Mark Messina - COO at Geek +

(Joel Beasley at 00:00:00) Hello, my friends. Today we are talking to Mark, the COO of Geek Plus, and we discuss robotics equipped with AI driving automation in warehouses, how 5G is opening new doors for futuristic technology, and how cloud computing is making mass deployment of AI and robotics possible. All of this right here, right now on the Modern CTO Podcast. Here we go. This is the Modern CTO Podcast.

(Joel Beasley at 00:00:38) Are those robots in your background?

(Mark at 00:00:40) Those are robots in the background. That's right. We had some robots running around for you today.

(Joel Beasley at 00:00:44) Is this a video loop or is this for real?

(Mark at 00:00:47) Those are real. Those are for reals. That's our picking robots.

(Joel Beasley at 00:00:51) That's crazy.

(Mark at 00:00:53) Yeah. So they manage all the inventory. So those yellow pods that you see, that's the racks that inventory lives in. When you place an order, a robot goes, picks up one of those shelves, and that's where it all starts.

(Joel Beasley at 00:01:06) You know, I had seen the video a long time ago on Amazon, like inside an Amazon warehouse where all the robots were doing that. And then I got to see that there were companies that were making these. And at that, it's the moment when it occurred to me that other people need this picking technology as well.

(Mark at 00:01:22) That's right. That's right. It's cool stuff. It's definitely different. We've kind of turned normal operations on their head, so to speak.

(Joel Beasley at 00:01:30) How long has this been a thing?

(Mark at 00:01:33) You know, I think really since probably 2006. It goes back quite a ways. You know, if you kind of go back to the genesis of all this stuff, it really I think started with Webvan way back in the day when Webvan was trying to do grocery fulfillment and they failed. And in the end, in the summary, one of the reasons was they couldn't reduce cost to actually pick an order enough. So the founder, one of the founders of Webvan, left and started up another startup which turned into Kiva.

(Mark at 00:02:10) And the onion that they peeled was intralogistics. So in the warehouse, how to deliver an order efficiently, right, cost effectively. And, you know, if you think about your traditional warehouse, somebody's walking around picking. I think a lot of people when they shop online, they don't really have a good idea where their stuff comes from, right? You know, if you really go all the way back, your stuff comes out of the dirt, goes into materials, goes into a factory, goes into production, all this. But if we just talk about the delivery piece, you know, I think when people place an order online, they don't realize somebody actually has to go move that material, put it into a box, label it. Somebody has to carry it, deliver it, all this. And so, you know, the founder of Kiva, he said, well, I'm going to invert that. I'm going to have the inventory come to the person. So instead of a person walking ten, fifteen miles per day literally in a warehouse picking orders, the stuff just comes to them and they stand there and pick it. And so it was a really catchy idea. And around 2012, Amazon acquired that company. For them, it was a lot of money, almost $750 million, maybe a little more.

(Mark at 00:03:20) And the rest is kind of history. That group in the Boston area spun off a whole bunch of different robotics companies. You know, you've got Fetch, Locus, Six River, companies like IAM Robotics, Geek Plus, et cetera, all sort of came out of that concept of inverting the way the operations worked. So it's been a kind of a neat time, you know.

(Joel Beasley at 00:03:42) And these aren't small businesses either. I mean, you have over 800 people at your company.

(Mark at 00:03:47) That's right. That's right. These are large entities and we support world-class players. You know, you think about the major retailers of the world and what they're doing and how they've had to respond especially with coronavirus where a lot of the shopping has gone online. And so it's fueled a fire that was already underway and pushing the industry in this direction of automation.

(Joel Beasley at 00:04:13) By the way, they're beautiful robots. Last night, I was sitting on the couch watching BattleBots with my wife and kids. And she was asking me, oh, you know, what's going on tomorrow? Who are you talking to tomorrow? And I was like, oh, you've got to see this. And you are exactly right. You hit it perfectly earlier when you said most people don't understand this because I showed her this video that I don't know if you guys sent it to us or not, but it shows the picking actually happening in a fast-forward motion. And it's really upbeat. I loved it. I thought it was really slick marketing material to understand what you do. And we'll even post it in the show notes since I'm talking about it. But I shared that with her and she's like, whoa, this is what? And I was like, yeah, when you press order on Amazon, you know, there's people, they're picking. And she's like, that is so cool.

(Mark at 00:05:00) Yeah. Yeah. We have a lot of different robots. So we've been at this since, I think, 2013 or so. So we've got a lot of intrinsic knowledge that we've developed by creating our own, we have our own warehouses. So in China, we operate our own, we own and operate our own 3PL, which means third-party logistics. And so, you know, the world's biggest shopping day is 11/11, right? Singles Day.

(Mark at 00:05:30) And we process a tremendous amount of orders. We set records every year, more and more every year. We're breaking new records. And in order to be really good at what we do, we own and operate a 3PL because it's not only do we use it to process orders for our customers who use us as a service, but it's the best lab, right? So when you're actually operating something, it's very different than when you're just providing a service. When you have to live with it, it's a whole different ball of wax. So we live with the software. We live with the hardware. And we have our own factory in China. Our R&D is in China. Our factory has, you know, more than 30,000 pieces per year capacity, and we're building a second factory. So we're expanding. We're bringing manufacturing and integration to the States as well to serve this market. But, you know, the point of why I mentioned that is our software has been around for all these years and we continue to add to it.

(Mark at 00:06:27) So every quarter we introduce 200 new features. That could be bug fixes, right? Because there's stuff that happens. And every time we install a system, it's like a fingerprint because even though it's robots and racks and all of this, those building blocks, how you apply them are very different for each client. So it's a fingerprint. And they may find some weird corner case. So we do bug fixes, but we also do feature enhancements and feature requests. And we release, you know, 200 plus every quarter. And so the software is very mature. And if you're not too familiar with what the software is, our software manages the inventory.

(Mark at 00:07:04) So when you, if you see the cubbies, the rack behind me, you can see it's got lots of little cubbies on it, right?

(Joel Beasley at 00:07:10) Yeah.

(Mark at 00:07:10) So each of those racks can have 150 cubbies. A cubby can be storing one particular type of item, just golf balls. Or we can say, put golf balls, pencils, and shavers in the same cubby. Just mix them. The software will determine how different they are so the human can reach in and just pick because you don't want to put, like, you know, golf balls and the only difference is a number on the golf ball, then the picker is going to go crazy trying to pick the golf ball. So our software will manage all that inventory. Do we do very high-density storage? Do we do random storage? Do we do low-density, fast-moving? Lots of different ways you can manage the inventory. And this software controls all that.

(Mark at 00:07:50) But then we also control, so we've got a number of different modules that we run and we do traffic management. We do heat mapping of the inventory. We do sorting of the inventory. We can run peripherals. So if you want to have conveyor or bin picking, a vision-based bin picking robot to replace a human, we run all of that on our platform. Very broad, very deep software platform. And as a result, the robots you see behind me are kind of like our bread and butter. That's our picking robot. And we call it, it's picking but it's also in the industry known as goods-to-person because it brings the goods to the person. Very simple.

(Mark at 00:08:28) In the picking series, we've got probably, I think, five different capacities of robots. So we can handle pallets. We can have big loads, small loads, all this. But then so that's, I'll call it a vertical. Then we also have factory robots. We do moving robots. We do disinfection robots. We have forklifts. We've got many different platforms. And then within each one, we've got different capabilities and different capacities.

(Mark at 00:08:54) So because we have our own software that we've developed and we have deep knowledge in AI and machine learning and all of this, and in the hardware space, we've adopted an automotive approach to engineering. So we build for reliability. We build for quality. We build for low cost because we build for volume, right? So we, like I said, we've got 30,000 capacity per year. We're building to make a lot of these things. So we drive the cost out, the reliability up, all of this, and that creates a really robust offering that, you know, you've got low cost, you've got reliability. And if you're running a distribution center or a factory or, you know, anything like this, we've got a broad enough product base that we can fill in a lot of the gaps under one platform, which is really nice from the client perspective because they're looking at it and saying, okay, we've got one throat to choke, right?

(Mark at 00:09:46) If it works, if it doesn't work, we know who to call and it's all under one interface and it's all under one service plan and it's all one quality. So it's a big advantage in the market.

(Joel Beasley at 00:09:57) So how did you meet the executive team at Geek Plus? How did you actually get involved with the company?

(Mark at 00:10:02) It's kind of an interesting story. I reached out to the CEO back when the company was, I think, maybe like 100 people or 80 people. It's still a pretty small company. And I reached out and introduced myself to him because I had, like, at the same time designed what we call a sorting robot. And he released his sorting robot on LinkedIn. And I said, hey, that's a really interesting sorting robot. And I sent him a sketch of mine that we were doing as well at that time. And it was just, it's kind of funny because I reached out to him and I don't think he knew what to make of me. Here's this guy from the States. He's in Beijing, CEO of a small company, and I had just left Amazon and he's looking at me going, who's this guy? What's he want? Does he, you know, maybe he wants money or something. Who knows, right? So it's really kind of a weird intro.

(Mark at 00:10:52) And after a while, you know, I traveled to Asia a lot. So I went over and I met him and we kind of broke the ice. And, since then, you know, that was, it's been more than five years. And so, you know, when they decided to open operations in the States, I had a meeting with our GM in Hong Kong. And yeah, we just, we had already been kind of collaborating and we kind of knew each other. And so it was like a natural fit to open operations in the States and that turned into the Americas. So we cover Chile to Alaska, right. It's really a big area and we have this massive market and it's good. It's really good. So yeah, I met him by kind of a funny introduction.

(Joel Beasley at 00:11:38) So are the robots pretty? I just, I keep seeing them run around in the background. It is, by the way, I think you win the award. I'm going to tell Harry from Zoom. I'm going to be like, dude, this guy's background, he's got moving robots constantly doing things back there. And I'm just curious, have you ever, like, maybe Friday afternoon, everyone leave, have you ever ridden on one? You ever just, like, jumped on one?

(Mark at 00:12:05) Not only that, I put my dog on one. I've got pictures of him riding around.

(Joel Beasley at 00:12:08) Oh, that is exactly...

(Mark at 00:12:09) I've got a four-month-old pug puppy and he comes into the office with me. So we have a 10,000-square-foot office here and he's the office dog and so I have a video of him riding around and he's just adorable.

(Joel Beasley at 00:12:22) It's so funny that you say that because in the prep for this episode with my production team, they were like, you know, we've got to ask him if he rides one. I'm like, I don't know. I don't know if we should do that. But I had said that, so you know what? You would have made this video better because we watched your high-energy video in the prep call. I was like, they should have just, like, when it was going, they should have just put like a dog on one of them.

(Mark at 00:12:46) I have funny ideas, like, do we just do, like, a video where there's somebody kind of, like, sailing by in the back on the robot? You know what I mean? But it's not allowed. I'm not supposed to do it. So, you know, if you're one of our users, don't ride on a robot.

(Joel Beasley at 00:12:59) Do not ride them. This is a make-believe conversation that we preset up.

(Mark at 00:13:04) I'm fully monitored. Somebody's watching for safety to make sure that if I, you know, there's no potential for injury. That'd be funny.

(Joel Beasley at 00:13:13) I wouldn't be surprised if you looked at your security footage and caught, like, an employee riding one.

(Mark at 00:13:17) I'm not the only one. I'm sure. I'm sure. We actually have, for one of our videos, I think, around Christmas time, we have a rainbow unicorn in one of the inventory bins. And you can if you watch, you can kind of see him passing by in the background.

(Joel Beasley at 00:13:33) There you go.

(Mark at 00:13:34) Just little, like, cameo stuff.

(Joel Beasley at 00:13:35) That, that's what makes it fun when you're making these videos and just putting that touch on it. Like, I was listening to an Elon Musk interview and they were, you know, he describes how they're building their Tesla cars and he's like, yeah, we kind of just do stuff because we think it's cool. It's an exercise in coolness. And I was like, see, those little moments are what makes the culture and the work. And that's how I run my business. And I like other companies that do that.

(Mark at 00:14:00) You've got a guy who launched a Roadster into space and, you know, makes flamethrowers, right? This is, that's fun. Super fun, right? So, right. Got to have fun.

(Joel Beasley at 00:14:11) So I was having this conversation with Yoav. He's the CEO of a company called Oro, and they do B2B commerce, you know, storefronts, which I'm curious to find out if you connect to the commerce storefront. Well, I guess you would have to, right? But that wasn't what we were talking about. We were talking mostly about open source stuff, right? And I was curious, how has open source impacted your business or how do you think about it?

(Mark at 00:14:36) To be honest, we really don't. Like, I think in general, ROS is a great thing because it's just an enabling piece, right? So robots themselves, not just, I don't think, I think in the future, I look at robots definitely as a commodity and they're becoming commoditized. But when you think about the traditional industrial robots that everybody thinks of like six-axis welding robots, this kind of like big industrial robots, those have definitely become much more of a commoditized thing. The ability to stitch together low-cost vision systems, you can put a neural network on a USB stick now, right? That's been around for a couple years now. And then the ability to access compute in the cloud all on like something like a Raspberry Pi running ROS, and then you can control, you know, you've got this very lightweight, low-cost piece of compute hardware that you can control, like, amazing robots with. It doesn't matter one of these robots, a six-axis robot, whatever. You can find a way to connect it up to it. This is really spurring innovation.

(Mark at 00:15:45) I think it's great. Our software is closed. It's all proprietary. But I think in general, ROS is a great thing, especially when you see things happening like First Robotics. Dean came in and introduced that years back, and it was just awesome. And the things that these kids are doing because of software like ROS, it's great.

(Mark at 00:16:06) It's how we're going to stay competitive in the future.

(Joel Beasley at 00:16:09) Is ROS like a suite of tools?

(Mark at 00:16:12) So ROS is Robot Operating System. It's a standard, like open source platform that you can just download. Think of it like Java for robots.

(Joel Beasley at 00:16:22) Yeah. Got it.

(Mark at 00:16:22) It's an enabling technology. That's awesome.

(Joel Beasley at 00:16:26) I was looking at BattleBots. I had that guy on, the guy that's like the producer of BattleBots.

(Mark at 00:16:30) Oh, really?

(Joel Beasley at 00:16:31) Yeah. Since then, I hadn't seen it in ten years. I'd seen it way back in the day. And then we had them coming on the show, so I had to catch up on BattleBots, which quite honestly sounded like the biggest dream job in the world. What are you doing? I'm watching some BattleBots for work. And ever since then, I just keep watching it at night because the kids love it. And I'm just thinking, man, I want to build one of those BattleBots. Have you guys ever done that? Geek Plus done it?

(Mark at 00:16:57) We haven't. But actually, I started a team a couple years back at iHerb and hired a bunch of people really, really quick. And one of the team building things we did, we actually went to see BattleBots in Irvine or not Irvine, Orange County, somewhere up there.

(Joel Beasley at 00:17:14) Yeah.

(Mark at 00:17:15) Seeing it in person was really cool. It was definitely a cool team building thing. And as far as dream job, like, maybe, maybe. I think building the bots is kind of more of the dream job. Like, if you've got ample time and funds and you can just set up a team of six or eight people and start building these robots for fun, like, that's super cool because putting all this stuff together is—they're remote controlled, so you're really testing your ability to make a machine that can take tremendous abuse. And if you win against one BattleBot, you never know what's coming next. These things have sledgehammers on them, flamethrowers, all this. You never know what you're coming up against. And it's really, really cool to watch that show and see—you watch the bot come out, you're like, oh, that one's definitely going to win. But then two minutes in, one of the wheels fell off because it wasn't built right or his competitor just did something so unexpected. And you're like, I didn't see that coming. It's really a good competition.

(Mark at 00:18:11) I think in the future, I have to imagine in four or five years, they're going to change the rules and they'll have to be fully autonomous, which that's going to be a whole different level.

(Joel Beasley at 00:18:22) I asked him if it was allowed. He said, yeah, they're allowed. But he said one of the hammers on one of them actually uses AI to determine how to position, how to hit the weapon on it.

(Mark at 00:18:32) Yeah, best hit.

(Joel Beasley at 00:18:34) Because if you think about it, they're going really fast. Some of those ones—that was the biggest difference that I noticed from watching ten years ago to today is how quick they are now. They're so fast.

(Mark at 00:18:45) Yeah. Well, we see that everywhere, right? So if you go back ten years, brushless DC motors—when you look at the Tesla symbol, the T above it, it's got a little arc. That's essentially a quadrant of a brushless DC motor. This technology is very, very efficient. And that's how we're getting, between these very capable batteries and this very efficient motor technology, that's how we're getting EVs. Well, that's the technology that's made these robots and BattleBots so quick and powerful because you're able to just transfer that energy so quickly and so efficiently. You go back ten years, we were still using brushed motors, very inefficient. Power density and batteries wasn't close to what it is now. And at least for the hobbyist market, it was hard to find that stuff. Now it's eBay and Alibaba and Amazon. You can find this stuff everywhere, and it's great.

(Joel Beasley at 00:19:38) How far do you get from your core business model with R&D? Because, I mean, robotics is such a huge industry outside of warehouse picking, right? So people are making advancements for EVs and battery density and things like that. So when you're doing R&D, are you playing in those areas? Does it make sense for you guys, or are you focused on other things?

(Mark at 00:19:59) My background is twenty years in R&D and ops. So R&D is really near and dear to me, and I'm definitely well rounded in that. I love manufacturing. I love factories. I'm a factory guy. So I think the short answer to your question is R&D drives commodities. So let's say you're a Tesla or us, right? We have a very specific battery requirement, battery technology. And one of the things that we worked very hard to do is develop a battery partner and a battery spec and then of course the battery itself that can last for many years and many, many charge cycles. So it's a very different battery than a battery that's used in, say, a Ford truck or a battery that's used in an EV or something like this. So we have specific needs, but we don't get into the development of the battery, right? That's a chemistry project and truly a massive manufacturing project because when you set up a battery factory, you're not making a couple of batteries for a customer. That's why when you look at a C size battery, it's a specific C size or AA or like what used to be in your laptop, like a 14870. There's a bunch of standards, and that's because it takes so much development to package a battery and get the chemistry to work and then get it approved from a regulatory perspective and the supply chain to source all the materials. And then of course, the manufacturing equipment—when they turn this thing on, they want to make billions of batteries. So if you unpack the average battery pack, it's actually made up of lots of individual battery cells that are welded together. And so somebody like Geek Plus or like a Tesla, they have specific performance requirements and that translates into chemistries that some really, really bright person or team solves, and then they figure out how to manufacture it. Manufacturing for these things—like a battery cell line, the machine that actually makes a cell like for an EV battery, it's probably the size of 100 feet by 50 feet by 40 feet. These are massive machines. There's just raw material going in and batteries coming out the other side.

(Mark at 00:22:21) So it's a huge investment. But we definitely work with those teams to make sure that they produce a solution that works for us. More often than not, unless you're making billions of products that use each of the multiple cells, you're not going to be able to drive a particular solution. What you need tends to fit into a piece of a puzzle that somebody else also wants because this is where you get the economies of scale. There's a big shift right now to solid state batteries, and the name of the game with batteries is many, many discharge cycles. And then so that defines how many times you can discharge it and recharge it. And then of course, how quick you can get energy in and out. Think of a battery like when you're filling a stadium. It's very easy to pour people into the stadium when it's empty. As it gets more full, people can't find their seats. It's the same thing with pushing electrons into a battery. So there's lots of games that they're playing right now to get batteries that you can charge very quickly, but that don't heat up because when you push those electrons in, they heat up and that's what causes the battery to not have a long life.

(Mark at 00:23:36) So we've seen a paradigm shift in batteries and that's what's made EVs possible. They have this great 300-plus-mile range and thousands of cycles, but it's got to get even better than where it is today. And companies like Toshiba and Panasonic and Tesla themselves are driving that kind of technology.

(Joel Beasley at 00:23:58) What's next? What's going to happen to this? Like, what's coming down the pipeline in five years? I won't even go ten years out because ten years is hard to see, but what's happening in five years?

(Mark at 00:24:08) Well, yeah. I think we're sitting under this avalanche that's kind of waiting to fall, and in a good way, though. We're not going to get dead in this avalanche. That's a good thing. 5G is coming on board. So you've got the ability to have this massive pipeline to the cloud. The cloud's established, but cloud compute continues to grow. The applications available continue to grow. If you think back ten years when you had your Windows machine, it would get a virus every day. You'd have to wipe it and refresh it. Now, get a Chromebook. Who wants to update software? Your Chromebook can last you five, ten years because there's no compute happening there. You just need a good internet connection. So you have the connection through 5G. You've got cloud compute. Edge compute is really becoming prevalent where you've got minimal processing actually at the site but then you do the heavy lift in the cloud and the 5G connection makes that work. And then it's the devices, right? That's where we're really going to see things change because think about it right now. When you bring home a device, you have to connect it to Wi-Fi. And in a short while, you're just going to bring home a device, you're going to scan a QR code and there's no Wi-Fi. It just connects to 5G and then it'll pop onto your Wi-Fi when it needs to, but the connectivity is going to change a lot of stuff. And then battery power density. So as you've got higher power density, which means the batteries are smaller and they last longer. So now you can have all sorts of devices that are going to be built into your clothing. They're going to be built into the most mundane thing, like your blender might have connectivity in it because it wants to know—you may want to be calorie tracking or maybe your scale has connectivity. You just take a picture of what's on your scale and it automatically knows, okay, it identifies it as minced chicken, which is a lot of compute to do that, but because you can connect instantly and seamlessly, it can do that. And now you know there's so many calories. So think about going to the gym. Most gyms aren't connected, but why not? Why isn't your phone just pointing at the thing you're going on? And it knows, okay. And it should be able to tell how many reps you're doing because it's connected to the device itself. So you want to do a training session. You're not logging like we used to have a log book and now you can do it on your app. Well now it just does it automatically. So the connectivity and the compute and the ability to have devices that are just persistent because they have batteries that last days without needing a charge, not a day but days and weeks. This is going to change a lot of how we live. And then of course, five years out, I think autonomy, autonomous driving is definitely going to be a lot more prevalent. You want to go a few more years out and forget about terrestrial driving and driving in cars and start looking at three space, right? We're not going to be sitting in traffic because our autonomous vehicle is waiting because somebody fell off a bridge and it stopped autonomous traffic. No, no, no. We're going to be in the sky. There's no point to sitting on the ground anymore. And I think it's going to be really cool because what are we going to do with all that space that we tied up for roads? And you'll still need roads, but you'll need roads for less stuff. So maybe the 5, which is 14 lanes across, goes back to a three-lane highway and we repurpose that space for green space for you to take your dog for a run or mountain biking. Go do something more creative with it. Who knows?

(Mark at 00:27:45) I think the future is amazing. It's also in some ways amazingly scary when you think about all this data that we just—we produce a lot of data without knowing it and what people can do with that data is great but also just use the chainsaw analogy. Chainsaw was a great invention. If you had to cut wood with a manual saw or an ax before, chainsaw is great but look what they did in the Texas Chainsaw Massacre. It can do bad things too. So data is going to be really interesting and we need to be very careful about the companies we trust and the governments we trust with that data. Anyway, that's probably a lot more than you bargained for with that question.

(Joel Beasley at 00:28:24) No. I love it. You think like I do. How do you deal with the AI stuff? And let's talk about just the United States. One of the things I had heard in—by the way, did you see the Elon Musk interview with Joe Rogan, the most recent one?

(Mark at 00:28:38) No. I haven't seen the most recent one.

(Joel Beasley at 00:28:40) Okay. Two things. First thing is the Tesla Roadster is going to hover. So get ready for that.

(Mark at 00:28:47) That's what—

(Joel Beasley at 00:28:47) He said in the interview. He's like, we're going to be out—

(Mark at 00:28:49) It's going to be mind blowing. Definitely want to see that.

(Joel Beasley at 00:28:51) That's what I was saying. And he said it's going to be out in two years. I was thinking to myself, whoa. You must have a prototype then, sir. I don't know about you, but I don't say that stuff unless I have a prototype of it. But so I'm very excited to see what comes of that. But the other thing he was talking about—but I didn't get a clear answer on, they didn't get a clear answer on was, like, how do you create some government body that oversees AI? The advancement of AI. Have you thought about that at all?

(Mark at 00:29:24) I have, and I don't have an answer for it. AI by definition kind of takes on a life of its own. So now you're basically creating a handler, right? So are you limiting the ability of AI? Are we saying AI can—where do you start to peel that onion? Because I'm less worried about the AI than I am about the people who hold the leash. You know what I mean?

(Joel Beasley at 00:29:51) That's true. What about the AI food data? Right? AI food data. Well, the AI's food would be data. That's how the AI would grow. That's what it would have to get to grow. So how do we—do we have a government agency right now that's concerned with data? Because I keep seeing these clips on YouTube of these really old people who don't even understand how cell phones work interviewing Zuckerberg. And I'm like, guys, this is water and oil. Let's get some smarter people in there, please.

(Mark at 00:30:17) Yeah. Let's at least get somebody who can call bullshit. Because he spent a lot of stuff at those hearings. Anyway, and he got away with it, so good for him. But I think as far as the AI—we give up—so first off, people, when I talk about what do we educate people, kids in school, about their money, about their finances. Well, the currency of today really is your time and your data. I mean, at the end of the day, that's probably the most valuable thing. You look at the big tech plays, it's all about data. And the stuff that happens in the background with people's data, most people have no idea. So I think educating people about what happens with data. What are they giving up and at what cost? I think most people, for the advantage you get from using Google, they would still give up all their data.

(Mark at 00:31:03) And go back, like, I don't know, 18 years. Your cell phone, it was like a flip phone, right? It had no processing. You were happy it could connect. That's it. If you could send a text, bonus.

(Mark at 00:31:14) But back in that time, your credit card company knew a full six months just from your purchasing history that you were going to get divorced before you even had an idea. So they were already having, you know, applying data science to very basic data. The data stream you've produced today is so rich and so full. Like, it's too late. I tell my kids, like, your data is already out there. Forget about it. It's your kids you're going to have to protect because you and me, we've given up our data just from our searches and we're pretty much—they know our fingerprint, whoever consumes data. And you would have a very hard time reeling any of that back in. It's too late. You can't walk it back.

(Mark at 00:31:56) So educating people about what happens with their data and how valuable it is because people make money off of your time sitting on the Internet watching YouTube. They make money off of your searches. They're always monetizing this stuff. So, you know, educating people about that, I think, is really important. And then, you know, having a government agency—I don't know. I have a hard time when people say I'm from the government, I'm here to help. It just never pans out well. You know, so I have no idea. This goes back to the chainsaw analogy. It's really, really good, but it can be really, really bad.

(Joel Beasley at 00:32:32) At least there's some good, you know, OpenAI. There are some good actors out there who are putting money behind artificial intelligence so that if we did have, like, a battle of the Alexas and the Siris, that's like the next war, right? Alexa versus Siri.

(Mark at 00:32:49) I'd love to see, like, one of those two—I won't say which one—go evil and then argue with the other one, you know, like good devil, bad devil. That would be pretty funny.

(Joel Beasley at 00:32:58) Sure. Some creative guy or person has put it on YouTube or something. Yeah. All right. So when this whole COVID thing happened, that just put your business in a frenzy. What happened there?

(Mark at 00:33:11) COVID's been, like, really good to us. I hate, you know—

(Joel Beasley at 00:33:15) Hey, me too. I've been—it's, yeah.

(Mark at 00:33:18) It's driven a lot of businesses to look at their operations very critically and say, okay, our brick and mortar stores are kind of on ice for right now. What do we do? So they pushed a lot of stuff to 3PL, third-party logistics, to fulfill online orders. And then they very quickly swung into action to update their websites, become more present, really push marketing online, get into people's lives. They're doing a lot of kind of guerrilla marketing with, you know, getting into TikTok and YouTube and channels that aren't really the norm. But that's driven a lot of business to their e-commerce platform. That's the goal. And so then, you know, think—go back to where we started, which is, you know, somebody walking up and down an aisle picking at, say, 40, 50 items an hour. By the end of the day, these people have walked, you know, 10, 15 miles in a day. It's not easy. It's not easy to find people to work these jobs.

(Mark at 00:34:10) So, you know, these companies are really under a crunch to kind of modernize or die. And at the start, everybody's like, COVID's going to go away in a month. Like, I thought it was going to peak out and then drag on for a few months and then be done, back to normal. I think a lot of people thought this back in March of last year. And here we are a year later and we're still, especially in the state of California where I am, we're still mostly on ice. You know, there's other states where they're kind of back to normal. But, you know, these national and international players, they're dealing with this major impact to how they do business. And so when they look back, they're like, yeah, we'll just kind of get by. Well, it's dragged on long enough that by, like, June, July of last year, people—you know, clients were coming to us and saying, okay, this isn't really going anywhere too soon. And we were going to update our—you know, we needed to update our solutions in probably two, three years. But now we need something right away.

(Mark at 00:35:09) Like we need an answer quick. What can we do to get more efficiency out of our operations? So a lot of budgets that were on ice suddenly opened up again. And all of a sudden it was like, all right, you just constricted the veins and all of a sudden there's massive flow. So we have this very strong pipeline and our applications staff has been just flat out developing proposals and applications. And those things have really fast-tracked through on our client side and gotten funded. And so we're like, we have robots. Like, you know, we just had a site where we had 28 containers show up of all the robots and racks and everything. We're doing deployments fast. And our systems are kind of cool in that, you know, we don't go into a building and you have to put down miles and miles, like literally 20 miles of conveyor at, like, a thousand dollars a foot and racks that are bolted to the floor and all this.

(Mark at 00:36:03) Like, our robots, we go in, we put stickers on the floor, we create a grid, then we build the racks, and we put up some workstations. The whole thing gets safety fenced in, and it's like, boom, roll in the robots, and you induct your goods, and you're online. Right? And the AI starts doing its thing. We're live in, like, three to five months, whereas, you know, if you were looking at traditional hardware, you basically—it's like an 18 to 24, 18 to 30 month deployment cycle. So there's a lot of money to be left on the table if you don't get your system up quick.

(Joel Beasley at 00:36:35) What does the sales process look like? If I come to you, I—let's say I have a warehouse. Actually, my parents have a warehouse because they own a weight loss/health facility. I'm not describing it well. They would be so upset at me right now. But they own—it's called Peaks of Health and, you know, they've got like a 10,000 square foot warehouse because they do vitamins for their customers. They have doctors on staff and they do all sorts of health care type stuff. But I was just curious. I know you deal with, like, much larger size. But for people that are listening and that might have factories, what does your sales process look like? How deep do you go? Do you actually start designing models and mock-ups? Like, how do you sell a system like this? It's huge. It takes a lot of time and money. What does that look like?

(Mark at 00:37:22) Yeah. I want to dip into something after I answer this, which is that things are going small. But the process is like this. You know, we get connected to a potential client and we either connect to them directly or most likely we use our channel partners. So we have integrator channel partners who already have clients and they're always, you know, kind of out beating the bushes for new clients, right? And so you've got a warehouse. Our channel partner will connect with them. They'll find out, okay, how many orders—you know, they just get some basic data. What's your operation like?

(Mark at 00:37:57) How many orders a day are you doing? What's your SKU profile, which is, you know, how many different items do you have? What's their physical size, inventory mix? You know, what's your peak demand? What's your nominal demand? And from this, we'll go in and we'll just say, okay, this is how many square feet you need, which is almost always like 30% less than what they have. So they can expand in the same building without having to get a new building. And then, you know, we basically cost it all up. And so we give them kind of a rough, what we call a ROM estimate or a rough order of magnitude estimate. And then they, you know, they're like, yeah, we're interested or not, right? We either have capital budget or we don't. It's not what we expected.

(Mark at 00:38:36) And sometimes our solution isn't a fit, right? We are a very flexible tool, but we're not always the right fit. So a lot of times the end customer will be soliciting proposals from a number of different technologies. And very frequently, we're the winner, but, of course, not always, right? So we're happy to have that discussion. If the ROM estimate is accepted, then we do a full detailed proposal, which we do, you know, very detailed CAD layouts, and we do software integration discovery. And then we have a very—we do an analysis to understand the binning, and we've got different robots, right?

(Mark at 00:39:14) So we may also suggest other robots. We've got our pick areas which you see behind me, but we also have sorting robots and we have shuttle robots which carry totes. And then we may bring in other pieces like, you know, six-axis vision-based bin-picking robots. And then we put this whole thing together, give a detailed proposal, and we have a dialogue, and then we get our PO, right? And then after the PO hits, three to five months later, you're up and running depending on the complexity and our—you know, what's our capacity? Do we have robots in inventory? Are we producing for you? Blah blah blah. So it's a very, very easy process.

(Mark at 00:39:48) And we try to make it easy. It's one of the advantages that we see ourselves having is that we're very easy to deal with.

(Joel Beasley at 00:39:56) Nice. Nice. And I like it because all of this business is just advancing robotics. Like your business growing advances the entire robotics industry as well. And I was curious, like, when am I going to—you know, the movie I, Robot with Will Smith? Aside from all the death and destruction, the idea of the humanoid robots being able to interact with them. I see all the pieces.

(Mark at 00:40:19) Like, I've—

(Joel Beasley at 00:40:20) —seen the minds inside of the Siris and the Alexas. I'm seeing the arms come out from, like, you know, the picking technologies. I'm seeing the vision systems improve. It feels like we're just running towards this human. And I was having some conversations. I was like, where are we going to see these first? One person shared with me that we'd probably see it in, like, elderly care first because there's money there. There has to be a business model for the humanoid robot to exist. Like, companionship is not enough of an industry, I don't think, early on to get investors in. But from what my experience has been, just talking around to people, is that they think the industry that'll move the humanoid robots farther forward will be, like, elderly care.

(Mark at 00:41:09) Yeah. It's possible. Anything interacting with humans is tough, right? Because there's just high risk of hurting somebody. So you really don't want to try out your first pass at technology in a situation where the optics of accidentally killing grandma are not good, right?

(Joel Beasley at 00:41:25) Right.

(Mark at 00:41:26) And it would truly be an accident, but that kind of stuff could happen. But you're right. So first off, we're sitting at this awesome confluence of technology. Again, 5G's putting compute in the cloud. If you've got a robot, you don't have to dedicate a lot of energy to processing or a lot of weight for the battery or for the compute module itself on the bot. You just give it connectivity. Let it communicate via 5G to the cloud that does all the heavy lifting or to the edge and then the cloud, right? So when you have that data, right?

(Mark at 00:41:55) So now you've got a robot, right? Think about self-driving cars. Today, you're driving down the road. The car is totally on its own. It's got a lot of compute on board. So a lot of your mileage is going to compute. If we were doing compute in the cloud, you're driving on the road along with 40 other EVs, autonomous EVs. They're mapping. That map should be shared. And so after a day, you've got an extremely precise map of everything in three space. But let's say, you know, a fridge falls off a truck. Well, your car doesn't have to worry about discovering it because a car two minutes before you discovered it and you got to share data. Well, it's the same thing when you think about humanoid robots, right? All of that path planning that happens to move the limbs or move the fingers or interact, let's say, with a coffee pot because you're giving grandma a coffee, right, you're assisting.

(Mark at 00:42:43) Once it's learned, the machine just continues to improve the model. Instead of it being a discrete model that lives on that particular bot in that particular house. That's where things get really exciting in terms of penetrating markets, whether it's, you know, humanoid assisting robots or maybe it's just a hotel reception or a bank teller robot, right? That kind of stuff. You've got—like, I love RealDoll Robotics, right? The folks over in—they're not far from here in California. They do—what's the name of that company?

(Joel Beasley at 00:43:18) RealDoll. RealDoll, right?

(Mark at 00:43:19) So RealDoll has a lot of tech built into creating a presence, right? And that's an art, and it's a technology. So then if you put the underpinnings of an elegant machine that can wear this sort of skin, and then you put the—you know, then you get power density and very fine motion control through using brushless DC motors and very fine gearboxes and nitinol. Like, there's all sorts of technologies you can bring to making motion appear very human-like because the compute that we do to do this kind of stuff is amazing. It's getting a computer to do is hard. But once you've taught a computer and you can share that information between—now, all right, it's quite easy. And the fact that you can share it means you can make a model that's uniformly bad. Hey. No problem.

(Mark at 00:44:10) Because it gets uniformly better as you teach it, all right? So that's awesome. And so I definitely think we're going to see the humanoid stuff, right? Probably my favorite robot company, aside from my own Geek Plus, right, is Boston Dynamics. These guys, the stuff that they're doing is just—they're truly, like, bleeding edge in terms of being able to do compute that is super, super elegant and then translating that to motion.

(Joel Beasley at 00:44:41) Have you bought one of the Spots?

(Mark at 00:44:43) What's that?

(Joel Beasley at 00:44:43) You can buy them for, like, I think $75K now, the Spot.

(Mark at 00:44:47) Spot's super cool, and it's a really useful app—like, it's really useful. But, like, their Atlas. Like, you can't help but just be like, holy shit. Did that thing just do a backflip? Wow. Right? And then you watch the dynamics when they do it in slow motion. The dynamics of the recovery when it lands, it's just elegant. From an engineering perspective, it's a work of art. And so, like, you know, they are at—we're really at the start. Like, it's taken a long time, but as we go further down this path, we accelerate exponentially.

(Joel Beasley at 00:45:15) Yeah.

(Mark at 00:45:15) Because the compute stuff going to the cloud, that's going to be like a step change in what happens. And then, you know, batteries still have some catch up to do. Motors are getting really good, but they're going to get better in different types of actuators. And then you start to bring in mass production to take the cost of a Spot, which is—let's say it's 75K, right? But let's say that thing cost $5K. Wouldn't you have a Spot to be your security guard at your building? Heck, yeah. Sure, I would, right? And so—

(Joel Beasley at 00:45:56) I don't have a Spot to hang out with me and help me host the show. I'm going to—

(Mark at 00:45:58) —have a Spot just to go get me a drink. It'd be as cool—for $5K, right? I mean, who wouldn't want to do that? And that's where we're going to go with stuff like Atlas. Like, Atlas is right now, you can't—you can't even do it because the government wouldn't want you to have one, right? They don't want to let people have that kind of fun. They want to save that for the battlefield. But it's going to come. And when it does, it's going to be super cool. When you come home and your Atlas Light—built, you know, made you dinner and vacuumed. Right? Forget about the Roomba. I want this thing wearing an apron and—go take my Dyson and vacuum for me.

(Joel Beasley at 00:46:33) Right. Have you seen the robot hands that make the food too you can buy?

(Mark at 00:46:37) Yeah. Yeah. So I mean, think about that, like, in your restaurant and then go back to, okay, I'm fighting for $15 an hour. Oh, but I can put this thing in my restaurant, and it's gonna attract people just because it's there and they can watch it. It's already in Asia.

(Mark at 00:46:51) Like, I go to Asia, I can get tea made by pretty traditional six-axis robots, but they're so entertaining to watch that I'll pay like an extra 50¢ for it. But it's not that expensive. And these guys are making money hand over fist because they run 24 hours a day. And this is the start. They're using, you know, expensive robots.

(Mark at 00:47:10) When you start purpose-building robots, I was at Mattel and the toy industry is extremely competitive, right? These guys, they do magic to make Barbie and Hot Wheels and all that stuff. And when you talk about the technology that goes into making these toys, it's amazing. And they fight with the fact that you're making something that is razor-thin margins, very competitive.

(Mark at 00:47:32) So if they can bring in automation, it's great. But when they look at, okay, what if I wanna do... like I wanna reduce the cost operating my factory. Well, put in a vision system. Ten years ago, it would've cost you, or even five years ago, it cost you $100,000 to put in a vision system that had enough accuracy and acuity to inspect paint on a toy car. Now, you know, we did a project where a team at SDSU put together a neural network on a Raspberry Pi with a camera that had capability to do this kind of inspection for like $100.

(Mark at 00:48:07) You connect that up to the cloud. Now your whole factory can be reporting data real-time on what's the quality, what's the throughput, right? So we're gonna see a lot of changes. And arguably, unless we get some really bad actors, it's gonna be very cool.

(Joel Beasley at 00:48:22) Nice. Dude, I could talk to you all day about this stuff. But as we start to wrap up, I mean, your business is probably growing. Are you recruiting? Are you looking for talent?

(Mark at 00:48:28) Yeah. Yeah. We're hiring. Check our page. We've got positions open around the world, not just in the States, around the world. So we're about 900 people now, and we have some extremely aggressive numbers just as far as putting robots in the field in the next two to three years. And we're definitely staffing up to get behind that. So whether, you know, we've got manufacturing, we've got field service engineers, deployment, application.

(Mark at 00:48:58) I mean, you can come to this company and learn a lot too in terms of, you know, technology. Plus we're a global company. So there's a lot to learn in terms of working with the global culture.

(Joel Beasley at 00:49:11) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.