13:35
Hello and welcome back to Equity TechCrunch's flagship podcast about the business of startups. I'm Kirsten Korosec, transportation editor here at TechCrunch, and this is the episode where we bring on industry experts to help us explore a trend in the tech world and dive deep. And today we're joined by Al Cohen, the founder and CEO of Humble Robotics. Al, welcome. Thanks for having me, Kirsten. Yeah, I'm excited to have you back on the show. We have spoken before, but you've never been on Equity. To give the audience just a sense, Humble Robotics is the startup that came out of stealth in April with about $24 million of
funding, I believe led by Eclipse. And what you're doing is developing a fully autonomous cabless electric hauler to transport freight. And I do want to talk about Humble, but for folks who don't really know you, I mean, you have this incredible background of working at a number of startups, starting your own in the Bay Area, all around autonomous vehicles, robotics. I feel like you know where all the skeletons are buried when it comes to autonomous vehicles in the early days, or maybe have some good stories to tell. Yeah, you know, so I've had actually like a very lucky career, I think. If you think about kind of the companies I've worked at, it's been a little bit of a blessing, but just as a quick
background for me, you know, I've been in the Bay Area now 20 years, all of it working on kind of these new startups, deep tech, physical AI, whatever the sort of like the newest way to call it is, but I have had quite a number of experiences, I think. I think I'm up to seven startups, maybe eight startups in my life. All at different stages and phases. You know, I've seen industries rise, I've seen industries fall, you know, it's like it's been quite an awesome journey. Um quite lucky to see it all, but uh yeah, we've seen a lot. I want to kind of talk a little bit about this. I kind of joke a lot in my newsletter and um actually with friends about how we're you know, back in 2016 all over again. And it feels very
reminiscent of that time, but that time in particular, I would say between 2014 and say 2019, you were at Apple Special Projects, you were at Otto, which was the self-driving trucks company that was then bought by Uber ATG. Then you went to Anthony Levandowski's autonomous vehicle technology startup Pronto. And give folks a sense who weren't in the Bay Area at that time and weren't involved in startups, like what was that time like in the industry and just being a part of startups? So, I'll actually go back even further, like I'll talk about even like 10 years before that cuz it parallels a little bit the story. So, so I got here in 2007 to the Bay Area, and at the time, you know, the industry that was most
exciting and interesting for an engineer was electrification, right? It was electric vehicles of all kinds. You had Tesla at that time being a very small company, um you know, barely keeping the lights on, but the whole Bay Area was kind of a wash in companies trying to do electrification, right? And they were developing motors and batteries and motor controllers and all this kind of cool tech. And that rose very quickly, right? And then uh today a lot of tech you know, is outside the Bay Area actually. Although there's still like a really healthy DNA for it. And so you see this kind of like you see these companies this sort of like hype build around a concept, but it's it's not just like it's not just hype, it's actually
like a lot of people just like putting all their brains toward something that's exciting and you create a little bit of a moment and a movement. And so you know, I did the electrification thing for 10 years. And then when you get to the 2016, 14-16 era that you're talking about, what kind of started to a lot of people to come together was autonomous vehicles and a lot of that came out of like you know the DARPA Grand Challenge this big this competition where you know people showed that autonomous vehicles were possible for the first time. And it created this moment kind of frothy a little bit like where people were like really excited about autonomous vehicles. A lot of companies and dollars
went into it and then it kind of rose and then kind of had like a little bit of a moment where people were like wait is this technology going to work? But the arc of the technology sometimes is longer than people's attention span, right? So it's like if you think about the electrification from 2008 you know today electric vehicles are kind of ubiquitous but that also had kind of like a rise and then a fall at a moment. And autonomous vehicles were kind of the same way. So it was a very early industry when I started in it. The technology there wasn't any real core technology that had been developed that was like really ubiquitous yet. We saw these companies start and a lot of us just continued working on it for the
next 10 years and like self-included. And now again in 2026 it's starting to have a moment again. But I just think of it more like the arc of the technology was just like a 10 15 year arc and you know our attention spans kind of come and go on that sometimes but that's been the story. I want to talk about that attention span because there were startup founders that have persisted. I mean some of them their companies imploded or you know got acquired you know maybe had a good exit but essentially they weren't like working on that startup anymore but they've sort of stuck to it specifically with autonomous vehicles. So when you look back I mean who was having the short attention span? Was the problem
more with the investor community that they were sort of chasing whatever that next new thing was. we crypto had a moment now AI labs have a moment Um, that the founders have sort of been at it all along, or actually was it some of the builders that, you know, had a bit of a short attention span? No, the core industry is it's actually like a fairly small industry, and a lot of the people that have been in it since I started are still in it, right? It just it becomes a little bit of a grind sometimes, right? It's like you just keep working on the problem and doing the next thing, like the cameras aren't good enough. Get the cameras better.
Okay, the cameras got better. The autonomous technology, the brain of it, you know, wasn't quite up to par. Okay, let's grind on that, get it better, right? And so, I actually think the builders, the founders, and a lot of the industry people, for the most part, have stayed in it. And they've stayed in it and uh just kept going. You know, investors, they invested at the beginning, right? And then they invested sort of in like the 2016 time frame. But once they have made their investment, it's really up to the founders to kind of take that investment and carry it forward, right? So, you know, I don't think it's a problem that the investors kind of go to the next, you know, the next technology they're trying
to kickstart, because their job is to, you know, see these ideas, like an a founder has an idea, their job is to kind of pair that with capital. Then it's up to the builders to really make the make a thing of that. It's just that sometimes the expectations around technology, especially if you're thinking about something like software, right? The expectations are that materializes in three, five, seven years, something like that. It's possible, you know, exits for the company. For some of this harder technology, you know, sometimes it could be 15 years, 20 years, 25, right? It could take a very long time to get the technology to fruition. But then when it does, it felt obvious all along, right?
Yeah, totally. I mean, I think in terms of the hype cycle with autonomous vehicles, which was a very real thing in 2016. I mean, you had companies like Cruise getting acquired for, you know, a billion dollars. But I think what was fueling that at the time was really automakers and other companies like seeing something really magical and wanting to be a part of it and then just really throwing huge amounts of money at it and then of course facing with the realities of like scaling autonomous vehicle technology is maybe a little bit harder and longer than expected. Would you say that there are other industries that you've been a part of outside of autonomous vehicles that have sort of that have mirrored that
same experience where you know large OEMs and manufacturers sort of jump in and fuel hype and then you know sort of pull out and you're left with a little bit of that trough of disillusionment moment. I think it happens to almost every industry you know and I think every new industry kind of goes through that cycle like you know we're we're all human and I think we get excited about especially when you have a magical experience right like going into an autonomous vehicle for the first time is a magical experience right you're like whoa this vehicle is driving itself and you as a human generally know how complex that can be and everything
associated with that and you're seeing it happen and you know it's not surprising that people sort of like ran towards it. I've actually had it happen in industries that I've been part of. I think the three industries I have participated in are electrification solar autonomous vehicles and I think it's happened to all three right basically large investment followed by like wait can this work followed by a wait it does work. Mhm. But the moment where it does work sometimes can take yeah 5 10 years. I think it'll happen to literally every industry right it's like people get excited people are excitable. And I think one thing that does happen right
is that you know some people crowd to a topic because it is exciting and then as soon as the excitement wears off they go to the next thing right that's also fairly human nature but you're always left with this core group of people that just stick with the problem right they're like I'm captured by this and I want to drive it as far as it can possibly go and see it through to the possible end. Yeah. Yeah, I would say the joke is so your joke is that it's 2016 all over again. The joke at Hummel has been it's 2016 all over again. But part of that is because we have a lot of the same people that we worked with 10 years ago at the
at my company but broadly speaking in the industry that have just kept at it for 10 years and are now starting to see the fruits of that labor. Or actually in the case, you know, correct me if I'm wrong but I believe one of your co-founders to your point of like it being a grind like was part of autonomous vehicles, took a bit of a break, now is back you know, participating also in Hummel. And I do want to talk about Hummel robotics because the cabless piece of it is really interesting. To paint a picture in folks' heads like this is, you know, when you're looking at like a big rig or a large truck, you know, there's a place for a person to sit, there isn't in yours. What is the idea behind Hummel robotics and why that
cabless design? Like what's important about that? So when I was thinking about Hummel and then what I want to do next, right, after working on Thomas Freight for 10 years, I was thinking, okay, what if you back up and you look at where technology is today, what is the simplest possible robotics platform to move freight? Now that was the sort of the premise of the company. It's like what is the simplest way that you can move freight from A to B and uh I was like, okay, well what the simplest way I can imagine is just a platform, right? So you say, okay, well that's that's a wild concept like just a platform that moves freight.
But is it feasible with the technology that we have today? And so you start a little bit of exploration on that. You say, okay, what will this cost, you know, what's what's the state of the autonomous technology for driving it, what's the state of the hardware, you know, what are the rules and regulations around doing something like that? You just sort of like go down like this list and you find yeah, okay, there is this is a feasible project. This is there's something here where freight could just be moving on platforms. Then I you know that at the same time I started seeing platforms like this in the US and in China. So that was kind of interesting moment for me too. So in China they have you know they do have
smaller platforms that I'm developing but basically like the same idea it's just a platform for moving freight. And in the US in places like Long Beach container terminal like that's within a private port area. There are also these platforms moving freight and they're kind of going from like crane to crane. Like okay, so there's like some basis and foundation for this. And you know that you start putting a business idea behind it and say okay like this is a pretty futuristic idea like the idea that we would just have these sort of catalyst trucks. We've checked off is it technology technologically feasible? There's like a business case behind it. You know there is we talked in depth to say hey like are
you willing to fund an idea like this to see where it goes and all that kind of comes together and materializes and that's how we got to humble. The idea though if you would have that let's say the towards let's say 10 years ago back in the original 2016 that we like to joke about. Do you think that you could have produced what you have produced so far? Like technologically speaking and would investors have backed you at the time?
Well you know to your point about it being a frothy time like I think investors probably would have at least looked at the idea at the time but I don't think the technology was anywhere near close to ready at the time. You know that 10 years ago you know some of the stuff that we take for granted now like you use like a chat GPT or you use like a cloud and see kind of like what that tech is capable of. That wasn't on anybody's radar or maybe it was on some people's radar in 2016 so it wasn't on mine. So we were doing very crude algorithms I would say at the time in comparison to the tech today. And so, I don't think it would have been feasible 10 years ago because the tech
wasn't ready. And it wasn't just software technology, you know, we didn't have cameras that we thought were at like a reliability grade and we didn't have the ability to sense in all weather conditions. It's kind of like it in the those frothy times, you don't know what you don't know. So, we didn't know that some of these aspects were going to be harder missing. Like you kind of had to try, see what the results are, go back, iterate, and then try again. Now, today it's a little different story like we've been working on it for a while, we know the state of the technology, we know where it needs to go, the gaps, what's missing. So, we have more understanding. So, so your question is like was this feasible in 2016?
I think that would have been quite early for an idea like this. Would have got funded in 2016? That's That's maybe a different question for that. Yeah. Well, to your point though, it's like you don't know what you didn't know at the time. So, of course, it's like are you thinking about that? Like what is it that hasn't been uncovered yet that would propel humble robotics even further like a new area of technological breakthrough? I mean, there's a lot of focus on certainly AI models, but do you think about that at all? Like what would propel humble robotics even further technologically speaking that maybe is really on the fringe?
Well, today we still have to do a lot of engineering around the brain, right? So, there's not a brain that can drop into our platform vehicle that would just drive it with no additional work. I don't know if our engineers would love that or hate it, but it maybe would be a blessing if we could just drop in, you know, a perfectly working open AI model or something and it just drives to the destination and we'd be like, "Great, we're done, you know, easy." So, there's still quite a bit of engineering that we have to do. What we are projecting and predicting is the rate of improvement of these models and intersecting that with where our business is going to be. So, today we have technology that can do
that can take this cabless vehicle and do like relatively short routes. Like we're talking under 50-mi kind of freight moves. We can do that with the tech today. Um and we can do it in certain locations, right? It's like if you look at um how autonomous vehicles kind of progressed, you know, they started in like sunny areas and then they kind of avoided weather, right? And then they started tackling weather and they encountered all these scenarios that they haven't seen before and we work on those. It's kind of the same thing kind of the same progression. Like we have some limitations on where we would take this cabless vehicle today. But tomorrow, and where we see the
intersection point of a of the business, we can take it much further. What we don't know is whether we are correct about the rate of improvement of the technology, but given how fast it's moving, we feel pretty good about that. One of the things that you have mentioned to me when we previously spoken is about vision language models and you seem pretty bullish around that and not just for Humble Robotics sake, but I'm wondering just for the industry, how do you see vision language models propelling the industry in a way that it didn't, you know, even 5 years ago? A lot of the work that I have done over the last 10 years um in autonomous vehicles was driven off using lidar technology, which is like an amazing
technology. It's like really cool. Like you get to see the world in 3D by firing lasers at like every possible point. It's like incredible technology, obviously. Um a little brittle, uh a little expensive, but I think it fueled it propelled a lot of the Thomas tech, right? You know, the dream I think would be to drive solely off of like a single camera, maybe two cameras, kind of like a human does today, right? I think that's the most cost-efficient way to do autonomous vehicles. And also something that like Elon Musk at Tesla has been proposing and pushing for quite some time, which is that cameras will eventually be good enough to be able to do that, but there's been a question mark about you know, are they there yet?
Yeah, very he's been very bullish on it for 10 years. But I would say 10 years ago, the tech wasn't there, right? So, to kind of to your other question, too, right? Like, the tech has really progressed a lot, and you're now sitting on this place where that question is still debated, right? Cameras versus non-cameras or whatever it is. But it's certainly trending in the direction where you can derive camera-only into basically everything that human can. That's the trend. And so, we, you know, at Humbolt have become really bullish on that, too. And when I started the company, we experimented with these what we call pre-trained models. Basically, a vision model that has been trained on the entire internet, right? For the most part, or, you know, somebody has put a
lot of effort into training the model. And we were blown away by the results. I mean, like, the especially when you had 10 years in the industry doing it a different way, like, we were doing a lot of lidar work, to see the vision model quickly understand scenarios that we would have to spend like months of engineering time trying to solve, and just have it really recognize traffic cones, recognize traffic light scenarios, an officer holding a stop sign. All those in the past used to be, you know, a lot more like hand-crafted scenarios almost that you would try to teach the model to solve, and we just kind of got them for free. So, for us, it started this idea of like, hey,
like, what's maybe go all in on the vision models. We're not so dogmatic about the technology that, you know, we wouldn't use lidar or radar. Like, our vehicle today has lidar, radar, and camera. Like, we use all three of those sensors, because you just want to be as safe as possible. So, it's not really about a about an or even about which technology is better. It's just like, at the moment, we see cameras being pretty awesome, and we're really bullish on these vision models, and their rate of improving has been tremendous. Yeah, but the stakes are high in autonomous vehicles and you know, you're going to you know, put redundant sensors on there and eventually it'll get to that point. The interesting thing in
sort of the startup scene is technologically speaking things change and also as a startup founder also somehow stay the same. And you were you were a founder before and you had an exit with I believe Spark was acquired by John Deere. And I'm wondering when you were leading that company compared to now, is it more competitive, harder to get talent because of what has actually changed in Silicon Valley, which is aside from you know, more expensive real estate, there is a lot of folks focused on specifically physical AI, anything to do with like defense tech, robotics, and AI. And then of course all these AI labs. Is it just a lot harder or is it very similar to back when you had that company?
I look at it a little differently. I think that there are more people gravitating toward the space, which I think is amazing. Like the more we want people to go into robotics. I think robotics is the future. I'm I'm very as you can guess I'm very tech optimistic, right? Like I see the value and like how our society can transform through robotics. And we want engineers to go into this. And I'd prefer engineers go into physical industries than go into, you know, more B2B software products. No shade on those, but I think you know, we I would love for more engineers to go to robotics. I think what's happening is that it is drawing people in. It's drawing, you know, new graduates in. You know, you graduate today and if you're
excited about robotics, there is a healthy industry to go into. So I just think the industry is growing and that's exciting. I would say the entire time I've been in Silicon Valley, the talent wars are there and it's always competitive. You know, the way that we built Humbolt was uh you know, I've worked in a lot of startups and people tend to follow each other when they have liked working together and so we got the band back together from like seven different companies and put them all under one roof, which is like a really fun way to build a company. Oh, and I should mention the roof that you're under is actually a throwback to 2016, right? So this was Cruise's first out of the garage into a physical office
space place. That's right. Yeah, this is currently what you're looking at it right now. This was Cruise's I believe first, yeah, first real building that they were in and they did a lot of amazing work here and we've had a Cruise employees come in and like be very nostalgic for their time here, which is really fun. But there's good DNA here, right? And we're very thankful to Cruise team because they did an amazing job, you know, making the building better. Yeah, but to your point you talked about people following each other from startup to startup and then you have a whole new batch of folks who seem interested or drawn in because of the rise in popularity. At the same time, you know,
I've talked to people who have said that there's an interesting kind of poaching wars that are happening. Not that is new, but where they're drawing from. So robotics companies find themselves in competition with autonomous vehicle companies. So robotics companies not working on, let's say an autonomous forklift or self-driving trucks because there's that same skill set, which is applied AI. And I'm wondering if you're having a similar experience and if you've had to be creative of where you find your talent. Yeah, I mean it's there's definitely a talent war and I think you're 100% right that those industries have a lot of overlap in the talent that they look for. A vision model is something that
you might use on autonomous vehicle and it's something that you might use on a humanoid robot. It might be something that you use for your, you know, drone inspection program, right? It's like all that's like common talent. The way I think about it is this, you know, I will paint a vision for where Humble goes. We get excited about where the our industry is going and we get a lot of interest of people who want to kind of participate in that. So, it's it's not been too hard for us to find good talent, especially for the size that we are. But, it is always it's always difficult and I think something that probably keeps up a lot of
entrepreneurs and founders up is like, where am I going to get the talent that we need? Uh and I would just encourage everybody who's excited about robotics to just get into it because it's a great time for it. Well, then you have more people that you can hire, you know, so that's that's benefiting you. Yeah, I'm wondering to sort of close things out, I'm just wondering if you know, you kind of gave a little bit of advice there which is encouraging, you know, people to get into robotics. But, for but for the founders out there who are trying to like do they just need to be thinking about how much money they're raising these days and how to apply that? Because certainly there's a lot of
startups outside of the Bay Area and Silicon Valley, but there is quite a bit of activity in this area and a lot of opportunity for those people who are looking for it. So, how should startups be thinking about how to manage these like very competitive salaries around either how they form their company or you know, create the culture or how much they're raising? I think still people are drawn to companies with great cultures and that is you know, I'll sort of answer that in a reverse way. I actually think companies should really start with the culture they want to build and draw people
on that before they try to draw them in on just competitive salaries and compensation. For a lot of engineers, they have options. They have many options and so and usually they have an option that pays better than the one that they ultimately choose. They'll choose based on the company's vision that speaks to them, the culture that speaks to them. In some ways it feels a bit more like the 2016 mindset which I think changed over the years where you know, people gravitate toward the project that really captures them. Um and you know, the compensation in the area pays well and well enough for it for some of these roles. I would encourage founders to build that amazing culture. You know, you don't need,
especially these days, you don't need these enormous teams. You know, I've been part of teams of thousands of engineers. That's a very hard ship to maneuver, right? When you have thousands of engineers and getting that coordinated and it rarely works. Actually to have so many people. So, I would also encourage founders to think not about culture, but about being efficient, right? You know, um give people the responsibility that they want and then give them the roles that they're excited about. Um and then they will come. That's how you win. That's how you win people over. Compensation is one part of it. I think that's that's been true for a long time, but feels very relevant now.
Yeah. Well, they have a lot of choices and if you look at what the salary ranges are, they're, you know, up there, 300,000 or so. So, it's you kind of have to offer something maybe a little bit different, more unique. We are about out of time, but where can our list I know it's Fed by, right? Yeah. Fed by. Where can our listeners connect with you or find your work at Hubble Robotics online? Yeah, so you can visit our website hubble.robotics.ai. I'm uh always available to connect. You can find me on LinkedIn or find me some other way, but uh you know, I've been in this ecosystem a long time and I love chatting about this, obviously. So, uh
always happy to connect with excited people. Well, there's the open invite. You'll probably get a lots of inbound now. Thank you so much, Al, for joining us and to our listeners, thanks for listening to another episode of Equity. You can find Equity at Equitypod on X and Threads. Talk to you next time.
