Simon Maselli - Xvision AI

Welcome to another Founder Wisdom podcast. We have Simon Messali with us today. Simon's a friend, a client, uh, he's the CEO and founder of xVision AI. They're rebuilding, uh, traffic infrastructure with edge AI, 100+ deployments across 6 countries, uh, based in Melbourne, Australia. So Simon, welcome to the pod. Tell us a bit more about yourself and what you're up to.

Yeah, Charles, awesome to be here. Thanks for the intro. Yeah, look, just out there to build or to transform the way that traffic infrastructure is built. Today we've got a situation where traffic infrastructure is really expensive to do a full rebuild on an intersection. It's about $1.4 million and takes months and months, 18 months on average to retransform these intersections. With our product, you can, um, yeah, do the same sort of upgrade, uh, in the space of hours. So it's a, it's a really transformational product.

Yeah, and it's something that we all experience, right? Like, we've all been stuck at a silly red light with no one passing on the other side, and it's like so frustrating to be stuck out there and to be inefficient. Um, I guess like us founders and CEOs like value our time the most, right? So like, we, we're We're the ones typically feeling the most pain, but I think anyone that's driven and that is stuck in the traffic or is stuck in a system that is inefficient sort of gets crazy with it, right? So it's a problem that a lot of us can, can be like, oh shit, this is important now. Why isn't this being fixed as quick as we would want is the question.

Yeah, I think, uh, before we jump into that question, I think, uh, you know, it's probably worth mentioning that I feel that the bigger problem that we're solving here is, is really this safety aspect. So definitely there's a, there's an element of congestion in there and an element of time saving, but you know, you can't really put a value on people's lives. And, and you're a father now, Charles, you know what it's like to have kids and all of a sudden your, your world changes, right? And you sort of start to look at things a little bit differently. And since I've had children and, you know, uh grown up a little bit, I feel as though that safety is definitely something that's important. So looking at things like those more vulnerable road users, pedestrians, motorcyclists, you're also a motorcyclist, I know this, so, you know, making sure that these sorts of road users are also protected is, you know, what we're all about. So to answer your question, your question was why isn't it being addressed today, and it is being addressed today. There's no question that road authorities are trying to solve this problem. The problem that they have is that their solutions today are not scalable. I mentioned earlier, you know, $1.4 million to upgrade an intersection and include all of the new technology is just not scalable at any sort of large scale. I can't think of a better word. Across the city. You know, where our product X-Vision AI is, is really out to, to change that. Uh, you know, we're making it much more easy to install, uh, you know, reducing that time to, uh, to get the implementation in there and also significantly, uh, reducing the cost.

Interesting. And yes, it's been implemented, but at what speed is the question. So you've dealt with various governments, or at least you're knowledgeable about various countries out there. You're in Australia. What municipalities, what governments would be the most receptive to your solution and why?

Yeah, awesome. So there's two levels of government that are really interested. There's obviously the road transport authorities. So these are the guys that look after all of the main roads across the country. And in other countries, there's very similar bodies. US has a similar similar configuration in their bodies to what we do. So that, you know, there's got the major authority that looks after the major roads, and then when we bring it back to the more urban areas, we've got the local governments, the councils, and those sorts of things. So there's really that, you know, those two segments that are interested and that we've received most traction in today. However, we are starting to see a shift, which is really interesting. And there's a lot of, um, uh, private organizations that are definitely taking an interest in this technology, and they're seeing it in two aspects. One is they're using it to be able to build their, um, you know, their estates much faster and much cheaper, uh, and manage things like logistics yards. So, uh, you know, we've got a customer, Australia Post, who has a lot of trucks moving around. And in some of their larger facilities, we have these cameras that help identify safety and near misses and the incidents between the different vehicles and all of those sorts of aspects. So that's really the interesting story from our side.

And for example, would there be smart cities that would move faster on that? I don't know, Singapore, Dubai? Or even like, uh, fantasy cities like Neom, like, would you feel that these folks would, uh, take decisions faster and adopt your product quicker?

Yeah, definitely. I think that some of those cities, uh, definitely, uh, like to adopt the new technologies. One of the, um, one of the things that's really holding us back at the moment is achieving certification. So we are very different to a lot of the other providers out there of similar technology. There's guys out there doing, you know, AI video analytics that sort of have a similar outcome from an analytics perspective, but what they cannot match us on is being able to directly influence the traffic at the intersection level. And that's something that we've spent a lot of time and effort to get right, being able to actually interact with the existing controllers and the existing infrastructure to provide a deterministic, you know, response to what's happening on the road. So looking at behavior and those sorts of things. So the thing that's really slowing us down at the moment is getting that certification. So at the moment we've got a program called our Extreme Validation Program, and what this is doing is this is collecting evidence from our cameras, and they're installed in some really extreme locations. But the really cool part about this, Charles, and I think you'll like this, is we're doing things like hiring racetracks and sending cars around at extreme speeds and, you know, putting things like crash test dummies on the road to, you know, bring the cars next to the crash test dummies to really understand the way that our product performs in those extreme circumstances. So when we bring it back to a road where speeds are much more conservative, We would hope people are much more conservative on the roads. Sometimes you don't see that, but, you know, and this Extreme Validation Program, as well as, you know, validating the product, it's creating that evidence that when we move into these new markets, we've got the body of evidence to put in front of those authorities and hopefully get through their approval channels in a much faster way.

And these certifications, they cost, they take a lot of time, right? Are they worldwide certifications? Tell me a bit more about that.

Yeah, so each jurisdiction has its own certification process. The good news is that most of those certification processes are based on engineering standards, and therefore, you know, they're kind of repeatable. But again, each jurisdiction has their own set of tests that they would like to see performed at their own speeds and their own scenarios. So again, this Extreme Validation Program is about collecting that evidence before we even approach these authorities to make sure we've got everything together that when we get into— let's say we move to Thailand next— when we get to Thailand, we've got all of the evidence already collected, you know, we presented to the government body and we get the— I don't want to say tick and flick, it's the wrong word, but we get the equivalent Exactly, yeah, equivalent done as quickly as possible.

Yeah, nice. Um, yeah, how dangerous are roads, right? Like, generally speaking, I'm in Mexico, I would say like very dangerous here, like especially on a bike, right? Like, uh, it's funny, right? Like sometimes I, I go on the road and it's like, wear your seat belt, you know? They have like these panels, like, dude, why, why did you put these panels if you allow me to be on a motorbike, right? It's like the, the great irony here. So like, I would say super not safe, but in developing countries, probably a bit safer, uh, generally speaking. I think like there's more laws, more rules, people drive, drive a bit safer, but still very unsafe, right? Which leads me to the, the next thing. A bunch of people talk about self-driving cars I think you're approaching the problem differently. You're like, not everyone can afford a self-driving car, right? It's not given to anyone. Let's take it from the government perspective and give safety to pretty much everyone at scale. Is that like how you approach it?

Yeah, let me try and give you two answers here. So the, the first answer is the self-driving cars. It's definitely something that we've thought about, and it's definitely something that we're addressing directly. So self-driving cars definitely have a capability on board to perform whatever function they need to function to do their self-driving. So what we do is we've built in V2X into our product. So we're able to provide another set of eyes from a different perspective. And when the cars have the capability and when they have the integration to V2X, so that's vehicle-to-everything, We're going to be able to provide that information to them, and we're already working with a couple of manufacturers to try and work out the best way to do that. And there are standards that are being put together at the moment that will make this a lot easier. So again, yes, definitely it's going to change the landscape. But the other part of this that you were talking about is the developing nations, and the developing nations are actually the ones that are showing the most interest in the product because most of these developing nations are becoming more, um, what to say, westernized. And I'm, I'm, I'm using that word loosely.

Well, care, care more about their taxpayers, right?

Like, exactly, exactly.

They don't want to lose their source of money.

Exactly. So they're trying to, trying to protect more. So, so here's a, here's an interesting stat for you. 1.14, I think it is, million lives, uh, lost each year on roads through death worldwide, right? Worldwide.

Yeah, that is fucking crazy. That is, uh, yeah. How much, like, compared with, uh, I don't know, diabetes or heart attack? I think it's like 10 million a year, but it's still gigantic, right? It's like, it's a very strong leading cause of death. It's like top 5 or top 10 for sure.

It's top, top, top, uh, number 1, uh, source of death between 19 and 20-something.

So That's not health-related, that's like—

yeah, that's all, all, all causes of death. It's the number one.

Yeah, for that age group, homicide or suicide and all of that, like on the road is— yeah, because our bodies are not made to withstand impacts of like even though over like 40 kilometers an hour, we're not— we're just not made for this as humans.

Yeah, and people don't realize they're massive forces even at low speeds. The forces generated by a, you know, 1-ton car or or, uh, you know, even heavier in some vehicles, is massive. And like you say, the human body just can't withstand that. I don't know.

Um, yeah, I can't imagine my bike, by the way. I mean, I'm— my excuse now is like selling my S1K for motocross, right? So that if I crash on a tree, it's my fault. But just like standing at a light, if a drunk guy comes behind me and hits me, I have very strong chances of— well, 100% I'm ending up in the hospital, right? So incapacitated, but also very high chance of dying. So it's just not smart.

Yeah, um, Charles, it's, it's not yourself that you have to worry about on the roads. Like you say, it's, it's everybody else. Yeah, it's everybody else. Um, and then again, I used to race motocross when I was younger, and, um, I, I've had a couple of fairly, uh, bad accidents. Yeah, yeah, so I don't get on the bike any, anymore at all for, for any reason. But, uh, Yeah, no, it's, it's great fun.

But isn't it fascinating also how there's like so few accidents in the midst of chaos? When you think about it, like, there's so much cars, right? And so much chaos. If you've ever driven in like developing countries like Mexico and so far, and how few accidents there is in the midst of all that chaos. Like how there's like two human brains that adapt with one another and sort of like avoid one another so that we don't like mutually die. Like, isn't that fascinating as well?

It's like an unwritten rule there, isn't there, that, you know, that, um, people just seem to find their way. I've spent a bit of time in Thailand and Vietnam and some of these, these sorts of countries which are very, very similar. And, um, yeah, you don't see many accidents. You see lots of, uh, fender benders, you know, there's lots of rubbing up against one another and bumping into each other, but there's nothing serious. But Charles, I think a lot of that has to do with the speeds that they're traveling, uh, because they're so congested, right? So again, if we pull it back to the, the product that we, we have there, the X-Vision AI, we're out there also, one, to save lives, but number two is we, we want to solve that congestion problem and help people get around the city faster.

So I remember being in China, and the most crazy I've seen was like 5 vehicles trying to get into one lane, one bike, one mototaxi, one motorcycle, and two cars. And I was like, dude, I'm getting stressed out for my driver. And he was just like, ah, it's nothing. No hand signal and just drive aggressively so that he's sure to get a spot, which I think is the technique when you're driving these environments. Like, take your fucking place, right? If you're being a wuss about it, people are definitely gonna take advantage of you. Um, how would your AI make sense of all of that chaos?

Yeah, that's a really great question. So we are looking at the velocity and the trajectory of all of the vehicles, so we're determining behavior based on those, those inputs. So while it looks chaotic, you know, to somebody who doesn't understand the velocity and trajectory of all of those, uh, you know, different vehicles coming in, to something like the AI, this is actually a pretty simple problem, right? The AI calculates it all, it works it all out, and it, you know, it puts it all into perspective. So definitely, yes, it's scary to the unknown, but once you unravel it and, you know, put some math and some science around it, it all makes sense, right?

And how does that intertwine with law enforcement? Let's say that there's a bunch of people breaking rules, going on a red light, or just like cutting another guy, it's not their turn, right? Like, uh, how do you make sure that these folks— I mean, punishment's still a good strategy to enforce behavior, right? Like, would you agree with that? And two, how do you, uh, intertwine in these systems and making sure that law enforcement is applied?

Yeah, so definitely there's a law enforcement angle, and, uh, you know, the speed and red light fines, uh, some of the obvious examples. But I really feel that changing behaviour of drivers is the goal here. It's not about hitting people with a big stick. It's about helping them do better, perform better, or act better on those roads. So for example, in places where you have a large number of people crossing and not everybody can cross the road at the same speed, you might have some people with mobility aids or some some elderly people, what we can do is watch that road, watch those people crossing, and hold the red light for longer. Rather than basing this on a timing system, we hold the red light longer so that, you know, the vehicles don't actually encroach and enter that space until everybody has safely crossed the road. So again, you know, you're talking earlier about that chaos, you know, the X-Vision AI is really about untwining that chaos or simplifying that chaos and just applying some simple changes to people's behavior to, to make it safer and less chaotic for all.

And what about integrating with various apps, let's say a Google Maps, a Waze, a Uber, to coordinate traffic, right? Or even with an ambulance or a fire truck that needs to pass. Like, any thoughts on this?

Yeah, exactly. So, um, the integration into Google Maps and those sorts of things is not really on our roadmap. It's not really our, our wheelhouse. So, so that's cool. There's You know, there's a horses for courses, I think, I think is the saying. But definitely giving priority vehicles prioritisation is something that's important. So when we detect a flashing light approaching, so from an ambulance or a police car, we're able to do things as I, as I was just talking about, like tell the signal controller to give priority to this particular vehicle. So hold the red on the other lanes, uh, until the, you know, till the red light, uh, till the, uh, flashing light has, has come through.

Um, so, so you spot the truck from a distance and then you give it the, the access. So how far can it see, the, the hardware?

Yeah, the hardware can see 50 to 100 meters depending on the configuration. So there's a few different lenses. Yeah, so it's a lot, it's a lot. Um, so, um, yeah, we can spot it out a long way. We can make that change on the control of the signal controller. We can hold the other lanes red or even change them in time to allow the priority vehicle to come through. And then, once it's been through, then re-look at the congestion and what's happened with this disruption and then make some assessments on what needs to happen then to reduce that congestion as quickly as possible.

Is there also sound data coming in? Do you analyze some of that or just vision?

Yeah, now we've taken a bit of a Tesla approach. We've gone vision only. So there's some advantages to that, there are also some trade-offs, but we made an engineering decision that we wanted to focus on getting one technology working the absolute best that we can rather than trying to, you know, bring in a whole heap of different sensing elements. Has some challenges in itself.

Interesting. And is there a component in which it can contact the authorities if there's an accident or such an event?

Yes, that's one of the base use cases. So yeah, definitely. If we detect a crash in a roadway, you know, we send an email or an SMS to the control center to let them know what's going on.

Um, yeah, baby pass, uh, there. What are the next steps for you, uh, Simon? Tell me a bit more about your goals.

Yeah, so our goals are, uh, we're currently in a, in a raise. We're raising $10 million to achieve the certification that we were talking about. So we're really trying to get all of this evidence, this extreme validation in place upfront. So as you mentioned earlier, that's a fairly slow and costly process. When I say slow, it's, you know, a year or two. The next step for us is to get to 1,000 intersections. So at the moment we're about 100, getting close to 200 intersections now. We're purposely trying to keep small because we want to keep our risk profile low before we achieve certification. Once we achieve that certification, the, the first milestone for us is to get to 1,000 intersections. To give you some perspective, there's about 100,000 intersections across APAC that are ready for the technology, and across US is about 300,000 intersections that are ready. These are intersections that are, you know, ready to have to have the system bolted on today. If you sort of expand it out and and look at some of these other edge cases. There's, there's a lot more, you know, the market's much, much bigger.

And what are the readiness criteria? What would make an intersection ready versus one that is not?

Yeah, so it's already got, uh, signal controllers there. That's, that's an important piece. We need to have an existing set of traffic lights. Um, we need to have a clear line of sight. So there are a number of intersections, um, where there isn't a clear line of sight. There's, uh, you know, things like bridges in the way or other, other objects in the way. So So they're put on the— I don't want to say too hard basket, but they're put in a more advanced basket and they need some additional thinking about. So yeah, like I say, the criteria really is that it's signalized and it doesn't have the AI already built in from another provider.

Sales, on the sales side of things, what's the challenge to sell to governments or municipalities?

Yeah, so sales side of things, it's, it's actually quite, um, I don't want to say easy, it's the wrong word, but for us it's quite a straightforward sales process. The certification means that we get built in as standard specification to an intersection. So once we're a standard specification, once we're a standard part of that, um, that build We just get written into the specifications. So then anybody that is bidding on building that, that roadway or building that intersection, they have to come to us to, to buy the product, you know, and implement it.

Tell me a bit more about, um, yeah, launching the company in itself as well. Like, uh, what's your main motivation behind that one and how challenging it's been for other founders listening.

Yeah, look, uh, business is, is hard. Um, actually, I, I've been fortunate to work as an engineer all around the world. I got to go and solve some massive problems, uh, you know, in, in the Netherlands and massive problems in Dubai and some of these sorts of, some of these sorts of countries. And, um, actually coming back, um, you know, to starting a company was really hard. The technology is, you know, not so difficult. It's, you know, being an engineer, it's straightforward. You just solve one problem after the other. But business isn't a straight line. There's a whole heap of different moving parts. And, you know, you have to— what'd I say? You really have to learn a lot more about a lot more different things to start a business, you know. For me, things like marketing is hard, right? You know, you put an engineering problem in front of me and it's simple. I'm sure there's others out there that marketing is something good. And that's, that's the other thing that I've learned. You have to find the right people to put into the right, uh, the right space because, you know, that gives them the opportunity to excel and, and lets you multiply yourself, I guess, right?

Yeah, it's like pick your poison, right? Like when you're, uh Working for someone else, you just follow the marching orders, you get paid anyways most of the time, but you do expose yourself to getting fired, right? Like, um, and when you're a founder, you do have like full autonomy. It's nice, you have freedom, but, uh, you face like all these different challenges and these fires. So guess it fits some personalities, uh, better than others. But yeah, at the end of the day, you're building like life-saving product, uh, I think that's a pretty cool thing. It's a pretty strong motivator. Like me doing raising capital as a service, right? I'm not like directly saving lives. Uh, so, so how does that change your approach to things? Does that motivate you in the dark times?

Um, look, I'll tell you what motivates me the most in dark times. It's, it's the people around me. So we've built like a little family here, um, you know, to around this product. And the dark times, what really sees me— sees the— what helps me see the light is knowing that if I don't perform, if I don't steer the company in the right direction, these guys aren't going to have a job tomorrow, you know. So it's, it's really about making sure that everybody is looked after. It's, uh, you know, and not only, um, our direct employees, we've got a lot of indirect employees, right? We've got guys that help us with R&D, we've got guys that help us with government grants. We've got, you know, a whole extended business, and if, if we don't perform, uh, you know, a lot of these guys suffer. So that's kind of my motivation.

Yeah, thanks for being on, Simon. Where can people find out more about you and Xvision?

Yeah, they can go to the website, I guess, is the best spot. Xvision AI, so it's xvision-ai.com, or you can reach out to me on LinkedIn, follow me on LinkedIn. Simon Maselli, S-I-M-O-N M-A-S-E-L-L-I. Uh, yeah, look forward to catching up with you and your listeners, Charles.