Founder Focus

And we are live. Hello, everyone. Welcome to another episode of Founder Focus. I'm your host, Jade Wang, and I run the startup program here at Cloudflare. And today we are joined by our guest, Igor Marinelli, one of the founders of Traction. Welcome to the show, Igor. Thank you. Thank you so much, Jade. Thank you, people from Cloudflare. What is that in your background, Jade? Oh, this is the lava lamp wall. Okay. It's an entropy generator. It is one of many forms of entropy that adds to the sort of entropy soup that is part of... It ain't real though, right? Or you took this picture or you just found it on YouTube? This is a picture. This is a virtual background. I'm not actually currently in the Cloudflare office. This is from the San Francisco office and I'm actually currently in Austin. Oh, okay, but it's from the Cloudflare office. That's nice. Yeah, we have a lot of lava lamps and they generate a lot of good entropy. So very briefly, can you tell us what Traction is? Of course. Traction is basically a software hire startup focused in helping industries and machine breakdowns. So we basically were able to predict very fast when a machine is going to break and which type of problem that is and more like a prescriptive advice. So whether it's misalignment, looseness, we have a lot of technical information to share, but I don't think it's the goal. So it's just like online monitoring of anything that's rotational. So we don't work only with industries. We work as well with hospitals, airports, facilities fields, because even DHL, the logistics, the conveyor belts, because well, if you're buying on Aliexpress or Alibaba, like those things, they need to come fast. And if one of those conveyor belts stops, you're going to be pissed off. So we help companies monitor that as well. And we say, hey, it's about to break, go there and fix that. So tell me a little bit about how it works in the user experience. You have sensors. Do you have one on you? Do you glue it to the machine? Oh, yeah, of course. Right here, for example. Cool. This is one of the sensors that we have. And you attach that to the machine somehow, right? Exactly. And it listens for vibrations. So it's like Shazam in that it listens and says, that sounds funny. Based on my machine's model, that sounds like a fan is misaligned or something like that, right? Is that how that works? Exactly. We're inspired a lot by Shazam algorithms and their infrastructure. It's really amazing what they've built. So basically, you attach this to machines. There are some magnets that help the fixation. You don't need any more than that to attach. It's really instant. And it starts sending out the data through a mobile carrier. So we don't depend on any infrastructure or Wi-Fi infrastructure. Generally, it's a pain in the ass to try to integrate with infrastructure, industrial Wi-Fi. It's always a problem. So we bypass that. And exactly, it collects the spectrum of vibration, which pretty much, it's very interconnected to sound as well. And the form of analysis, it's very similar to say, hey, when you're listening to a song at your home and like Beyonce is playing or Rihanna is playing when you use the Shazam app, it's so interesting because you can be at your home with subwoofer or just like a radio, very different in terms of bass and treble and so on. Or go to just a show and someone's screaming in the back and still Shazam is going to say, it's all the single ladies. So it's incredible. It's because of some transformations that you do in the way to the frequency domain. And we do the same. And we're able to provide very, very assertive techniques of, hey, this is weird. Like, it's playing X, Y, and Z. You need to fix. And the reason it's very different is that it's very, very different from any type of machine learning models that you're going to build on more generalistic way. Like if you open Medium nowadays, there are a bunch of articles teaching you how to build generalistic machine learning models, which is like, we know something is wrong. We just don't know what, you know? You get just this chunk of historical data, feed the machine and like, yeah, those like data points here are weird and you exceeded the threshold. But that's not, it doesn't work for industries. Like you can't, they need a prescriptive medicine. It's just like our solutions is like a doctor reading the x-ray and saying, hey, it's this problem that you have and you need this type of surgery. And that's why it's very challenging and very complex. But like our assertiveness model, so just for instance, engines, pumps, gearboxes, bearings, we have a bunch of rotational assets label already. And it varies from the minimum 75% to maximum 95%. So you imagine even in the lowest level of assertiveness, 75% is like seven out of 10 others that you're going to receive are going to be assertive. You're going to the machine, sure of what the problem is, and you're going to fix that. And the proportion when you work in the industries of how much they're paying and how much they are saving is like, it's huge. It's more than 10X, you know? So, and the challenge to work with industry is really the cost of acquisition per client. And really like just mining more clients because they're generally more hard to find. But once you find that, it's really easy to convince. And like the lifetime value of this client, it's really long for all the base. They just, when they enter, the product is good and they stick. And just like- And then they buy a ton of sensors. They're like, we'll put it on all the conveyor belts. Okay, now we put it on all the other machines. Exactly. They don't even need to buy. We offer in just any type of like SaaS model. So it's really renting the sensors. You pay monthly fee. Most of contracts are annual plans, but like really you're not just possessing anything. You don't need to just put a bunch of money in the beginning as a setup cost. So we increased a lot the adoption of just predictive maintenance, especially for SMBs, for small and medium sized clients. Because essentially like predictive maintenance has always been a privilege for the hugest of the industries, right? Let's dive into predictive maintenance a little bit because a lot of our viewers dialing in are from the software industry where- I imagine. Where they're not working with hardware on a regular basis. So I imagine like if I'm a factory floor manager with a lot of equipment, with sensors, I'm looking at a traction dashboard and I'm looking at all the machines and maybe in a given day, some of them will pop up and say, you should check out machines like 256 and 312. And those haven't broken down yet. Is that right? And then, but I should go check them out because they are likely going to have this, that or other problem. And so that's really changing the uptime, right? Because you don't have to wait until an actual failure. Exactly. And funniest thing is that we even have a window to say when the machine is going to break. Like we have a timetable that we can appoint. Hey, it's going to break in 30 days or 60 days. But we noticed that because of the human element, we can't say, hey, you have 30 days to fix this machine. Humans are postponing mechanism naturally. So they're going to let for the next day, for the next day and the next day. So we say, it's breaking out. Then they're going to check and it's really a problem. So then they fix immediately. So what we see nowadays, when we started, we were afraid of the false negatives. What is the false negatives? Like the missed failure. So essentially machine has broken and we didn't say that a failure was about to happen. But interestingly enough, that has never happened. What happens more is like, we say the first time it's going to break. We say the second time it's going to break. And the third time, and then it breaks. And then the manager comes in like, really? You didn't check Traction's platform? I can't believe you're fired. So it happens quite often with maintenance technicians. So we're like, you got to trust the system. I know you have years of listening to these devices yourself. And sometimes they really listen like, nothing is wrong. Just like, I remember my grandma, when entering elevators, she would be like, this is about to break. And then next day, it would be just the warning sign in the front of the building saying, we're fixing this elevator. Sometimes, but if you can't hear the failure, it's just like too late. Forget about it. It's already broken probably. So let's share some cool news. You recently just announced a funding round. Can you tell us about the good news and who some of the participants are and the new people you've added to the board and what you're going to do next? Sure. So we recently raised $3.2 million as a seed phase. So it's very uncommon to raise a seed phase at this value for hardware startups. We know that FinTech is like 10, 15, 20, 30 million dollars at a seed phase. But generally, harder startups, they go much lower. So we think we got a very, very good deal in terms of how just to kick off and just start this new phase of the company. We have half of the investors are US investors and half of the investors are focusing lots of market in Brazil. And the funny thing is that hardware was once kind of radioactive for venture capitalists. And we know that it's not a big deal. But the thing that I've seen with US investors is that they've passed so many deals. So they missed so many deals in the past that they're a trigger like, oh my God, one more hardware startup and I can't miss it. Because they've seen IPOs, they've seen like unicorns coming out of this field of once they thought it was like unsexy to invest or something like that, just because you have essentially like it's a piece of hardware. And the hardware, the thing about our company is like hardware is not a deep tech. We could do with potentially existing sensors in the factory because all the technology is in the software side. But hardware for us, it's a mechanism of opening your market. So go-to-market accelerator that if you're going to use infrastructures, it would take one to two years to integrate with Siemens devices, Schneider or any other manufacturer. I think it's universal, right? You can name a few as well. And in terms of predictive maintenance, we also got like very good people in the board. So the CEO of Siemens in the Middle East, she's joining us and she's just absolutely amazing in terms of like product distribution and so on. And that's funny because like in some ways, any manufacturer like Siemens, they're kind of competitors in the bottom line because those manufacturers, they're a bit of incumbents and they're not going to say that they don't have your product to the end customer. They're too proud to do that already. So they're going to find a way to build that tailor-made to you just like consulting, but it's going to cost a lot. That's why we say like we're democratizing predictive maintenance that once was just a privilege of huge factory, huge plants like Coca-Cola and so on. But like all the other mid-tier and the small market are totally unassisted. And really what there was left for them was just like, just hold. And when it breaks, you fix it. So I would be just tired of that and there's not enough solution available for them. So that reminds me of something that you wrote about in one of your blog posts, the story of a fan and the toothpaste box. Can you tell us that story? Yes. In industry phase, it's funny because like there's a tendency to go very deep tech, like it needs to, I don't know, like this, all the, also this robotics, you need to do everything. Like it needs to self-fixing up your machines to be very high tech and so on. But most of the solutions that actually work and actually like, it's gonna just thrive in this scenario. They're just simple, just simple. And I used to say like simple, but not simplistic. It needs to be simple so that everyone can understand. But if it's too simplistic, everyone already thought about it and are doing themselves. But like most of the solutions, you just need to just be a little bit more creative. And the thing about this... Can you tell the story? Because I don't think most people who are viewing have heard the story about the box. Is that, I don't think it's only, it's a Brazilian story, but like there was once a factory and they were trying to figure out how to kick the boxes of toothpaste that were empty. So a bunch of boxes in the conveyor belt, they were just like leaving empty and just would essentially arrive empty at the supermarket. And then you put it on the shelf and then a customer would buy it open and it's just empty. You bought an empty box of toothpaste. So that's very frustrating. And engineers have come along and just like, they hired consultancies. They hired like industry 4.0 experts. They hired it all and they're like trying to build a robot with lasers and like infrared and trying to wait. One of the solutions were like, okay, let's wait and then kick when just an empty box appears. But then there's a lot of problems into that. If your conveyor belt is not aligned enough, then just two kind of toothpaste is going to go along in the same time and you're going to kick them all. So you're going to lose product. And that's like very, if you have 3% of mistake, it's going to already be bad. So there was just one intern there. They're like, hey guys, I have an idea. They're in front of the conveyor belt. And if like an empty box appears, it's just going to eventually blow out of the conveyor belt. And it just worked fine. And I think the solution of the fan is actually one of the most common in conveyor belt industry nowadays to just like discard some products that goes out in a favor like that. And it costs literally like you go to Ikea or Costco, you buy the fan for $30 and there you have it, very sophisticated solution. It's such a great, elegant solution to an industry -wide problem. Yeah, it's 4.0 for sure. It's like really high tech. So I'd like to shift the focus a little bit onto sort of a personal journey story. Can you tell the story of how you met your co-founders and how this company got started initially off the ground? Yeah, this is a very interesting story. So I went to do computer engineering and I met my co -founder in 2015. So essentially six years ago, when we count in terms of years, I think like, oh my God, we're getting old. So it's just like, we just moved in to the same apartment to start building just a crowdfunding platform together. And that was essentially like this platform we were building for three years. And we were supposed to kick start some projects, like various projects and so on, mostly in health sector. So a lot of surgeries, I think this is very common in US probably, like a lot of surgeries are really expensive. And if you don't crowdfund them, there is no way that these children is going to get the treatment that they need. And it was really by accident that we had this idea and we started building that. Again, just like we wanted to build something cool and we just don't think so much twice. We're like, let's start then. And one of the problems is that it got so social that we didn't attract any investment of private sector, neither the public sector. So we're like, huh, this is weird. And so one, so essentially we're just like, okay, let's turn into an NGO and move to the next challenge. And then next challenge, I went to study at UC Berkeley in California, go bears, whoever's watching that. And I went to do like the extension of my post, my undergrad. And that was one of the first times that I actually started meeting other hardware companies, really interesting like this. I didn't even like so much hardware in the first place. There's not enough space in Brazil to build that. So although we've done computer engineering, I was not so much into that. So then I started building a software to predict health chronic diseases. I think this was one of the points that we really got interested in the data part of the solution, but like not the data visualization and all this thing that's like power BI, just because the companies nowadays are just buying a bunch of data scientists that essentially you just do dashboards. Like you have no other role. It's building dashboards and it's not so intelligent and so on. Like, I think the challenge of being real time data base is really incredible and really hard. So, but that didn't help, didn't fly because of the team. So me and Gabriel were really nerds, like just focusing in building the software. So just like, ah, we don't care about sales. We just want to code and just build the better product and so on. But then we learned like one of the first lessons in this, not only about what you go, it's not only about the product and so on. You got to have interest in all the areas. And then I came back to Brazil and the thing is that maybe another connector there is that my father was a maintenance manager and Gabriel's father as well. So in a lot of barbecues that we're having, we're talking about just the general idea of working with industry. But that seemed too far. I think for anyone, it's like, no, working with industry doesn't make so much sense. I don't know. Do you know anyone who works in industry? Probably just probably not. I mean, the fact that you're able to have a conversation about what your father's pain points are in his job is just so insightful. That kind of information is out of reach for a lot of people who, for whom working within industry is probably seems further away. That's really true. And the thing is that, well, it seemed really far, really hard fetch, but we're like, it was again, just really this decision slash accident that I said, okay. And it wasn't even about opening the company. It was like, we were kind of like idle for a while and figuring out what we want to build. And then I thought, I'm going to work at industry. It's like, if I'm not going to go now, I'm probably never going to do that. And you don't have the interest once that you get inside, it's like, it's like magical. But so, then I went to work in a pulp and paper factory. And so this paper factory was like more like countryside of Brazil. So I moved in and so one, I was supposed to stay like one year there. And then I went to work with the maintenance team. And I was like the only kind of software engineer that the company hired a bit, a bit of an anomaly because like industries, they generally don't hire soft engineers. They just outsource that anyone who's there, it's like production or maintenance line. So, but I started like having dialogues with them and get more and more interest in the maintenance field. And the reason is that like, when I was talking with the production people, they're like, yeah, whatever, like we're producing 1% more, 1% less of paper, who care. And there's more like laid back and so on. But maintenance was like, oh my God, it's going to break. It's not going to break. I'm going to be here like to midnight. And so what they were so just like this energy. And if you think about the, just like the timeline of industries, you think, okay, robots are going to replace the production workers. Unfortunately, like it's just like anything that's kind of, it can be manufactured by people essentially can be manufactured by like robots, but maintenance, it's a much more like timeless concept because if you don't build a robot that fix other robots, and it's like really complex. And it depends in series of factors because it's not mechanical thing like this, that, this, that, this, that, and that's it, you know, assembly. No, it's like figuring out what the problem is. And then unbolting this, that, fixing that. It's like, this is going to be just probably still done by people at least a hundred years, you know, which in production line is like 20 years, 50 years time span. So they're like really energetic and really honest people. So I really like to work with that a lot. And then I came back and I quit my job and I was like, Gabriel, that's it. We got to open a company here. And then we started traction that was in the beginning of 2020. So it is like basically out of an accident, but we really got interested. And the thing is that another thing that happens with startups that start working in industries, and that's an advice that I give for any of my future competitors is basically like don't fall so much for the proof of concept category. You know, like an industry generally, one of our mistakes in the past is that we kind of started developing our hardware inside the customer. So that has some benefits and some problems. Like first benefit is that the customer is actually paying for the development. So it's fostering the development. But like the huge con about that is that you kind of get blind of looking at the necessities that the actual market has and the all the customers pool has. And like you're focusing more as a consultant to solve a internal problem of that customer. And that like, this can just throw your product in the trash. You know, and I'm seeing many startups going from client to client to client, never actually finishing the building of their product. And like, and then you essentially, you turn out to be a consultancy. And nothing wrong to be in consultancy, but like a startup, you're kind of, you were born to just find product market fit. And if you don't find this, it's like very frustrating to go from client to client. It's a different solution that you're building, you know. Oh, absolutely. Right. Because if you, if you go into the, if you find yourself in the consultancy trap, your revenue is proportional to your headcount. Yes, a lot. And it's not a exponential curve. That's true. About headcount, we can talk about like one thing interesting about hardware companies. It's just like, that I was noticing those days is that actually we were doing some parallels with the teams, right? Like how different is to build a hardware company. And if you're a SaaS or a software company, you're just have three people. You're probably fine for one, even two years, depending on your SaaS. Like, and the thing is, the moment our product is shipped to the client, we become SaaS. We're like customer support, our customer service online and so on. What do you have as a problem? Helping them onboarding and so on. But before that, it just looks mostly like Apple, you know, in a very low stage. It's like a bunch of hardware engineers, like a bunch of labs and so on. And to be able to just fasten the curve of hardware development is much harder than the software curve. So we had to start, for example, with at least 10 people to be able to create something that was actually a bit relevant, whether in a SaaS, probably just me and my co-founder will be fine for one year. So like, it's just really those three gravities that you have. You have the hardware, you have the software, and the data science compound together here. And then you have sales and marketing in another gravity. But like adding this extra gravity, it took a lot from us in the beginning. Because like, I think, as founders, we put a lot of our own personal money inside the business, which is kind of uncommon for SaaS. You're going to do that maybe to hire one or two people. But like, we had to do that to hire people and to just like buy hardware parts and to do hardware projects. And so it's overwhelming. And a lot of things can go wrong. And it goes in a lot of time. So that's why it's very challenging. Thank you for sharing that insight with us. I think that's very appreciated by a lot of our viewers who aren't really, you know, dealing with hardware in the day-to -day. Yeah, but just probably don't, you know. But if you're going to, I'm happy to share more advice. You can just drop me a message on Twitter or email. So in our last two minutes or so, I wanted to touch on a blog post that you had written about a book that you had read. Would you like to recommend this book that has influenced how you think about things? Yes, of course. So, well, I like, essentially, there's three books that I think I'm going to recommend. Like, one is How to Raise Successful People by Esther. She was once a Berkeley professor. And it's a really great book. It's generally about her raising her daughters. And they went out to be CEO of YouTube, CEO of 23andMe, and so on. It's like, there's kind of some elements there of how she raised their education to be very liberating and just very entrepreneurial. Another thing is, like, hard things about hard things. It's just common sense. Yeah, it's really great. It's really suffering. And when you, you can feel what's been felt at that time. And also Traction with O, the name of the book. It's like a very simple book, but very cool to identify just channels and sales distribution. Awesome. Thank you for those recommendations. I've read hard thing about hard things, and I also second that recommendation. Yeah, you've got the other two. All right. Thank you so much, Jade. Yeah, thank you so much for being on our show. It's been a great episode. And thank you all viewers for tuning in. And yeah, join us next time. Remember, hashtag Bluecap. Let's go. All right. Thank you so much for joining us. Transcribed by https://otter.ai