Transcript 0:00 [upbeat music] Welcome, Daniel, to the IoT For All Podcast. Thanks for being here this week. Yeah, thanks a lot for having me, Ryan. 0:09 So to kick this off, I'd love to have you do a quick introduction about yourself and the company to our audience, if you wouldn't mind. So my name is Daniel Lang. I'm now the chief marketing officer for Toradex. 0:18 So Toradex is a company, it's more than twenty years old, and we do embedded computing. 0:23 So we really grow up building system, uh, on modules, and we are used by over three thousand active customer in a lot of verticals, in a lot of, uh, verticals where reliability is very big, like medical, industrial automation, and so on. 0:38 And then also today we'll talk about Torizon. 0:40 So Torizon is a product from Toradex, and for Torizon, I'm, uh, the general manager, and this is a, a offering with a operating system and a SaaS offering, which really makes it much easier for people who build these products, where we provide the hardware, uh, it makes it easier to build it, to develop for it, and then also keep them up to date, maintain them, and also big, big topic nowadays is cybersecurity. 1:04 There's a lot of new regulation. Also make sure, uh, people can comply with all these big regulations coming now. So on the Toradex side, so you all are-- what type of hardware are you focused on? 1:15 We are focusing on Arm systems, so we are, we are doing Arm, like what you know from your cell phone, and then we build system on module, so small computer modules, so people can integrate that into trucks, into medical devices, into robotic system, into energy devices like electric charger, power walls, all of these stuff. 1:38 Everywhere you need computing, and you need to be power efficient, you need to be small, and it need to be re-really reliable, that's, that's very, very common. 1:46 So I wanted to ask you kind of a high level question to start. Um, so if we're look-- if we look at the embedded space, um, IoT For All for us, we've been in the space since two thousand sixteen, two thousand seventeen. 1:56 So give us a, like kind of a, a high level overview of, of where the embedded space is right now, kinda how you, you know, see it, where it's kinda come in the last few years, just to kinda set the stage for, for our audience. 2:08 Of course, over the years, like anywhere in computing, uh, you know, power and capabilities really grow. So at the moment, I would really say the big buzzwords, the big topics is AI. 2:20 Uh, you know, at the edge, you now can do even generative AI at the edge. There is this trend we call, like, tiny ML. 2:27 So, like, couple years ago, you needed big GPUs, uh, big, uh, PC system or cloud system, and of course, today they're still the, the one which are in the news because there's a lot of money, uh, too. 2:38 But now you can actually bring down AI even to microcontroller and very small power optimized system. You can run it on the edge, we call on device or on the edge, uh, so you don't need to send up your data. 2:50 This is a lot of advantage in privacy, it's more energy efficient, and so on. 2:55 So this is really a trend we can see, and nowadays almost all products we have have some kind of neural network accelerator, so you also don't need to run that on the CPU anymore. 3:05 You get special hardware, like in the past you get video accelerators, and you get GPUs. Now almost all our products are available with neural network accelerator, uh, so you can, uh, get quite some performance. 3:17 So I, I would say this is a very big trend, and then on the other side, security, right? 3:23 It's really driven, especially in Europe, uh, but also in the US with FDA and so on, governments really pushing to make stuff more secure, uh, because there is so much fraud and, and, and things like that. 3:35 And then of course the wars make, uh, IoT system a, a good target to bring up a, a factory and things like that. So I would say security and AI are, are really big trends at the moment. 3:46 And with the AI piece, where are you seeing as the biggest, I guess, changes that that's enabling or the biggest advantages that it's enabling? 3:54 Like what-- you know, what are you really seeing AI do from like a, I guess a, an ROI standpoint on the devices and the, what it enables for the use cases? 4:02 'Cause, uh, you know, a lot of people talk about AI, but there's really not a tangible benefit at times. It's more of a buzzword. 4:08 So from your perspective, when you're talking about AI, what, what are the things that people are really getting with AI being pieced in there, uh, now with a, with, you know, across, uh, the different areas? 4:17 I mean, one thing of course everybody knows, and it's kind of the early on on the thing, was predictive maintenance, right? So you can measure vibration and things like that. 4:25 I think this is kind of the textbook example everybody brings. But I mean, we, we are in so many, uh, different application. 4:32 I think just now, j-just the meeting I had before was about a company which does medical devices, and when you do, for example, getting into a human and you, uh, use heat to close vessels, there is smoke, and it's very hard for the surgeon to do that. 4:49 So they remove the smoke, so it's better to see for the surgeon. We are really-- We have a lot of medical things, so there's a lot of application to support, uh, things, to detect cancers and so on. 5:00 So, so this is a very big one. Another one, we also see a lot of application, it's already deployed, let's say speed cameras. We do a lot of, uh, speed cameras, smart cameras, and so on. 5:10 And of course, you can now do so much. So for example, in Brazil, we have a customer, uh, they have a lot of fraud. People fake the number plates on the cars, right? 5:18 So even you read the number plate, but now with AI, they can really easily add additional stuff. They can id-identify the kind of car it is, even the brand. 5:27 So now they match all of this so they can detect this number plate should be on a Toyota, but it's on a Ford, so something is off, right? So this dishonest that. Food processing, we are big. I could go on. 5:39 So we see a lot of things. And what's very special at Toradex for this embedded, we are highly fragmented, right? So each our customer comes with use cases you probably didn't even think about, right? 5:49 Because they're so niche into something, and then they come up with the ideas. It's really not that we drive them, right? We enable to get easy access to the neural network accelerator and that. 6:00 But the idea and the concept and the thing like that is, it's our, uh, it comes from our customers. So let's shift over and talk a little bit about your new product. 6:08 Tell us a little bit about why it was developed, just the story of that and, and, you know, just kinda set the, the set the stage. 6:14 Got some follow-ups to ask about it, but I'm just curious to get a, get a sense of, um, you know-Why does it exist? So Torizon, uh, started actually to help our customer who, who de-- uh, worked on Windows CE. 6:26 So maybe a little bit older people, they know there was something like Windows phones around, uh, at the time, and there was also an embedded solution which uses the same technology a-and of course, that didn't go anywhere. 6:37 There are not no Windows phones anymore, right? So we needed to migrate these people over, uh, we a big fan of Linux, open source, and all of that. 6:45 But if you come from a Windows world, was very hard to start with, with Linux because in our industry, the most common is a Yocto project. It basically means you build your own Linux distribution. 6:57 And coming from a Windows guy, Windows person, application developer, and you ask your first time, "Hey, you need to build your own Linux distribution," not just using Linux, you have to build it, maintain it, and all of that. 7:07 It was just too much, right? So that was kind of the trigger, and we did a lot of research. We, we asked over two thousand engineers in our industry what they need, where they see the market going, and so on. 7:18 And so we developed this Torizon, which basically is an embedded Linux really designed for all these use cases I was just talking about, but making sure they do not need to build their own, own Linux system. 7:30 They can really focus on their application, on their car detection thing or their medical cleanup of the, of the video and all of that. 7:38 And at the si-same time, we already saw that people will be more and more connected, right? So all the IoT, we really see that coming. 7:47 And we also saw that security is very important because we are not into the consumer industry. We are really in this industrial and, you know, security camera, and we already saw security, and we decide to go with a 8:00 very robust security framework. A little bit complicated one. This comes from the automotive industry. It's called Octane, and it's designed to protect against state level attacks. 8:10 So not just your fourteen-year-old guy who wants to bring his neighbor's camera. It's really governments who wanna bring down, let's say, BMW fleet or like they wanna do this one. 8:20 So we adapted that, and at that time, people were a little bit, "You guys go a little bit overkill here, right? 8:26 Uh, I mean, it's IoT, so," uh, but we were able to abstract the complexity, so our customer doesn't really see. It makes it a little bit harder for us. 8:35 And now in the meantime, a lot of the stuff, it's in the regulation, especially European Cyber Resilient Act, some FDA. Now we are in a very good position because we did that work ahead of time, and it's really good. 8:49 And, and state sponsored attacks, it's daily, daily business here in Europe with, you know, Ukraine and things like that. 8:56 So this is now really a topic, and we invested that early on, and now in-- we are in a really good position, and it's appreciated by a lot of our customers. 9:05 Tell me a little bit more about how Torizon reduces the time and cost to kinda meeting those standards that, that we have today. You know, the EU Cyber Resilience Act, as we mentioned, FDA cybersecurity guidelines. 9:16 How are you kind of helping address that and future-proofing technology stacks to keep your customers ready for, you know, whatever is the next thing that comes? 9:26 So all this regulation, one really big point of there is all kind of the realization whatever you ship today, it's probably not secure, right? 9:35 Even you did very good, you know, analytics and all of that, somebody will find some, you know, exploitable vulnerability, and you will need to patch. And, and you can go through, this is really a realization. 9:47 So the first thing we provide is a very secure, very reliable over-the-air update system. 9:53 And many of our customer already had some kind of over-the-air update system, but it worked maybe ninety-nine point nine percent of the cases. But our system is really made that it works always. 10:05 There's a lot of reliability checks and rollbacks because even many of our customer had over-the-air update system, they didn't wanna use them because th-they know, uh, maybe, you know, the, the internet goes down, power goes down, customer makes mistake, they break a device, and you don't wanna break a car manufacturing line, right? 10:24 You, you're gonna be in really big trouble, right? So a lot of things, reliability, uh, we built in, and then of course also security because you don't wanna have the OTA system being the main attack vector, right? 10:34 So, so this is one thing, and there's a lot of other things. So there is secure boot. 10:39 So when you boot our s- our system or an embedded system, you can basically lock it down, and almost all the vendors like TI, NXP, and so on, they all have features integrated in their chips, but they're not that easy to use. 10:52 So we make it easier to use them. We abstract that a little bit so our customer don't need to be embedded, uh, Linux security experts. 11:01 Then I think another one, I mean, I could go on and on, but one other really interesting one is the vulnerability monitoring, so that's also regulation. 11:08 So everything we ship, we monitor for known vulnerabilities, and then if there are known vulnerabilities, we let our customer know, and we have humans translating the vulnerability for our customers. 11:21 Because if you scan a Linux, you get hundreds of vulnerabilities. Most of them are not exploitable. So we are explaining to the customer, "Yes, you have to exploit. 11:29 You know, you have to worry about it in this case," and this stuff. And we ca-- we, in our products called Vulnerability Manager, and I think this is also very useful. 11:37 And then there are many more, but I think this, uh, you'll, you'll get a little bit idea how, how we are tackling that. 11:42 Last thing I wanted to ask you about the regulations, um, how do you think these regulations help product quality? 11:49 'Cause a lot of people can see them as restrictive and stuff, but how are you seeing them help product quality get better? 11:54 If you look at security, and we did the service in the past, say, "Hey, how important is security for, for, for you?" And everybody said it's pretty important, right? 12:03 I think like four point five out of five or something like that. 12:06 But then we also ask the customer, "Rate the security or stack rank it against other things, time to market, uh, pri-- you know, cost of the product, ease of use, features, and so on." It was never really highRight? 12:20 Because the value of security is for a, a device manufacturer was not very high. 12:25 And it really now leveraged the, the playing field because everybody who wants to sell in a certain market need to, uh, comply, uh, at the same, same thing. 12:34 And I really think companies who have high quality requirements typically also do a better job at, at security, and so they can differentiate it. 12:44 So your competitor may not even be able to sell, they maybe get blocked, they may be late to market if they don't have, uh, the quality mindset also on the security side. 12:54 And then of course, overall, it will make sure that your, you know, uh, your camera at home or, or your speed camera is not used in a, in a bot attack on, on, on something, right? That, that... 13:07 To be honest, in the past, many times the end customer didn't really got the penalty for this, right? But now it, it really helps, um, you know, that stuff like that doesn't happen any- anymore. 13:17 From your experience in the industries you work or maybe just even in the industries that you observe, which sectors do you think are most vulnerable right now given everything going on? 13:25 I think a lot, one we call critical infrastructure, right? So power. So we are in EV charger, power walls, solar converters, and all of this, right? And if I would be a government, I mean, it's not if. You see it. 13:40 Uh, they, they let drones flying into, uh, power, you know, stations. But, you know, the, the more elegant way is to just do a, a, an attack, right? 13:49 And if you look at Venezuela, power went out there, probably a cyber attack, right? So, so this kind of thing, definitely power, and then industrial, uh, manufacturing, right? 14:01 So factories and things like that, I think they're also definitely at risk. 14:05 So if we look out, you know, three to five years from now, what do you think the biggest shifts that are coming for embedded and IoT teams in general, where are you guys focused on? 14:13 Like, looking kind of more, more forward-thinking here. I definitely think the, the AI stuff is go- going more mainstream. It will be... 14:19 You know, at the moment it's a hypey thing, right, and you maybe know this hype curve, right? I think we are very high up that. 14:26 I think it will be get, get kind of normal, gonna be used i- into product, like, more natural. Um, I think this is that. And then also I would say the whole agile process, uh, will get more better, right? 14:40 So I think in our industry or so, Tesla really showed, hey, you can update even a car. Um, and you see more and more companies adopting this model, also forced by the vulnerability patching and so on. 14:52 But we still see there's a lot of catch-up. So we still see a lot of companies quite behind. I think in the consumer IoT device may be a little bit different. But in our industry this will still take time. 15:04 So I, I, I think, you know, looking out three years, this, this will be maybe then very common, while today I think it's still not that, that common. 15:12 If you could give one piece of advice, maybe brutally honest advice for people out there, companies out there listening to this that are product teams shipping c- like, connected devices, what would that be? 15:24 Like, what's the one thing that people really need to be paying attention to, or that one piece of advice that you found to really, you know, really be helpful? 15:30 If you wanna ship into the European, uh, market, really check your supplier what they're doing about CRA, right? 15:37 Really look what the CRA is if you don't know it, and then go over your supplier, contact them, what are they doing? 15:44 Because we do know, and it also affects us, right, that some of our supplier will stop product because they cannot comply with the CRA. It's not worth, you know, to continue them, right? So maybe a very concrete, right? 15:55 There's a lot of other things of course you have to do, but, you know, checking CRA, go through your, you know, main supplier list, understand what they're doing, what's their plan, so you're sure you not get surprised that there is maybe something not available anymore or you cannot bring that certain product, uh, CRA com- uh, compliant. 16:14 This has been a great conversation. I, I... It seems like y'all have a lot of very exciting things. You're playing in some really cool industries. 16:19 For our audience out there who wants to learn more about what you're doing, follow up, touch base, what's the best way they can do that? 16:25 Yeah, of course, check out, check our website, right, at www.torizon.io, or I don't know if you can share my LinkedIn. Of course you can, uh, always, uh, uh, contact me and reach out. 16:37 Uh, we have offices all over the world in each, in each continent, uh, almost, local language. Uh, so you can check, you know, reach to our local offices. 16:46 Everybody at Toradex is quite excited, uh, to talk about and help you and learn more about your product and, a- and make you a success. 16:54 Because our own, our business model is really we only profit if our customer are successful. Uh, so we are, you know, we are all in to support you and, and, you know, get you to market and be a suc- uh, a success. 17:07 Thank you so much again, Daniel, for taking the time. This has been a great conversation. Uh, excited to get this out and hopefully talk again soon. Thanks so much. 17:14 Thanks for an opportunity, and it was very nice, uh, talking to you. [upbeat music]