The Wealth Cockpit · Episode 8

Why an Amazon robotics engineer moved to the woods

Cover art for Why an Amazon robotics engineer moved to the woods
Why an Amazon robotics engineer moved to the woods
26 min 13 sec Download MP3

Show notes

Tait Carroll spent 25 years building automation: twelve years running his own factory automation firm in the Pacific Northwest, then four and a half years at Amazon Robotics as lead engineer on one of the biggest robot deployments in North America. Then he moved to the mountains of northeast Oregon and started Larix Robotics. We talk about what fixed-bid engineering teaches you about estimating, how he turned around a roughly $4.5 billion Amazon project and saved about $2 billion of it, and why he picked forestry by running a list of business ideas through a screen instead of chasing a passion. His target is pre-commercial thinning: hand crews walking young, replanted forests with brush saws, deciding which trees stay. By insurance rates it's the most dangerous job left in forestry. Five years ago a machine couldn't tell a tree from a shadow. Today it's hard instead of impossible, and Tait has a working prototype that finds individual trees in a real forest. We finish on where the same technology goes next: fuels reduction and wildfire. A detour from real estate, but it's the same thing I look for in a deal: a real problem, a clear screen, and someone honest about what's still hard. Disclosure: I'm an advisor to Larix Robotics and a shareholder. This episode is not an offer to sell any security. Larix Robotics: https://larixrobotics.com Want the show notes? Email wealthcockpit@neelypi.com.

Transcript

SPEAKER_00 0:00

Hey, I'm Brent Neely. Welcome back to the Wealth Cockpit Podcast where we plan the flight before we take off. Today I have a special treat for you, a little divergence from real estate, but it's still I think it's very interesting. Tate Carroll from Larryx Robotics. He has been an entrepreneur. He worked for Amazon in automating some massive systems inside their warehouses. And now he has started up a new company that's in the forestry vision and fire prevention technology. I've been working as an advisor for him on the business planning side of his startup and wanted to kind of share what he's doing with all of you guys. So if you have any questions for me or want to get the notes from the show, feel free to shoot me an email at wealthcockpit at nealypi.com. Otherwise, I'll uh kick it off on the show. Hope you enjoy. Welcome, Tate. And uh, can you tell us a little bit about your background and and uh yourself?

SPEAKER_01 1:07

Sure. Um thanks for having me on. It's kind of fun to talk to more people about what I'm working on. So um um me, like from business perspective, um, I'm an automation engineer. I work on full stack um automation system problems. Um, my I've been working in this field for about 25 years. I came out of school with an industrial engineering degree, which is not the not the normal start for what I do, but what I've it's kind of put me on this track where rather than designing a robot or an electrical system, um, I really have enjoyed working on these like business layer problems where you look at the stack from like what's the business objective and how do you turn that into an engineering problem? And how do you solve that engineering problem to deliver to a business layer? Um and that's what I've done for really the last half of my career after, you know, after getting in early, early career started stuff.

SPEAKER_00 2:03

Yeah. You um so you originally started your own engineering firm, right?

SPEAKER_01 2:08

I did, yeah. I got out of college and I worked kind of normal early factory jobs. Um, and I've I've always been like really torn between do I like business, do I like engineering, and like really I like both. So I got this uh sales engineering job kind of early in my career, and it let me kind of do both. And from that, I built a pretty good network. Um, and I also saw this opportunity for um where I was living. Uh, there was no contractor to really kind of fill this gap of delivering custom systems. So I ended up doing that and it worked pretty good. I did that for about 12 years and was able to build it up into a nice little business, nothing like venture scale amazing, but a good business to pay all the bills and buy some toys too and have a lot of fun.

SPEAKER_00 2:53

Yeah, yeah. What uh in layman terms, can you tell us a little more like what sort of systems and problems were you solving with that business?

SPEAKER_01 3:03

Yeah, in that business, um, we built custom equipment for factories in the Pacific Northwest. So um it was mostly either inspection systems or material handling for factories. So it was everything from um like if you ever buy Georgetown uh beer in Seattle, um, every can of that beer, if they needed a palletizing robot, they had guys working all day long just stacking beer onto pallets. And you can buy those systems, you can buy a robot that'll palletize whatever you want, but they're tucked in in Georgetown in this really in a footprint that they can't change. And they had this tiny little space to put a palletizing robot. So things like that, where we build a really custom palletizing robot that would fit into a space. We did stuff like that. We did uh welding robots for nuclear fuel rods assemblies, we did label inspection systems for airbag manufacturers and pharmaceutical vials and all kinds of stuff like that. But and it was really fun for an engineer. It's like every project unique. Um, really fun stuff.

SPEAKER_00 4:06

Yeah. No, it sounds uh fun. Yeah, problem solving on steroids. And yeah, yeah.

SPEAKER_01 4:12

And we did the work on fixed bid. I I would blind bid the work. I would be like, we need this thing, and I would put a spreadsheet together and go, like, well, I think it'll take this many design hours, this much commissioning hours. I think it's gonna take this much in parts. And it was fixed bid, so it really put me on the hook to like get good at estimating because the losses can be really big if you get that wrong. But the margins were good if you got it right.

SPEAKER_00 4:36

Yeah, interesting. Interesting. Okay. Um, and from there you you went to work for Amazon, right?

SPEAKER_01 4:43

I did, yeah. I I had plans for the consulting firm. We had a nice like kind of regional spot that I thought we'd be able to build it and sell it to a consolidator. Um, and then COVID was uh yeah, I didn't plan for COVID. One of my bigger businesses say, so it's not predicting uh come on, Tom. Yeah, yeah. So managed to scrape through. I brought on a partner, he bought it out, and the business is still running. So that's good. All the bills got paid, no bankruptcy. So and and in the grand scheme of things, that was great for us. So then I I was like, what am I doing next?

SPEAKER_00 5:15

Um funny because I was on the other end of that where I didn't plan for COVID, but my business just accelerated with COVID. It just went to the moon, which was a different, different set of problems.

SPEAKER_01 5:30

But yeah, thanks for having some salt in that one. That thing's good. I had to throw it in there.

SPEAKER_00 5:36

I had to throw it in there. But all right, yeah, sorry, carry on. So um, yes, sci-fi robots at Amazon. That was where you were left off, I think.

SPEAKER_01 5:45

Well, I I think there's a lot to learn from those, like those big events, like that's part of business stuff, is like those risk assessments. So what do you what do you build cushion for? And what do you say, like, well, you can't build cushion for everything. Um yeah, so I ended up uh going to work for Amazon Robotics, which was it was really like this is not in the plan, but um, I needed to figure out the next thing to do, and uh so I took a little time off in between. Um and yeah, started there, and it was really, really great experience. I was there uh like four and a half years. Um, I was hired in as the lead engineer on this enormous robot robotics deployment. Um they were they had a design that was in progress. I took over the design to shepherd it through um the concept and getting it to where they could start copying and pasting. And the project was like it was big enough that it was gonna be like 10% of all the industrial robots deployed in North America were gonna go to that project. Wow. Um, like probably the biggest robots deployment in North America, probably in the top five in the world. Um and so it was really it was a scale I obviously hadn't worked at before because it was like the biggest. Um really, really enjoyable to take all these kind of like there's like all these like concepts of in a factory, like you can save compressed air and you can save a little bit of money here and there. But then at that scale, it's like, oh no, if we do that, we can save 40 million dollars in compressed air every year. Um just like it's like really fun to take these theoretical things and all of a sudden they become really practical. Um, and you stop just ignoring the margins and you like start managing all those margins too. Um, but then the what I feel like my big accomplishment was there was this design was in process. It's a big organization and it has all the big organization dynamics, like people really focused on their promotion path rather than the good of the company. Um, there was a lot of momentum behind the project, and it was targeted, it was it was gonna be about a four and a half billion dollar spend to build this out. And it was still the overpriced, it was the wrong hardware for the job. Um, so I was able to, it was mostly the political work. The design honestly wasn't that hard to say, like, here's the function we need to do. What's a more effective hardware solution to do this? And then months and months of careful negotiating to talk to the right people and to not upset the wrong people and get that turned around. And was able to save about two billion dollars on the project and not get fired for it, which was close.

SPEAKER_00 8:26

That's pocket change. Two two billion.

SPEAKER_01 8:29

Yeah, yeah. By far the biggest projects I've gotten to work on. Um and uh but really satisfying to you know, like that's uh in Amazon money, it's important, but it doesn't change their business. Um but I always kind of look at it like whatever you're working on, whoever it accretes to, it's still, you know, in some way accretes to society, still some way makes things better. Um maybe that means we get more plastic crap we don't need cheaper, but I think I think it's not positive. What we choose to do with the wealth we generate is one question. I think generating wealth is an inherent good. Uh yeah.

SPEAKER_00 9:08

Yeah. I I bet the uh the other robotic supplier probably doesn't agree with you. Uh if you uh that you switch from, right?

SPEAKER_01 9:17

Yeah, yeah. They were they were one of the main they were one of the three names that you have if you're robots. Um, and so it's a bummer for them for sure. Uh yeah, and and it wasn't just about switching brands, it was about the form factor of that type of robot. Was um it was like using a a two-ton truck to deliver pieces with like no, it'll do it fine, but like why are you doing that?

SPEAKER_00 9:40

Yeah, yeah, yeah. I gotcha. No, that's interesting. It it's interesting to me to look just look across our, you know, the United States today, and the the biggest uh CEOs, the biggest founders of businesses are there's a lot of engineers. And uh that's it, that's how I've always looked at business is that you know it's solving problems is bringing value to to people, and the bigger the problem and the more people affected, the more um value, the more money and wealth you can create. So I, you know, I really agree with what uh you just said. Like you could look at it like, oh, it's greedy, and there's these people that are trying to just make the most money, but uh that's how it works, is like if you if you bring value, that's how it should work anyway. It it probably doesn't always work that way, but um if you can if you can solve problems for people, uh bring value to them, and then that's when wealth is created. So um that's pretty cool. That's it sounds fun. So tell us I guess what are these robots doing? Like there's like is it pitch black, there's no lights, and they're just going to get Chinese junk off the shelf and send it to me faster?

SPEAKER_01 10:49

Yeah, yeah, they're uh it's not pitch black, it's in the fulfillment centers. And actually, if you if you search for Amazon's Sequoia robot, um I don't think you have to look too hard. You can find it made the Wall Street Journal, but it's like buried in this whole system. So like I can go like that's my robot in the journal, you know. But you can find some YouTube videos and stuff pretty easy. Um up there was a media push where they uh sent us out. And so my robot was part of the whole Sequoia project, which was like this refactor. Um, and when you go in the nuts and bolts of it, it's like this is so mind-numblingly boring, and I can't believe there's value in that. But the volume of Amazon goods, when you save a penny on the delivery costs of one item, it stats up really fast. So the overall project was to containerize all the stuff in the fulfillment centers, um, you know, billions and billions and billions of items. Um, and by getting them containerized instead of these soft fabric pods, it solves a whole bunch of things. Um, so my machine took the containers in and out of shelves. There's these portable shelves that oversized, like Roomba-looking robots move around. These huge, like half a million square feet on every floor, four floors in a building. Um then those get out, the robot takes the container out and presents it to a person to pick the actual item out of the container. Um but there's a bunch of buildings, and there's three or four hundred of these robots in every building, and adds up.

SPEAKER_00 12:23

Yeah, the scale is is uh hard to imagine.

SPEAKER_01 12:27

Yeah, yeah. Like if I I go on site to a test site and on one floor, like we had one floor built out to get to walk, and it's all caged in in the like storage area, so that there's to contain this uh robot safety hazard. It was a half mile walk to walk around the perimeter inside.

SPEAKER_00 12:47

Inside the building, yeah. Inside the building, yeah. Interesting. Yeah, well, we gotta have our stuff now, not not tomorrow, not the next day. We need it now. Yeah, yeah.

SPEAKER_01 12:59

That was I gotta work on another project to make it more now. Uh working on designs for um instead of having regional distribution centers, having like these small little centers tucked in city cores. I gotta spend a bunch of time working on the architecture team for that, also. Of how do you make ideas and be able to squeeze that into a leased building and yeah, yeah, a lot of fun stuff.

SPEAKER_00 13:23

Yeah, yeah. Interesting. Well, let's uh we'll pivot a little bit. I I got to meet Tate um because he moved out here to my neck of the woods. Uh, if you're if you can see the video, the the mountains are in the background behind him or yeah. Um so uh yeah, he lives out here in uh Willow County with me and has started um a new project that I'm kind of uh advising him some on. But um yeah, maybe you just segue into that, Tate, and kind of tell us of why uh why you're doing this and what you think the applications are and things.

SPEAKER_01 13:58

Yeah, yeah. Um yeah, so I did move out here um and largely because of the view back there. I well, and I also I grew up in Northeast Oregon and I've always just thought this is um like don't tell anybody because uh we don't need more people here, but uh so I can find most cool places you can be and a lot of great people. I have long-term friend things here. Um uh so I moved really just for lifestyle reasons, and I've I've worked like pretty hard. Like, you know, I can I can say uphill in the snow both ways, and you know, I can honestly tell you I've had a job since I was five years old. Um, and I've just I've worked a lot and and then where life things kind of came together. I was like, well, I'm gonna try just taking a little break and see what I want to do next. Um and so I moved out here and I made it, I think I made it two and a half weeks before I got pretty bored and was like, okay, I'm just gonna go play in the mountains for six months. I gotta get to it. And I what I thought I was gonna do was I kind of had a list of like maybe this business, maybe that business. Um, and so I got to work on that list. I got through my number one, it has great margins, all these great things, no way to protect it. I would have gotten squashed once it started to grow. Um, so I was just kind of like marching down my list, and I there was a machine that I was looking at for this like timber operation that I thought I could make a better version of. Finally found market sizing data, and it was like a like a $10 million national market for the use of this, which is not enough to build a business on at all. Um, but in the same thing, I found this thing called DCT pre-commercial thinning, that was a decent sized market. And I I don't know anything about forests. Like I like walking through the forest and hiking and stuff, but I don't I don't know about forestry. So I dug into that more, and it really ended up the more I could dug into it, the more it's just like the classic thing you want to automate. So what it is is in clear-cut forestry, um, in high productive forests, like a vocastase, they clear cut, they go and they replant, which is nice to know it's required by law, they replant all the clear cuts, the trees aren't going away. But when they do is just like in your garden, if you're planting like a bunch of basil, you're gonna plant seeds, you have to plant too many, and then when they sprout, you thin them out to the right spacing you want. They do that the same thing with the trees. And right now, that's done all by hand crews. Depending on the forest and region, it's either chainsaws or these like oversized weed eaters, and you've got people walking through the forest um deciding which trees should be kept, which ones should be cut. And it's like really hard work. It's um if you go off of the insurance rates, it's the most dangerous job left in forestry. Everything else has been they've found ways to make it safe. Um, and then when I dug through it, like outdoor automation is really, really hard. Um, making a robot move where you want it to move, that's a really solved problem. It's done everywhere. The challenges are always on perception. In like in a factory environment or a logistics center environment, you structure the environment so you can uh so you can see well what's going on. But when you're like in agriculture and forestry, like it's so hard to all the conditions like just see like where is a tree. Is that a tree or is that a shadow? And when you but so all those problems, um I don't like buzzwords, I don't like the like latest paths, but like it really is like AI vision processing. I don't want to be on the AI bandwagon of like it solves everything. Um but in this case, what we can do now with AI processing of images, it makes this like a solvable problem. So the more I dug into it, the more it's in that sweet spot of it. There's huge benefits from automating, and it hasn't been done yet because five years ago this was impossible. And today it's hard.

unknown 17:47

Yeah.

SPEAKER_01 17:48

Rather than impossible, it's hard. So that's what we're working on is building uh doing the initial startup phase work to start building this business.

SPEAKER_00 17:58

Interesting. So when you're when you're talking about the perception, like you know, these these autonomous robots uh in the factories, there's like lines painted on the ground or like different uh indicators that you can put up where you're you're controlling their environment and what they're navigating off of. Is that kind of what you're talking about?

SPEAKER_01 18:19

Exactly. Like there's a whole like evolution where the old original ones they started with like magnetic tape in the floor. So they can follow a magnetic line really easily. And then as perception has grown, um, now uh autonomous mobile robots in factories, they work a lot like uh self-driving cars. They've got the same sensor speed, but it's still a more controlled environment. Like it never snows inside of a factory, it never rains inside of a factory.

SPEAKER_00 18:48

Um yeah, yeah. Yeah, no, I've been uh you know out in nature in the forest enough to know, yeah, trees leaning over, you know, bigger branches, or like it's it's a very uh unstructured uh for sure. So that's yeah, that's really interesting. So you think that um you think that that problem is solvable now at this point with the technology that's available?

SPEAKER_01 19:17

Yeah, yeah, I thought so. I started really in earnest on building a prototype last, well, right about a year ago, right now. Um, just like doing the analysis and prior experience of like this is doable. Um, and then I went and talked to a number of foresters at timber management organizations and consulting foresters, and the universal response was like, yeah, this would be amazing. Uh if you could do it, you know, like they're skeptical. Um it's fair. I think it's fair skepticism. So I I a year later, um, I have a prototype complete. Um, it's running basic functionality, uh, really basic functionality, um, but shows the concept that does work that we can detect individual trees in an actual forest environment. There's a ton of scaling work left to do to get high quality detection of what species and tree health. Um, but we're now at a place and we can go back and I've been having these meetings where I go back and I show Forester's video of it working, and the conversation is just totally different. It's not. Well, that would be neat to like when when can we use this? Can I invest? Like they're like exciting to hear people just respond.

SPEAKER_00 20:30

So yeah, right, right. No, that's exciting. That is uh that is cool. So I mean you've told me, so your your vision for PCT pre-commercial thinning is to be able to have sort of a um, I don't know what you call call this, but like a set of these five, five or six uh robots that are going out so that so that people can get out of the woods, uh running chainsaws and these brush cutter saws and things, and just have one person kind of monitoring, you know, this fleet of machines going through the woods performing the thinning. Um right? Yeah.

SPEAKER_01 21:06

Yep. Yeah, and I'm calling the fleet a flock because you think a flock a lot. So and it, you know, it's kind of kind of like a little shepherd tending his flock out before us. And I I think the work of that person is gonna be really similar to like if you if you have a room in your house and like you know, it's automatic, then you don't need to do anything except you need to definitely pick up all the USB cords on the floor and everything that's gonna get tangled up in. Um I think it's gonna be very similar if you had five roombasing your house all at once, and you wanted to get done as fast as you can. Just all the little things tend into those. Yeah. Which doesn't get to like full autonomy. Like, you know, when we think autonomy, we think like push the button, go away. Getting all the way there is like really, really, really, really, really hard. But getting a lot of the way there and getting a lot. Economic value out of being a lot of way there, that's where it becomes doable with current technology.

SPEAKER_00 22:06

Interesting. Well, another thing I've been thinking about, um, you know, as I've learned about your business and what you're doing is is the other applications. And living out here in the West, uh, in the US, and you know, wildfires being more and more uh prevalent and more and more intense and things, it it seems like uh have you thought about some of those uh different kinds of applications?

SPEAKER_01 22:31

Yeah, yeah, and that's yeah, actually a great intro. Because it really has kind of unfolded in the same way that we're talking about. This was found this market for PCT and was like, okay, you could build a healthy business around that. And that's what, like in all the business fundamentals, that's what we're underwriting now, is that business case and it all cancels. But like everything about this machine is actually pretty easy, it's just known stuff. It's a hydraulic diesel-driven machine with uh automated controls to drive it. That there's no real risk we can build more of the whole thing is can we perceive? And once we develop this perception system, we can use that in so many different forestry applications. Um the one that I'm personally most interested in, um because I I do want to make things better, and I don't want to just make money, I'd like to make the world better, um, is fuels reduction. A lot of all the wildfire problems we have, not all of it, but one of the major components is that we have built up this huge amount of dry fuel in our forests because we've been putting out forest fires ever since the Tillmook fire, which was like 100 years ago or something, that made this change in how we manage. We haven't let the natural fire cycle happen. Um, and that's one of the main reasons we're getting these enormous burns every year. And yeah, like you were saying, if you live in the West, you know there's like a month of smoke season now, which was not when I was a kid living in this same region. And so this the perception system, we can take that whole package, the hardware and software, put that on a new machine that's sized for different operations. Maybe it's a large machine, it has different functionality. Um, but it's really exciting because those markets and those business cases and the environmental impacts are significantly larger than uh than the case we're underwriting the business with right now.

SPEAKER_00 24:24

Yeah. Yeah. No, that's that's really cool. And it's a great uh great cause as well, I think. Yeah, a month, a month of smoke season might be understating it. We had a pretty good pretty good year this year. Maybe it was only a month. I feel like it was three months last year. It was just never never ending. Yeah. Yeah. So pretty hard. Yeah. So well, that's uh that's yeah, no, that's great. Uh so is it I guess is there anything else you want to share about lyrics robotics and and what you're doing now?

SPEAKER_01 25:01

Um I you wow, you just I didn't know that would be such a hard question. Um share. Yeah, we yeah, uh well, we can just wrap it up there.

SPEAKER_00 25:12

If um yeah, if if people want to get a hold of you, uh learn more about lyrics or how they might be able to be uh involved. I know you're starting with some investors and like trying to get it to the next level. How would they learn more or get a hold of you?

SPEAKER_01 25:28

Yeah, lyricsrobotics.com is our website. Um and there's links on there to be able to get to my email. I also have an unusual name, so you can also put my name into Google and you'll get to my LinkedIn page pretty easy. Yeah, it's not like trying to sell random metal parts from China to me that I don't care about, I'll I'll respond for sure.

SPEAKER_00 25:49

All right. Well, yeah, look up Tate Carroll on uh on LinkedIn or uh lyricsrobotics.com. And uh yeah, if you'd like the the notes from the show or any questions from me, you can shoot me an email at wealthcockpit at nealypi.com. But uh hope you've enjoyed the show and I really appreciate your time coming on, Tate.

SPEAKER_01 26:09

Yeah, thanks for having me on. This is fun.

SPEAKER_00 26:12

Yeah. All right, we'll talk to you later.

Disclosure

Securities offered under Rule 506(c) of Regulation D. Available only to verified accredited investors. This page is for informational purposes only and does not constitute an offer to sell or a solicitation of an offer to buy any securities. Any such offer would be made solely pursuant to a Private Placement Memorandum and related subscription documents. Real estate investing involves risk, including the potential loss of principal. Projections are forward-looking statements based on assumptions that may not prove accurate. Past performance is not indicative of future results.