving in that direction already. I understand the reason why. You just for simple you know, we need more staff to man the check out or just simple as that, or maybe even calculating retail space and how to distribute food and, you know, place products around the store. Where does this become a privacy problem? Because I know you probably had people talk about that and obviously in countries like China they're scanning everybody faces, you know, they use their faces to log into things without even needing another human being. There countries like Australia, Canada, America, it's a little bit more cautious on the side of privacy. So if you bumped into any challenges with this idea that what you're doing might be potential privacy she people don't want to be counted on to be scanned. As there been anything along those lines? Yeah, I mean this GDP are compliance, you know, like we take this very seriously and it's coming back to this whole idea of having edge devices. Edge devices right now, you know, they don't send that data to the cloud. Don't bet your streams. It's really just sending Metadata, right. So what what the people at for example, you know, like one of those retailers, will see is essentially all right, now, like two people are lining up, but they don't have they can't see who is lining up. All what they see is, all right, we had full capacity. Team. Now please open another check out on another counter. Or even here's the real deal about like these people counting and queue detection. It is really it gives an access seven off like what kind of changes, you know, like in their stall out, for example, what kind of effect they will have? Right, and they can track this every day, right. So like they're changing one thing now, like a week later, they can assess all the data and it's like well, we use moved one check out over over here, and all of the sudden, you know, we were able to reduce our checkout time by ten seconds perpose, which is massive. Right. And if you think about this, this is normally what consulting films are getting paid for in all like what a KPMG OR PWC? What they're getting paid for, you know, like sending people in there and making sure that those statistics come back to the retailers. Now, essentially, what computer vision allows you to do is giving an access to this kind of data twenty seven, right, without breaching any of these privacy, you know, problems that you just mentioned, because I believe it is a real problem, right, like if I was able to identify every time you go into store, right, I mean, yes, they that's the other side. There's the flip side of the of the coin where you could say, well, I could now target this particular person with advertising, you know, like with dietary requirements and all of that. Right, but we all know, Australian and like the Western World in general, right, it's not really seen as an advantage the customer, rathers disadvantage. I funnily enough, you know, like if everyone is okay, you know, for private, for facebook to collect all your information, right, we like for zero when you when you talking to it, right, like everyone is fine with that, but as soon as it comes to all right, now, like you know, we may target you when you go to a supermarket to, you know, make your way off to let the steals. But I get like we have to be compliant with, you know, whatever kind of regulations we come under, and I still believe you know, it is great the way we're doing it right now, but I see like a massive, you know change probably in society in the next couple of years where people are okay with that right if I mean, you know, if I go in the store, I don't drink normal milk, I just drink almond milk. Ride it'd be good, it'd be good to show me, okay, where is that almond milk right and right now, like you, doing this through APPs, but I believe computer vision might be, you know, like the better solution that in the future. Yeah, I can imagine like all the science fiction movies I've seen where an add pops up or your shopping and it's, you know, a Hologram telling you something. But, like you said, it could be a search based tool as well if you're wondering where this item is in the store and they already have data about you. You want milk, when you say milk, it's almond milk, it's in a'Le Four, etc. Interesting. Would you might taking his back, though, to the point where you move past that early stage. The apple is no longer scanned as a watermelon. It's actually working you. You're comfortable with it, rolling out to many, many stores. How do you then expand in scale, given this is up? Like you said, it's a hardware solution, it's a physical product, even though I think it's really a software problem. Ultimately, you still have this physical device you have to place in stores and you've expanded globally from Sydney, which is obviously far away from a lot of these places in Europe. How did you how did you go from that working first version to to convincing these big chains to actually give you a chance and, you know, prove it works for them? Yeah, one thing to understand. I believe around two thousand and eleven, the first fresh body is identification systems popped up around the world and you know, like back then, I believe IBM was involved in this. That then got sold to Toshiba. But...

...what happened was back then, obviously you didn't have that computational power. You know that you may have two algorithms, but you didn't have the computation of power in order to process for those algorithms. Like giving you right and essentially, a lot of retailers put a lot of money into this kind of technology back in the two thousand eleven. Right was when the end up being happening was that the majority of retails got burnt by it, and burned in a bad way. That, you know, like that had all these capax cost of machine algorithms, on machine learning algorithms that, you know, didn't have any relevance for the retailer. They just couldn't use it right. It would never work. Those initial problems I talked about earlier. That's exactly what they were facing, right, and all of the sudden was to call pm and everything came up, as you know, as a stake these kind of things, and for us, initially, what we did is we build up a team in Germany, and that was super important, you know, like to service our biggest clients, you know, and essentially, how you commenced those big clients, and comes back to this whole thing about the principle of, you know, having easy to use products. Essentially, what we did is, you know, like we took our systems and took this to the headquarter, right, and invited hundreds of executives, right, and maybe ten fifteen ended upcoming right to the store and we denod it in front of them, right. We gave them basically opportunity to demo everything to to their own folks, but anyone was super interested that if this works, yes, we will put it in everywhere, right. And but yeah, obviously a lot of skeptism. And you know, once they were able to show it to their superiors, to their executives, all of the sudden discovered really momentum, right, and that's that's how we did it in Europe and you in Australia, we actually use the momentum that we created in Europe to bring it back to Australia. So the first doors were in Europe and in Australia, right. But then obviously Australia is this small puppy syndrome, right, like you don't want to miss out when someone, especially in retails, you know innovative or you know like shining these new things. So like you want to do that here as well. And we ended up, you know, like having really good products. And then like the US. So we always knew in order to crack the US you need people that have a lot of existing relationships. You need basically gray hair kind of sales people, kind of people, they that done this already, you know, like maybe was twenty thirty years ago with something unique, you know, for us. In our case, it was someone that, you know, was one of the CO founders of a scale management software company. So he had done that, you know, twenty thirty years ago, sold to all these massive retailers out of their garage. Right. So essentially I messaged him, you know, like we had some very good talks and I was like, you know what you did that to twenty thirty years ago? Can you do it again? Can you do it for you know, what we are offering? He said, all right, I'm willing to give it a shot. And this is really now, like, I mean real life. If you know Albertson's, you know it's like one of the largest retailers in the US. There are couple of other ones that are unfortunately can't mention right now. That you will see where those systems will pop up in the next probably two three months. And you know, like you don't get access to these kind of customers if you don't have someone on the ground that has done these kind of things before. You know, I mean essentially, some of those like a multimillion dollar deals, right, you can't just like walk into the executive rooms like he please buy this, you know, and please buy it for fifty stots. It's not it's not going to work unless you have those relationships when you come into Canada and can I actually experience one of these myself. We're trying. We're trying, hopefully soon, hopefully by the end of this year, like we would. Got Some, you know, like you got the soulbees the lot last we got some. Some of them already interested and it's just for us now pushing them over the line to try these kind of technology because of this. It is mind blowing when it's working. You know, I'm watching. You left Australia now, like otherwise you would have the best experience. He Sue. So take us far with the future then. I know you already kind of mapped out some of the ideas with where computer vision might go sort of over forty years, but what with tolter in the next sort of five to ten years? Is it just a case of, like you said, it's like a winner take all kind of scenario? Is it just a case of trying to get into his many grocery stores around the world and just become the de facto tool for scanning whatever needs scanning in a retailer grocery retailer. And also what's what is the challenge? There's it's simply the relationships and the slow sort of convincing sale cycle you have to go through. Or is it really about building a team and every single country, and that also takes a lot of time. Yeah, there's a lot to unpacky. It is. Yeah, first of all, you need a team and you just need that team, you...

...know, like to really give those retell us the confidence in you, right. I mean they heard it all before, you know, software companies that tried to scale like form, if it's Israel, Australia or any other country. It is tough, especially for us. It's really crucial to keep those relationships, but also build up on these relationships that we're having with those retailers to, you know, get more of these new, next generation products. Exactly what I'm seeing right now is, you know, like we're doing fresh produce recognition on checkouts, which is great, right, but I also believe that the next horizon there might not be any more checkouts. You know, in ten years time there might not be any, right, or you will see them maybe similar to like petrol cars in twenty years. And I like everything's going to be electric and it's probably like how you see checkouts as well. So obviously we don't want to put all our money, all our innovation into something that might become redundant. So what you will see a lot from us is we will be going into mobile, and we already, you know, doing a couple of things that with mobile, where you know, you just use your phone in order to identify, you know, what kind of item it is, pairing this then with the weight that is measured somewhere as form a legislation. One of you how you do this. That there's another story. But I believe, you know, like us, we will go like into mobile, we will go into more areas where we don't necessarily need integration. Integration is always key, right. Integrating into a retailer system takes a long time. Yes, it gives you that competitive mode which just because you know others have to do the same thing again, right, which timeline sometimes can vary between six twelve months. But I believe the future will be not just an autonomous does the wants that you see them as on go, but it will also see like a hybrid version where people use them mobile phones or potentially even you know, their next gadgets. You know, right now I think everyone's got like these smart watches. You know, maybe it's smart glasses, smart contact lenses, whatever it's, it's going to be right. But using this and really that transformed this whole company more and more into a software company right now. You know, like I would say like from our revenue perspective, eighty percent of this goes back to hardware, twenty percent to software, right, but this will change. Now the next couple of years will we make more and more money from not just recurring revenue but also offering software packages without hardware involvement. You know, I I'm from a kappex point of view. That's that's essentially where we see this whole industry is going. You will see a lot from us that will support mobile phones, and not just mobile phones, also handheld scanners and these sorts. Right correctly, if I'm wrong there, if I'm reading what you're saying, it would be like I would go to the grocery store, I put everything I want to buy into my my bag, I bring it to an area. Whether I could then or I could even do this as I'm shopping. I guess I have my phone out. I take an item off the shelf, I just use the camera look at it. It adds it to an APP, which is the APP for that grocery store chain. I just put a straight in my bag. It's added to my list of things I'm buying in the APP. I just do that as I'm shopping and then I just click by kind of like I would do it with, you know, online shopping. Set them doing it in the real world. And this like the technology. You will need bar codes because the computer vision would just know this is a product. It could just read whatever the boxes and even the writing on the box and it would identify any kind of object or packaging whatsoever. Is that kind of what you're painting as a picture of the potential future? So this is already happening right now. It's happening with bar codes. So, you know, like the store set I was mentioning, say, for example, with who wolves, you know, like we're think life and you know thirty thirty thirty five stores where you do this exact thing that you just mentioned. You know, like you take a photo of the of the bar code, it skins. It already doesn't take a photo of the item and recognize the item, but just does the bar code. And now, like the part it doesn't work is the fresh produce. Right. This is why we have those whole checkout systems where you place the item down and then it just pops up with a bar code as well. Right. And but I'm seeing, you know, going forward you will see less and less of like these checkout systems and you will see more and more of mobile recognition, you know, like where you potentially don't even eat those checkout systems, you know, like because you do everything, as you just said, on the APP. Okay, yeah, that makes sense. Like we're getting used to doing it with qr codes, to look at the menu with with Covid, and it's like everyone sort of expects the other phone to be the interface for everything, or the watch or whatever or, like you said, a portable hand device that you could also use if you don't have a phone. In terms of growth of the company, I'm really asking this to as a, you know, a small minority time little investor, and what you're doing. What is the timeline for all of this? Like, I can see what you know, as long as you remain excited Marcel individually, you personally you you're going to be here for a long time growing this company. But where do you like? How...

...does a company like yours grow over the next two, four, hundred and ten years? Because it is hardware, so it is a bit slower. It is a slower sales cycle. Like you said, this may be a transition period to as you move away from hardware to more phone and APP based solutions. Obviously you need to keep raising funds to keep building the team. Every time you hire and new developer, you said you know you need to have more money to pay them. And I ask this because I'm a software web kind of guy and I kind of understand how that works. But in the hardware world I feel like you have to be really careful with your cap x, your expenditure. You know you're raising funds as you go along. You can't just bootstrap. It's very difficult to use the money you're making to keep growing. So how do you take the next sort of three hundred, forty, five, ten years in scaling and manager everything? Manage everything. You're the CEO, so you I'm sure you're thinking about this every single day. Yeah, it's a good question, you know. For us right now we basically bootstrapped, I would say, you know, bootstrap in a sense, with a little bit of captain until last year, middle of last year, and now we really have taken some fuel from venture capitalists in order to really go after those, you know, large market. It's rights, like we always said that if you just want to do fresh bodies, we probably wouldn't have to raise any capital. Right. It's just like you become capital efficient, you know, you got your subscription fees, you're selling the hardware with a margin. You also this is all working well, but in order to take on really the world and demoketized computer vision, you will need a lot of funding. And the reason why you need a lot of funding is to run experiments. So I'm a huge friend. I'm a product guy, right, like I love like running experiments and just finding out, you know, like after very short time, like is this working? Voice is not working? Right. But you need that internal conviction as well. I think this is super important. Right, we are internally, we were convicted that computer vision is going to be the future. Right, like if you look at fourth industrial revolution, computer vision is a massive part of this. Right, like in will transform society. We know that, right. So compleatualize. Nothing is missing there, just like how you get there. In order to run those experiments, we will need access to money, right, if that means, you know, like more venture capitalist or, you know, at once stage, taking the cup company public and then really, you know, I love that from Jeff be so's when he took Amazon public and in he said that, yeah, the stock market for some reason, like all our internal numbers went up, right, but the stock market, you know, like all our numbers went down right, but we knew, like inside Amazon, you know, like we had something really great here. And you know, I believe you know, like if you go public or if we become a stay private for, you know, a couple of more years. Nevertheless, like the future of the company, I think, is secured. Were just, yeah, really need to execute along the way, doing more of those experiments. And and this is, you know, like the company's Day and a right. We are product focus. We are very product driven company. In order to figure out like the right the right things in the right products along the way, you just need capital. Right now we are not, you know, like in order to keep growing, you know, like the fifty five people now, right, but I'll hope like in the years time, you know, like that we will double or triple that. And just because there's so much opportunity out there rights computer vision. If you think about this, everything that right now that we are seeing, you know, some of these can be taken care of, but machines and so much better, so much more convenient and, you know, more efficient for US humans. I love it. I love the Vision, I love the future focus here in the product focus. To maybe like kind of wrapping up here a little bit, Marcel, I'm curious, you know, being a product guy, you must feel like you want to have your hands involved with every experiment. But what really is a day in the life of Marcel as the CEO of Til Eiter? Are you still playing with product or are you more about hiring and firing people with your daytoday kind of role or building relationships? What? What do you do every day? Yeah, I'm still like massively involved in product. You know, I will never give that up. Think build gates. Like even like until the last day, he wasn't every product review meeting the company ever had. So I see myself the same way there. I do a lot of things in terms of hiring. Hiring is really what I'm focusing on now and it's about not the people that you're hiring two morow, but it's the people that you will be hiring in three months, six months, twelve months. Right. So we have a great network from people, you know, like through firms such as angels, you know, like where you have a lot of individual investors and everyone is willing to help. You know, I think this is what I love about angel is, but even like eleanor ventures of life, Fox means she's here from from Australia and they have a lot of people, all of them are willing to help and you know, potentially some of them could become your next hire, your next vp of engineering, your next vp of sales. Right. So that's that's a lot of what I'm doing. And then, obviously, you know, keeping the lights on, speaking to venture capitalist. We are in that process of, you know, just renting up for a series be so you know, I'm spending a lot of time they are and building relationships. You know. Obviously, you know, still the steward of the of the culture, the company culture. I don't see that changing much, you know.

And then, obviously, you know, thinking about strategy, really not thinking about the short term strategy. That like the next year, I think the company Really Scot that, but the next three, five, ten years, you know, like I think Elon musters, I grade at this, you know, like pointing out, like how the world would look like in five hundred years time and slowly taking it back right, like how does it look like in one hundred years, fifty years, ten years? What do we have to do in order to move to needle? And, you know, like I'm spending you know, like I'm spending moral life, essentially, you know, like looking for the right kind of strategies might like. And none of them will be perfect. You know, there will be a ajustments along the way, but I'm fine with that. Now. I love it. I love it exciting, citing industry work and being a product. I must be so satisfying to see these things roll out and perform and and then you get excited about the next generation and next experiment. Any kind of websites we can send people to, and you know that you're hiring twos or any specific job opening you you want to highlight before we end the call, a data. So we're really looking for ahead of data. Head of AI. I mean you can't have too many machine running engineers and, you know, the over always need to be super careful that they're not getting snatched up by, you know, like those big corporates, that Fang companies. They just have, you know, obviously so much money at their disposal. And but you know, from what we're seeing, we get a lot of great talent and the reason why we're getting that talent is because, you know, it is exciting. It's exciting what we're working on right. I mean we're talked about fresh produce identification. Of everyone is using this every single day they go into store or like even if you use online delivery ride like those online delivery people, you know, they're still using the same kind of technology, but they still need to check out, they still need to use our systems right. So, end to end, write like if it goes more into direction of online or offline, a bricking water, you know. anyways, like the technology will always be there. Yeah, I mean for us, you know, police supply, we've got tons of open rolls, but we would like to hire four. But yeah, engineers, software, hardware and definitely data and machine learning. Is that on that till latercom or do you have an external website you use. Yeah, so it's on to lettercom and that to web pages. Wants to like to leadcom, which is really about this whole vision of the mocktizing compry division. And then we've got to lead a retailcom which is, you know, the tailor to our retail customers. But yet just head on to Yourcom and have a look at jobs. Okay, well, I cut all the links in the show notes. Yeah, I really agree with this idea of, you know, working for a company like yours versus say, yeah, joining a Fang, but you might be just trying to make advertising improve their quick through rates and get people to pay more attention to an ad, which is, you know, it's a job, but it's not exactly the sense of fulfillment, improving the world and taking things to a better place. I think working for a company like like yourself at this stage will be way more exciting. Are those jobs all like remote nowadays? Like I know you guys, Australia, Germany are big parts of where you're at. It is it? Is it really a global remote workforce now? I mean it's not totally remote, because you have to have that element, and you see this. You know, apple in orders a lot of the employees back to the stores just because that what we're seeing. A lot of this is, you know, like from a creativity level, it is sometimes really good to have those white board sessions, you know, like to have have it involved to see, you know, what is possible. But we're getting there right now. I mean when Sydney, when the lockdown. So we can't go on your office anytime soon. You know, it's probably going to be another two months or something like that. You really want to grow, that doesn't necessarily mean, you know, like it can't be outside of our headquarters. You know, I would love to hire more people, especially like in the US, you know, where we see a lot of this. But again, it's probably like a base, you know, like we've got a base in Rochester in the US, in New York. But you know, like the day's potentially know that we will be hiring like people on the West course or on the east course. Definitely. Okay, awesome, all right, myself will wrap it up. Is there are an email address or twitter linkedin account? What do you prefer to give out for you personally, if people did want to get in touch for any reason? Yeah, I mean I can just give my personal address. I've got to, you know, like if I get too many mails on on bonds. So what is my cel dot hurts at till lead Outcom don't a you. So you know, like feel free to just send me message if there's anything I could do of you looking for a job or, you know, potentially even like being a new investor. Always appreciate it. Awesome. Okay. Well, I appreciate the time. I man, I'd love to do this in five years time and see where everything is that. I think that would be an amazing transition and what a story to share. But yeah, keep up the good work and I love I love hearing about it. So no doubt I'll hear through Brendan as well as things grow, as he updates the syndicates. I'm looking forward to that. And Yeah, great to talk to you. Yeah, thank you so much and thanks for having me there. You have it. That very fun interview with Marcel, the cofounder and CEO of the let a company. I've done an angel investment, so I'm hoping will become... Unicorn, as we all hope as angel investors that the companies we invest in will go big and ready has gone big in terms of a global footprint and all the stores are entering and all the partnerships they're opening up. So signs are good. You never really know. I think you can tell with Marcel. This guy really is a limit with where this technology can take us and I'm excited to see what new products and what new solutions come out of computer vision. It's an exciting world to be alive in. Right now I'm going to wrap up the show. I hope you enjoyed bested capital. This was episode number thirteen. If there is a friend or a family member or a colleague or maybe someone even a cofounder of your own who should hear this interview. Maybe you're doing a hardware software entry product to a new market, maybe using new technology, it's AI, computer vision, cryptocurrency, whatever it might be, but you want to hear a story from someone who's basically started with an idea, turn it into a prototype and now rolling it out two stores around the world. This might inspire you and or them, so share with them. Best of capital, episode number thirteen. You can tell them to go to my website. Why aarro DOT blog. Go to the PODCAST TAB, but also you can find us on spotify, Google and apple, podcast players, on audible and Amazon, where the podcast sections are pretty much anywhere. If you just type in vested capital or Yarrow, why a Ro Oh, you will find this show and then episode thirteen for the interview with Marcel hearst. Okay, that's it from me. My name is yarrow and I'll speak to you on the very next episode. Thanks for listening. Bye. By.

In-Stream Audio Search


Search across all episodes within this podcast

Episodes (86)