A Radiant Ride: Exploring Hyperspectral Technology and Founder Trials with Francis Doumet, CEO of Metaspectral

3, 2, 1.

All right. Welcome back to another episode of Founder Journeys. My name is Ray Walia, CEO, founder of Launch Academy, here with Francis Dumit from Metaspectral. Francis has been a OG Launch Academy member. I think you hear me a lot these years talking about OG members just because we have so many. But Francis is one that's really an og. He's there for quite a few years building multiple companies. His office was right next to mine. We got to see each other every day. A lot of respect for Francis and the teams that he's built over the years and the companies that he's built over the years. The latest one is Metaspectral. So, Francis, welcome to Founder Journeys. Please tell us, what is Metaspectral?

Yeah, thank you for having me, Ray. So Metaspectral's mission is to solve the world's most complex problems using next generation sensing. And so one example of that is hyperspectral imaging. These are sensors that gather up to 100 times more information than a conventional camera. And so for that reason, they're about 100 times more complex to handle, especially when it comes to data transmission, but also data analysis. So we've built technology, a combination of hardware and software that allows us to analyze this hyperspectral data in milliseconds instead of hours.

So very technical, very forward type of platform. I've obviously been following you guys for a while and I've seen some of the applications. One of the ones, so you're working with NASA and you're working a bunch of stuff in space, but the one that really got me excited was the recycling. And those that are familiar, when you're looking at recycling, like differentiating plastic and all this stuff, it's very manual, it's very time consuming, and it's not perfect. But with Metaspectral and the image analysis that you're able to provide it, it's like a 10x improvement, if not more.

Yeah, so recycling is a very interesting application because there are players that have been in that industry for decades, and there's maybe two schools of thought within those technology providers. One of them is using infrared technology, which is maybe a subset of hyperspectral. The other one is using conventional cameras. So computer vision as we know it today.

And just to step back. So what is it that the cameras are doing in that recycling process for those that aren't familiar?

Yeah, so they're processing essentially consumer waste moving on a conveyor belt at 1 to 2ft a second, and they're sorting it automatically. Using air injection or even robots in some cases.

And so the cameras are helping the robots or the air injectors identify when to push something off or push something into a different bucket or whatnot. Right. How to separate the recyclables.

Yeah, that's right. And then as they're doing this sorting, they're also. They can also provide insights into how good or bad the quality of your production is or how pure it is. Right. Since they're. They're also looking at things moving on the belt. Now, the technology providers that are using infrared, they just use a handful of bands or a broadband infrared. And it just tells you that this one object is of material type X. For example.

This is a plastic milk jug or milk container.

A plastic milk jug or PET Coke bottle. On the other hand, you have the other guys that are using conventional cameras and they're looking at the shape to essentially determine the material of an object. So they say this bottle looks like a Coke bottle. I've been trained on Coke bottles that are all pet. So I'm going to say that this bottle looks like a Coke bottle and it's going to be pet. But that's not always accurate. Manufacturers often change materials in packaging. It could be a change in the bottle polymer, or it could be the type of the label changes, or even the bottle cap might change. So this is where we come in. So instead of giving you an inaccurate determination of what the material is, we go pixel by pixel and we tell you exactly, pixel by pixel, what the material is on an object. And so remember, the guys using infrared just look at the entire object as a whole and tell you this one object is a milk jug and it's htp. But we take it to the next level and tell you that this is a milk jug. Yes, but it also has a label that is maybe polypropylene. And the bottle cap is actually a third material, like hdpe. And so this allows our customers to define very complex sorting rules downstream. For example, they're able to define things, find rules, like if they encounter a PET bottle that is wrapped with a label that is covering more than 90% of its surface area, then let's remove that bottle from the stream, because that label is going to be very hard to remove mechanically. And once we shred it, it's going to contaminate the stream because the label is made of something else than the bottle's body. So that's one key differentiator for us as we do material identification pixel by pixel. The second major differentiator is that we're able to detect and identify materials and differentiate between materials that are otherwise very hard to detect. We're talking about different types of transparent objects, different types of black plastics. Those are all things that look the same to a human sorter. They all look the same to a conventional camera, like a conventional RNGB camera. And they're also very hard for a conventional infrared system to detect because they're transparent. They don't bring back a lot of signals. So we've made significant advances in neural network architectures that allow us to essentially spectrally unmix or untangle signals and determine that there's a little transparent object sitting on top of a conveyor belt that has more signal. And so we use AI to essentially find these very faint signatures that give us an idea of what may be there and what may have otherwise gone.

So you guys must have a pretty deep database of all this information to benchmark against.

Yeah, but the one advantage of our approach, though, Ray, is that a single data point for us is a pixel, whereas the conventional AI, conventional computer vision, a single data point is an entire image. Right. So an object may have 100 pixels. And so that means that we can get away with maybe 100 times less data to train our models, but with.

That much more accuracy. Because you're going right down to the pixel.

Right.

That's so cool. And so I can geek out on recycling all day here, but that's just one sector that you're able to apply your technology to. We briefly mentioned NASA. Canadian company doing work in space. Those are always things to celebrate. So tell us, what is Metaspectral doing in the space race?

Yeah, space race. It's cool to hear that. And so it's been like a dream of mine to do stuff in space. I should just mention that we're a subcontractor to an American contractor who has the contract with NASA.

Semantics. Come on.

Yeah, it's just the mechanics.

No, we do want to clarify that. We don't want to put any false information. Information out there, but just the hoops.

We had to jump through to get that opportunity. But we are putting a payload together. It's a hyperspectral imaging payload that we're attaching to the International Space Station. It's going to be up there for six months. And our mission for NASA is to really demonstrate how we can actually derive results in real time from hyperspectral imagery. Again, it usually takes hours, days, weeks to process this imagery. And we're going to be demonstrating how we can do it within less than 15 minutes.

When we're talking about, well, what images specifically are we talking about? Is it image of the stars, images back Earth or.

Yeah, no, yeah, it's images of the Earth looking back at Earth. Yeah. So the, the payload is going to be pointed at, at nadir, which is essentially down, down to the Earth and we'll be looking at things. It doesn't. The mission is not so much about what we're looking at, but it's more about how fast, proving the technology and.

Showing the processing speed. Yeah.

How fast can we implement the data pipeline? But yeah, our partner with this project is High Speed Computing out of Florida and their expertise is in the marine environment. So they're going to be doing a lot of analysis with algae and, and things like that.

Again, very important subject matters for us to be as knowledgeable as possible. Moving into the future with the climate change and the changes in the planet, this is a really cool sector that you're in. How did you get into this? Like what is France's founder journey like? How did you get down this path to Metaspectral?

Yeah, well, when I was first.

Well.

When I was getting ready to go on the next chapter of my life, I guess I met my co founder Miguel, who was fresh off the Press with his PhD in AI and had this cool idea of how to better compress data using AI, because any of the algorithms that we know of, like JPEG and MPEG for video, they basically use the same hard coded formula for every image. And so Miguel's insight was, well, why can't we tailor this compression formula to every input so we get a much more performant algorithm depending on whether the image is a landscape or a portrait, for example. So we did that and we ended up developing an AI powered version of JPEG that we were trying to market as ways to accelerate web and accelerate mobile applications. But around that time our first customer happened to be in the agricultural space and they were doing a lot with infrared technology to measure and monitor how plants react to different types of pesticides. And so that kind of opened up our eyes to that whole market and we really found that the problem was an order of magnitude more important than for infrared imaging than it was for conventional imagery like JPEGs and whatnot. And so we made a 180 degree shift to that industry. And then gradually we started going up the value chain and doing more on the application side. So now data compression is one piece of the puzzle that allows us to do what we're doing. So we're all about real time analysis of hyperspectral imaging. But then we've also made significant advances on the AI side of data processing that allow us to do what we do.

Hello fellow tech startup founders. I just want to quickly interrupt this episode to tell you about us at Launch. If you're looking for a community to help you take your business to the next level, consider joining our LaunchPad program. As a member, you'll get access to investor connections, programming workshops, mentorship, and over $400,000 worth of perks. Also, don't forget to check out our other podcasts, Launch, AMA and Bits and Bytes for expert advice and stories from the tech community in Vancouver and around the world. Visit LaunchAcademy CA to learn more and start building your dream business today. That's LaunchAcademy CA.

Let me take a moment to shout out our longtime sponsor, Smythe. They are a leading independent VC based accounting firm specialized in providing free financial services and consulting for tech companies across North America. Smythe has supported our program and our alumni throughout their early growth stages by helping them structure and set up their businesses all the way to helping more established businesses with cross border operations and M and A. They combine industry knowledge with a proactive collaborative approach, empowering you to make more informed decisions as your business grows. So if you're looking for a trusted partner to help you drive your business to success, reach out to one of our Launch Academy's longtime mentors, Camelia Ho. For more information about how Smythe can support your growth, you can find Chamele's information on Smyth's website at smythcpa.com that's s m y t h e cpa.com along with more information about the various industries they support and the services they provide. So it was that initial compression technology that got you down this path. And for our listeners that are geeks like me and remember Silicon Valley, they TV series, Pied Piper, data compression, JPEGs. You were known as the Pied Piper of the real world and that company is called Compression AI.

That's right.

Yeah, yeah. Obviously I know this because they were at Launch Academy and that was also, it was fun, but it was also cool. Like no, we have a real life Pied Piper building that technology out and it's actually doing great things. And obviously that's morphed into so much more with Metaspectral. But as you quickly alluded to, like that's when you met Miguel, your co founder. So you got a team that's been built here that's two of you as co founders, but your team's like what, 12, 15 people now? Yeah, but you said that was an when you're ready to start the next part of your life. Prior to that you were a solo founder. So maybe let's dive into what were you doing as a solo founder and what are the stark differences of being a solo founder to a co founder.

Yeah, yeah, I started my previous company all by myself and I guess I did so at the time because I figured that I had all the skills required for that company. It was a mobile first company who were building next generation push notifications for mobile operators and it went well for a few years. The solo founder. It was challenging being a solo founder because all startups go through ups and downs and I think as a solo founder, those ups and downs are exaggerated when you're a solo founder because yeah.

You have you in an echo chamber and it's just going through your head over and over and multiplying.

Yeah. So the highs are way higher and the lows are way lower because there's no one to ground you when there's, when you have a high and there's no one to, you know, console you, I guess when, when things hit. Hit the fan. So, so yeah, you know, it's. Yeah. And then there's on the, when, when you go raising funds as well, there's. You also have to explain to investors why you're a solo founder. No one likes that. So I would definitely recommend being. Yeah.

One of the biggest challenges that comes up from a investor's perspective of solo founders is the question of people automatically think it's, oh, it's the workload. It's not so much the workload. It's like, okay, for you as an individual, why have you not been able to attract other people around you or be able to join your cause and really chime in on this because that shows a sign of validation. If you get a third party to come in and drop everything they're doing to join your journey and path, well, that's some validation that other people are believing in this and it mitigates risk. But again, as a solo founder, sometimes you want that autonomy. You want to be able to just do it on your own. You don't. You think you can move faster on your own. But now, having lived both sides of the table, what is the, the advantages of having a co founder and where are the challenges?

Yeah, well, the advantages are first and foremost for founders sanity, I think, like mental health, like having just someone there.

To.

Talk about the ups and downs, that helps a lot. And then obviously for just strategy, business strategy, bouncing ideas off one another, you both want the same thing at the company and so you're both invested in or you both have a material vested interest in the company. And so yeah, there's always the opportunity to have a dissenting opinion. For better or for worse. Right. Whenever you want it, which helps a lot. And then ideally your co founder has complementary skills to you, otherwise it probably wouldn't be a good relationship. Like in a marriage, I should say. So.

Yeah.

Having a partner, having essentially a life partner that has complementary skills and with, with which you're going to invest about 10 years in your life, you know, in this, in this, in this journey is something very valuable.

And then flip side again, having been a solo founder and now in a co founder situation, what are some of the things that you miss about being a solo founder? Obviously speed, like being able to move faster. You don't need to double check with somebody. But as we just said, sometimes double check with somebody is, is that grounding that you need? But what are some of the things that you were able to do as a solo founder that aren't as easy in a co founder relationship?

I mean things at the strategy level, direction of the company, what to build, even hiring. Right. So being a solo founder is like being a dictator. You can do whatever you want without asking anybody. But having a co founder is probably a much more sane relationship. If you have a board in your company, then that's how every company works. You tend to make major strategy decisions through your board. If you don't have a board yet, the board is basically you and your co founder, even if it's informal. And like any democracy, it slows things down, but for the better. Usually.

Cool. Now moving off of actual co founder relationships, just the journey of building a startup on its own. You've built multiple companies now and all various levels of success and I've seen you go through the ups and downs. What are some of the tactics that you've developed over the years to help you through those ups and downs for you personally?

Yeah, I would say it's very important to maintain a routine as a founder because it's very easy to, very easy to slip and get too invested into the topic of the day. And that could be a positive topic or a negative topic. It could be a new contract you have to get out the door or it could be a contract you lost. But if as a founder you have a routine, by routine I don't necessarily mean a 9 to 5 job, but at least if you have certain things you want to get done every day, you know, you want to work out 30 minutes a day, or you want to, you want to, you know, spend half an hour reading a book or, you know, just things that, you know, keep you sane is very important to have as a high priority. It kind of instills a sort of, sort of discipline that helps you as a founder not, not dedicate 100% to your startup. And that is a good thing. You, you, you as a founder, you want to give yourself to your startup, but after a certain, beyond a certain level, you're just doing damage to yourself. So working nights and weekends is totally fine as long as you keep a few things in your routine that keep you sane.

And as a seasoned serial entrepreneur, what are some of the things that you wish more entrepreneurs talked about openly back when you were just starting out? Like, this is our chance. This is one of the reasons we have Founder Journeys, is we want to talk about those, those taboo topics that a lot of entrepreneurs don't talk about. But we as entrepreneurs wish, like, why did people not talk about that more openly when I was starting out? Yeah, because it's that reality. Like, you realize you are not on your island by yourself. You, there are other entrepreneurs going through the same challenges and problems that you are. Like, what are some of the challenges and problems that you wish people talked about more openly?

Yeah, so the, the reality of startup life, like when I started my first company, you know, it was, it was like a, it was like a. What's the word? You know, I thought I was like, going into like a, like an El Dorado. Like, I'm going to be living, you know, the nice life. Because that's just what I read on TechCrunch. Like, you build a company and three months later you get funded. And then, you know, fast forward five years, you sell your company. Right. That's just the paint. That's just the picture TechCrunch paints. Right. But the reality is not that. It's far from that. I thought when I started my first company that, oh, it took two years to generate revenue. That's a big deal. But after you talk to people and. No, that's actually good. If you generate revenue after two years. And it takes some companies five, six, seven years to start generating revenue depending on what they're building. And so there is no, first of all, there is no, you know, rules set in stone with certain metrics that you have to do X by Y. Time period. One thing that I did appreciate a lot when I was doing my first startup was here in Vancouver. I forget who was organizing it, but it was something along the lines of called something like fuck up nights or something like that. I forget who was organizing it, but I remember a session very vividly by.

I think it was J. Lors Mann that set those up.

Maybe.

Yeah, okay.

Yeah, I remember a session, I think it was Jay. Jay from Picotick. Yeah. And his, he's a great guy and he great, great talker. But yeah, he got real really quickly and I didn't realize how, how many times like Pic almost blew up and he just got back on his feet and every time he would just kind of recover. And then now there's a success story having sold to Eventbrite. But I think we need more of these stories to keep it real because otherwise first time founders especially think they need to abide by certain metrics and KPIs and if they don't reach them, they start thinking they might doing something wrong. Maybe thinking that they might not be.

Worthy of the imposter syndrome starts creeping in like, okay, what am I actually doing here? Why am I getting up every morning dedicating all my time and energy and my money to this fleeting idea that I have. But one of the things I always said at Launch Academy is misery loves company. And the reason we try to make sure people are surrounding themselves with fellow entrepreneurs, especially ones kind of going through the same challenges and problems that you are, because that's the best person to learn from. Bringing in a mentor is great, but having a mentor that did something 10 years ago is very different conversation and very different type of education than you get from somebody that is a fellow entrepreneur that just went through those same challenges six months ago. Because it's more timely, but it's also somebody that you can connect with right away or even sitting side by side with people that are banging their head against the wall trying to figure out how to get revenue or how to get product market fit just as you are.

Yeah, exactly.

So how do you outside of your team spend time connecting with fellow entrepreneurs?

It's been harder since the Pandemic, honestly. Launch Academy was a great space when we were there. There are always people building new stuff and we would go out on lunches and meet people and build relationships that way. Since the Pandemic, we've all been working from home and beyond our circle, which are the co founders and then our immediate investor circle. It's been really hard to meet other Founders. We did go through a couple acceleration programs and we, we did meet people that way. One of our current investors hosts periodic founder retreats and that has been good to meet people. And we have also WhatsApp groups and Slack groups that help us communicate and kind of exchange ideas and questions when they arise.

Yeah, it's that human connection that really makes a difference though, right? Like talking to somebody via text or Zoom is great, but when you meet somebody in person, those founder retreats I find are. It seems like it's. Oh, it's just a gimmick by the VCs. It's just them kind of giving back to their portfolio. But once people go and they come back, they're like, wow, that was actually really good. Like getting to kind of sit down with fellow entrepreneurs in this environment and having our VCs there, showing us the resources that have. That's a great experience. Maybe just share what that experience looks like for the people that haven't had that privilege yet.

Yeah. So yeah, the VC kind of brings together their founders once a year. It's not always the same founders, at least in our case because they're just too many, but I think they send invitations to those who haven't been in a while and they show up and they spend usually two or three days, I think two days, two days off site and there's, there's programming, getting to know one another through certain exercises. Like for example, like you break out into small groups and each one kind of, you know, gets really personal with the other people.

Like a mini fuck up night.

Yeah.

What have I fucked up on this week?

Yeah, one thing that, or one approach, I guess that was really great was going around the circle and each one would tell moments, defining moments in their lives that define them as a person. So it could be things anywhere from your childhood all the way up to now. But things and events define who you are as a person and that really helps getting to know your fellow founders. And sometimes it gives you insights into how they're, what they're building today and what motivates them, what is their passion. Then there are also breakouts specific to topics that may be of importance to you now. So if you care more about fundraising, there's a special fundraising breakout with some special speaker or how to build a. How to build culture was an important one in this day and age. Post pandemic. How do we do that remotely? Something that was really important. But also it gives you an opportunity to talk to your peers who are in that similar situation and kind of brainstorm ideas.

Yeah. So all in all, very valuable experience for you and your team. Was it the team or just you or you and your co founder who gets invited to these?

Usually founders. I think they try to do it once. One founder per year. Yeah.

And so you may not be there next year, but there's an opportunity for Miguel to go there, or vice versa.

Right. Yeah.

This has been a really insightful conversation. I'm going to drop off with the two last questions. One is, what can we expect from Metaspectral and in the near future? And then the last one is some parting advice that you'd like to give to the audience as they go down their own founder journey.

Yeah. So Meta Spectral, we were. So there's one market event that is really exciting that is happening, and that is that there are a handful of companies, maybe five or six different companies, that are putting commercially available hyperspectral sensors in orbit as we speak. So this time next year, there's going to be quite a large supply of hyperspectral imagery for us to start consuming. And historically it's been really hard to analyze this data. Everyone that we've wanted to use hyperspectral data have told us that they don't have the expertise, they don't have a PhD on staff. So Metaspectral is really positioning itself as being the go to software for analyzing this kind of data using AI. We're really trying to make hyperspectral data more accessible. So in the near future, hopefully you'll see more applications of hyperspectral imagery from space using models built and trained through benettospectral software.

Super cool. The data just keeps coming. And what you can do with that data is going to be very transforming.

Right. Especially since the software that we're building, we've designed it to make it easy for anyone with or without a technical background to train these AI models. And so hopefully you guys even can start maybe training new new models to detect things you never thought.

That's actually third question I'll throw out there. So who has surprised you? Who's somebody that came along or you see using your technology that you're like, oh, I never would have imagined somebody using our tech in that way.

Yeah, the Navy. The Navy, yeah. They're looking to use our technology to look at paint and antifouling on Navy vessels, because if that coating is misapplied, if there's pockets of air inside, then that's a recipe for disaster. It's going to induce corrosion, and corrosion is like cancer, it spreads super quickly. So they want to be able to detect these inconsistencies at the time that the paint's being applicated so they can fix it before corrosion starts.

While the ships are in dry dock getting repainted, they're doing a spectral analysis before they are put back in the ocean.

Yep, exactly.

That's cool. Yeah, it's awesome. And okay, now your parting advice for our founder. Fledgling founders on the podcast.

The most important thing is to work on something that you're passionate about. Because. Yeah, I mean, starting a company is not easy and working on something that, that you're passionate about will essentially help you or make you keep going during those, those hard moments.

Yeah, and there will be a lot of them, a lot of ups and downs. But the good thing is there are a lot of ups too.

Right? Right. Absolutely.

Awesome. It's been an awesome conversation. Thank you for joining us on Founder Journey and we look forward to forward to seeing all the amazing things coming from Metaspectral.

Thanks for having me, Ray.

Hey there, fellow tech startup founders. This episode of Founder Journey was brought to you by Launch Academy. If you're looking for a community that can help you take your business to the next level, look no further than our Launchpad program. As a Launchpad member, you'll have access to everything you need to succeed, including investor connections, year round programming, workshops, offline socials, mentorship, office hours, and over $400,000 worth of perks. Whether you're just starting out or looking to scale, we have the resources and support you need to achieve your goals. But that's not all. At Launch Academy, we also have two amazing podcasts that you can check out. Launch AMA and Bits and Bytes. At Launch ama, we bring in industry experts to answer your burning questions about entrepreneurship, fundraising, marketing, and more. It's a great way to stay on top of the latest trends and learn from those who have been there before. And on Bits and Bytes, we highlight the tech community of Vancouver, sharing stories of innovation and success from people who are driving the industry forward. It's a great way to stay informed about the local tech scene and connect with other like minded individuals. So why wait? Join Launchpad today and start building the business of your dreams. Visit LaunchAcademy CA to learn more about our programs and how we can help you achieve your your goals. And don't forget to tune in to Launch AMA and Bits and Bytes for even more great content. Visit LaunchAcademy Cat.