Avoiding Automation Failures with ATI Cofounder Bob Little

0:00 Well, I'll just buy humanoid and he'll do everything for me.

0:03 Humanoids just left the research lab. What humanoids can do today is not much. Google Deep Mind said it's gonna take at least three years before we can show you something of value.

0:13 I'm seeing these robots doing amazing things. But then when I ask more details about how do I get from the show floor to my factory floor, you can sit down and learn to play the violin, but you cannot play the orchestra.

0:27 Automation is a development of an orchestra. Trying to get a robot to do what a person is doing is the fool's errand. When you have a person doing it, you will not believe how much that person is accepting change without ever telling anyone. If people feel that the automation is displacing them, they can actually hamper the solution.

0:49 We want robots to be intelligent enough to perceive, reason, and perform without programming it. An executive who says, I hear everything about these humanoids, take the word humanoid out, put the word physical AI in. Amazon is one of the biggest robot manufacturers out there, and you can't buy any robots from Amazon.

1:08 So the question is, are we going to go the Amazon route, which is going to hold back everybody else?

1:14 Robot safety isn't paperwork you fill out at the end of an automation project. It's part of design, integration, and operating the system from the beginning. For more than 35 years, the A3 International Robot Safety Conference has brought together the people developing robot safety practices with the people applying them in the real world.

1:34 This year's conference is November 3rd through the 5th. And for full disclosure, IRSC is an A3 event, and A3 is who I work for. See you in Detroit. Welcome to the business of automation. I'm Aaron Prather, the director of market intelligence here at A3, the Association for Advancing Automation. And we have a special guest today, Robert Little, who many of you know as Bob, is really popular on LinkedIn right now, to say the least.

2:05 I'm definitely always reading his stuff. But the whole thing is Bob's coming to us with 40 years of experience in the automation field with robotics. He is not just only an amazing person, he is an Engelberg winner, he is, as I said, 40 years of experience in this field. He founded ATI. Bob's been in this industry for so long.

2:26 Not only has he had some amazing successes, and I mean we could talk about his successes all day long, but we brought him on the show for another reason. Because he's seen a lot of failures. And like I have in my 25 years of experience in this field, uh, have seen a lot of failures. Bob's seen even more because he's actually been out there selling the stuff, getting the stuff in the in the factories and warehouses and everywhere else.

2:51 So, Bob, welcome to the show. And if you'd like to add anything on what you've you've seen over 40 years and your history, please do so.

3:01 Well, we're in a very interesting time period where robotics is seeing a technical change. Um, AI is the driver. And um I can remember dealing with robotics in the 1980s. And to be upfront with you, before AI, there wasn't a lot that has changed. The robots that we know in the 80s got improved on reliability, quality.

3:31 We've improved our safety standards, but honestly, the machinery and how it functions and works works very similarly. And there's nothing wrong with that because the these robots have been extremely uh reliable tools for manufacturing. The problem is that they're difficult uh to use and implement because of how rigid they are.

3:56 And rigid I mean by once you write a program, sure, you can rewrite the program, but the task of rewriting is so expensive that most people avoid it at all cost. And uh so today, physical AI is now becoming a name you hear more frequently. It's got a ways to go, but right now the AI can write the program, and you need to be the director, the manager of the robot, not the move here and move there person.

4:31 And that change is going to uh make robots more usable in the workplace. And boy, do we need it because manufacturing is going through a renaissance today. Um, we are trying to make more domestically than we ever have.

4:47 I've seen that transition myself of where it used to be where, yeah, you just programmed and the robot did what you told it to do, and that was it. So I guess one of my first questions for you is someone that's been in this industry and seeing this transition. Because I know when I go to a trade show, when I go to automate every year, or I go to any of the other shows, I'm seeing these robots doing amazing things.

5:13 But then when I ask more details about how do I get that robot or that automation from the show floor to my factory floor, there seems to be a much bigger jump than what it looks like. Can you talk about that? Of what what does it really take to get a robot or automation from that show floor over to the factory?

5:37 I think that people can be enamored by robotics. I'm enamored by them. I I think they're fantastic. I love machines that you can program and fiddle with. The complexities, though, of a manufacturing environment require 10,000 other things that have nothing to do with robots. And you can sit down and learn to play the violin, but you cannot play the orchestra.

6:04 Automation is a development of an orchestra of many moving parts that function together. And when something's out of tune, it's a big mess. So robots play a role in the automation on the factory floor. But the first thing you got to get right in an orchestra is less the automation and more the manufacturing.

6:26 You've got to know your process and you've got to optimize your process. Trying to get a robot to do what a person is doing is a fool's errand. It's a rare circumstance where you're going to want the robot to repeat a person's doing. Instead, look at the manufacturing line and look at the productivity and all the KPIs that are involved with that, and then rewrite the script from a bigger picture of what you want to see the end result to be and how robots fit into that.

7:01 That's a fascinating take because I think that's one thing that's really missing out there when I talk to the C-suite folks. They're always like, oh, well, we'll just buy a robot. And it's like, well, what is your process? What is the steps that actually makes the product? Are we missing out on that?

7:20 It's possible. I mean, ignorance, you know, is nine-tenths of the law. I think people take for granted what they know actually. I think they they they functionally know, but they may not be able to put it into words or writing. Um, they know what a good process is and a bad process is, and they have the intelligence and the capability, but they've never been asked to sit down and write the music.

7:43 And and and and that is a little bit more difficult to do, to put in what you know on paper. And that's what you're gonna need to do if you're going to evolve automation. You're gonna look at your manufacturing process, you're gonna take your knowledge, and you're going to rewrite it. Too often uh an integrator comes in, they're invited into the manufacturing process, and they're asked to see how they can improve or automate.

8:10 And they're gonna be very reliant on that integrator or a series of integrators they bring in because they're looking for you know competitive bids. You are in the hands of that integrator, that person's knowledge. You need to be part of that. You cannot hand off to the integrator and say, make it happen.

8:31 You need to be an every step of the way shadowing their thinking and making sure that your manufacturing is the way you want it to be, not how the integrator thinks it should be. Grant you, the integrators have a tremendous knowledge base they're going by because they've seen a hundred other manufacturers and they're going to deploy their information into that.

8:57 I was going to say prejudice, because in many cases, we tend to get prejudiced about how things should work based on a couple of things we learned in the past. But you, the manufacturer, have your own prejudices. So you need to come together with the integrator and bring that knowledge base as a one, not relying on the integrator to do it all for you.

9:18 That brings up a really good point that here at A3, our standards team is actually developing what they're calling part threes of some of the standards. So part one of R15.06 for robotic arms applies to the person actually making the arm, of like, these are the safety items you need to take into account as you're making this actual robot.

9:39 Part two, as you brought up, is the the integrators of like, okay, now it's time to integrate the robot into your factory. And I've I really give a lot of praise to our standards team of recognizing there needs to be a part three for end users. And what I think one of the key things that's in that new standard is you, as the end user, need to work with your integrator, exactly as you just said, Bob, of what is going to work for you?

10:06 What are those other risks? What are those prejudices? I think actually that's a really good word to use here, that not only you have as an end user, but what the integrator is bringing to the table, because as you said, they've learned from so many other areas, but they haven't learned your operations fully, maybe.

10:23 And you guys need to have that conversation. You need to actually talk this stuff through. And end users need to be an active per role in all this. So I think that's an excellent point, and I think that definitely would cut down on failures. Would you agree on that?

10:39 I'm gonna go full Ronald Reagan on you. Okay. Trust but verify.

10:45unknown Yep.

10:45 The relationship you have with the integrator has to be a trusting one. They gotta trust you, and you have to trust them. If you start to see that trust unfolding, get back to a place of trust or change integrators, or the integrator may need to fire the manufacturer. You've got to have a trusting relationship.

11:08 Um, yes, there's going to be a contract on this work, but if you're relying on the contract, you're going to be in big trouble. You have to have a trusting relationship. So the the point is that, yes, I want to trust that the integrator is giving me everything I need, and the integrator needs to trust the manufacturer's telling me everything I need, but we need to verify.

11:31 We need to verify uh through KPIs, you know, through measurement. What is the what is a what does success look like at the end? Let's agree on that so that when the day comes we have a written acceptance, we go back to those measured rules that we agreed to at the beginning. This is what success looks like.

11:51 Now we'll tell you, trust also involves making changes, and because you learn as you go. And and that's one of the sticking points I have seen is that these changes get people stuck. Do not be so rigid that you don't recognize that, hey, uh, we can't get there from here. It it ends up that maybe something I promised you, I have learned enough information that the process will not accept it.

12:21 Be okay with backing the bus up for a little bit, making the necessary changes, and there's going to be an expense to that, and then move forward.

12:31 Is there anything else we're missing here, Bob? Is there technology questions we're missing? Is there workforce expectations? Is there other things that could really derail a project outside of the integrator? Other things that we might, as the end user, have control over.

12:49 I would say it's like a minefield. We can't see the problems until we step on them. Manufacturing in many cases is quite complicated. When you have a person doing it, you will not believe how much that person is accepting change without ever telling anyone. They modify their processes automatically without command.

13:13 Automation requires everything to be written and performed functionally. And when it doesn't get what it thinks it should have, it fails. That's a mine. So do not expect to automate 100%. Um there's just some expensive processes that are very difficult to understand. When you have something coming into your manufacturing line where it has variability to it, um, it's very possible automation can't pick up on that variability.

13:43 Um, and that's where a human operator comes into play. So how the humans, the labor is working with the automation, they're taking the variability, uh, the non-standards um and the complexity out of the automation into a human labor, uh, and then they're functioning and working with the automation. So be careful.

14:10 Uh don't do not push too hard for full automation. A good integrator will tell you that and they'll show you the expenses of automating something that is is difficult. If you're willing to pay for that, that's another story, then then then fine. But understand that the more complex something is, the more likely that there could be uh failures that are unseen.

14:33 I think one of the biggest things I always stress with folks who are starting their automation journey and they're trying to figure out where they can start is I always say start with the lowest hanging fruit that's out there. If at the end of the line you're supposed to be building pallets and you don't have a robot doing that, there's your low-hanging fruit.

14:51 But I think you you brought up this interesting point of just how much knowledge is in our employees' heads of how they have to perform a task, knowing all those nuances that are out there. And I think we're gonna talk about AI more uh as we move through here this conversation, but I think that was one of the things I felt was really important was getting those nuances from employees so that if you did, if you were going to automate that process, it's like, oh, this is how they had to deal with this.

15:29 But then I think there's the other question of uh what happens to the workers after we do this automation? Should we be accounting for the for them as well?

15:38 You need buy-in from everybody, you need it from the labor that's working online, from the quality, safety, the operational managers, the executives, the engineers. Everybody needs to buy into this automation. If people feel that the automation is displacing them in their career, they are not all in.

16:02 They can actually hamper the solution. So you want to get everybody all in. You need they need to know they have job security, uh, that the automation is going to make changes, but it's going to be opportunities for people to level up to a higher position, possibly if they have the skill set and they want to do it.

16:24 But that anybody is that is displaced because the automation has a home somewhere else to functionally perform. Um, that I think is is a very important thing. I've seen what I would call successful companies do. Companies that are purposely displacing uh workers and not uh per performing well with them creates a morale issue and a negativity, which makes automation that much more difficult to be successful and future automation projects be successful.

16:57 The best stories I've read are the ones that consider the aftermath of how automation actually increases employment through um productivity because a productive factory makes more parts, gets more orders, and then increases the size of the facility. And that's what our goal is is we want to perform better, which means that we will get grow and be bigger and need more employees.

17:28 So don't think that automation is displacing people. Automation is putting value into the operation. People are finding uh higher places to move up, or they're finding different places to move that gives gives value to the company.

17:45 Now, I I know you're a big proponent of reshoring uh manufacturing back here to the United States. Is there anything that you're seeing as as you look at some of these reshoring efforts where you're like, I like it, but my concern is this?

18:03 I'm worried about the supply chain that supports the reshoring. Jeff Bernstein just wrote a post about this Chinese humanoid manufacturer just finished their manufacturing line to make humanoids. And then, you know, the complaint was, hey, we beat Tesla, you know, on being able to manufacture, produce a manufacturing line for humanoids.

18:22 I'm like, who cares? At the lowest level I have is to worry about the manufacturing line. What am I worried about? Well, how is Tesla going to support the components that go into the in into the humanoid? Are they going to depend on China for sensors and actuators? No, they're not. What does that mean?

18:42 They need to find domestic support if possible. If not domestic, then you need to find a allied country that you can trust economically to be there for the next 50 years that's going to support. So right now we're extremely dependent on Chinese products because they're so cheap, they're so inexpensive that we become dependent on them.

19:08 And it's it's killed off a lot of component suppliers. So what I would like to see, and I think is happening, is that sensors and actuators and products are being found domestically, or there's reshoring of those of those products coming as well. So when I see a very large plant like Haas, for example, Haas Automation is producing a brand new plant in in Nevada, but yet one of the questions I read in the article about it was well, they're still dependent on castings from outside the country.

19:49 Now, you brought up humanoids, and I get asked questions from a lot of folks, the decision makers that are trying to figure out where to start an automation, and a lot of and I ask them like where they are, and some of them have some automation, some are just literally getting started, but they say, Well, I'll just buy humanoid and he'll do everything for me.

20:11 And that's it. I'm done. What is your pushback, if any, on that kind of statement?

20:17 Humanoids are what I would call just outs just left the research lab. I mean, just left. Uh, we're in what I call a development phase. It's going to be going on for years. In this development phase, we're going to be experimenting with manufacturing. We'll be experimenting in the home. We'll be experimenting in the retail setting.

20:38 It's all experiments. We're undergoing a massive experimentation. So, what humanoids can do today is not much. You know, the the biggest application I've seen is they can move part A to part B. Okay. Okay, any robot can do that. Um we haven't really got humanoids developed enough to be to be ready. Now, having said that, I am confident that uh humanoids will have a strong place in our future, but when is hard to tell.

21:18 You know, I listened to who I would consider uh uh an expert, Google Deep Mind, who's really involved with with the software that makes these things tick. And they said to the public, hey, uh come back in three years, because it's gonna take at least three years before we can show you something of of value.

21:38 In the meantime, we're gonna show you all these experiments we're doing and how lovely they are, and uh and that's gonna be great. So we we are dealing with uh time as a necessity. And three years, by the way, is a guess. It could be longer. I just don't think it's gonna be shorter, is is the bottom line.

21:57 Now, Aaron, I've read some of your Writings on humanoids, and that you talked about the valley of death, and that's where a product has been hyped up, and then a reality check sets in, and there's a fall, and then you lose a lot of uh companies, and then it starts rising but at a slower rate. Um, so a lot of products have fallen into that.

22:22 It's it's very possible humanoids will fall into that as well.

22:26 So your recommendation right now to any CEO saying, I'm waiting on humanoids, is you might want to actually get started now on other stuff, and we'll see how the humanoids play out.

22:36 Why should we wait when we when physical AI is this is the real solution that we want? That is, we want robots to be intelligent enough to perceive, reason, and perform without programming it. And this not just robots, but it's the conveyor system that is talking to my robot controller to tell me when a part is there.

22:59 It's the PLC's programming that the ERP system can grasp and get information so that everything can be directed into manufacturing to speed up, slow down, move from part A to part B. All of that AI can sit down and control without having to sit there and program every single line of code. That is where the reality I truly see coming before humanoids.

23:30 And physical AI is going to bring us and make my FANIC robot and my universal robot talk to each other, perform with each other, and get tasks done without me having to do all the work. There now humanoids got to compete with that. So I can I should be able to wheel up my cobot from either of these two companies and put it in front of the job that I wanted to do, like loading or unloading machines or doing some assembly task, be able to show the task, you know, to the cameras, then it goes to a trial period, however long that takes, and then next thing you know, it starts performing it.

24:14 That's not a humanoid, it's a camera and a robot arm. That's where I see things going more short term today than long term. And so an executive who says, I hear everything about these humanoids, take the word humanoid out, put the word physical AI in, and then go talk to your standard companies like Universal and FANEC.

24:35 Where are you on the physical AI journey? Tell me more about how I might be able to implement this high flexibility I keep reading about.

24:44 Pure gold there, Bob. I I think that that's something that really needs to be reiterated over and over again humanoids are just one physical AI thing out there. There's so much more out there. And I know one of my last installs before I left left the factory warehouse was a physical AI robot picker. Just it picked up packages and moved them up, moved them along.

25:11 But that was vision, AI, and a robotic arm. It was not a humanoid, and it still brought us value. And I fully agree with you on every point you just made on physical AI. It's out there, it's going to allow you to do really interesting deployments, and they don't have to be humanoids. But where's the risk here?

25:32 There's physical AI is great, but we hear on the other side of AI, Chat GT, Claude, and all those still hallucinate. Should we be worried about hallucinations when it goes into the physical world?

25:48 Yes. Um, I think the worry is that uh physical AI is at the very beginning, it also is going through an experimentation phase. It's just going to be more rapidly developed than the humanoid uh uh sector, uh, because you're gonna be using what I would call standard off-the-shelf products, and then you're gonna combine uh the physical AI software with it.

26:10 But where you need to worry is safety. Safety is a big deal, and and what's gonna happen, what I see happening, is that we're gonna be starting to try to rely on AI for safety. Then, safety standard people, how do you measure that? That you know it's reliable. And I don't not have the answer for the audience here, and I think it needs to be developed.

26:33 But there, but look, driverless cars are robots. Driverless cars are making safety decisions every single day in the millions of calculations they're doing, and we are starting to approve them to go on the road. So if driverless cars are a robot and they become safe through AI, there's no reason why an industrial robot working on a line without a fence cannot work around people close by using the same conceptual physical AI driverless cars are using.

27:12 It's possible. Um, so the other the issue is you know, um that is holding back physical AI is physical AI needs hard data. It does not have an encyclopedia to read. It needs to see um millions of hours of manufacturing vision and touch and feel sensor feedback to know what a good process is and a bad process is.

27:39 It takes a lot of time, it's being worked on. Um I see a lot of teleoping to do it, I see a lot of simulation, so it's ongoing, but hard data is holding physical AI back as well.

27:55 So we talked about data on the show just a couple days ago, and one of the things we we discussed is who actually owns that data. Who, I mean, especially if now that robot is doing work in my factory or warehouse and is starting to collect data? Is that my data? Is that the vendor's data? Is that someone else's data?

28:17 Yeah, I mean, the bottom line is it is the the manufacturer owns their own data. I'm just bottom line. I'm just gonna say that. Would the manufacturer be willing to give up their data into a system that overall would make a product much, much better? And then they benefit from having that product come into their factory.

28:42 And then what the person who collects that data can say is your data is not going to be shown to be yours. It's going to go into what I call a neural network collective. And it it'll be such that we're not going to know how your process works. Nobody's going to know how your process works, but the neural network will know it will use that and it'll help it become autonomous.

29:09 I think some manufacturers will sign off on that as long as their data isn't directly used but is indirectly basically helping a neural network. But that's a big ask. And so it is a holdback. But what's happening is there are companies out there that are for a living are performing tasks through teleoperation so they can learn how to do it themselves.

29:34 So you have people using robots to weld, and there's a person behind it actually helping perform that. Um, there's a person doing assembly task, teleopting the robot to do it. There's a lot of teleoperation out there that is performing this. Now, who's going to be a big winner in this data collection is when the manufacturer owns the technology that they want to go in their plant.

30:05 So we talk about humanoids. One of the benefits Tesla has is they'll use those humanoids and to perform their own data in their own manufacturing network. Hyundai is another one for Boston Dynamics. So manufacturers that own the technology are going to be potentially big, potentially big winners.

30:27 So you're saying that we are going to see companies that today are not robot or automation companies potentially become robot automation companies in the future if they go that that route like Tesla and Hyundai, and the list keeps growing.

30:42 Yes, and no. Um Amazon is one of the biggest robot manufacturers out there, and you can't buy any robots from Amazon. They do it all in-house and they keep it in-house. And that that hurts warehouse logistics. Because if we had Amazon technology that would go out to everybody, we'd be so much further ahead because Amazon does great work in robotics.

31:05 Tesla is likely, you know, I don't, well, they say they're going to sell the Tesla Optimus, but you know, are they going to help their competitors at Ford, you know, with physical AI? I just have my doubts about that. Meaning they're going to try to keep as much in-house as they develop themselves to make their own cars better.

31:27 And Hyundai is the same way. Hyundai's got um so much manufacturing that they need to optimize through physical AI. Uh, I don't see Hyundai, you know, helping out Toyota. I could be wrong about that. So the question is, are we going to go the Amazon route with this, which is going to hold back everybody else uh and but make companies great, but individual companies.

31:52 Um my hope is, of course, is we have a breakout where uh everybody benefits, all the manufacturers benefit from the advanced technologies coming out of these companies.

32:02 Yeah, it'll be interesting to see how this plays out because as we all know, you told it beautifully, is when Amazon uh acquired Kiva, they Kiva had other customers besides Amazon. And Amazon, I think, gave them a year or so. Like, you have a year to get off of it because we're bringing everything in house.

32:23 And and then something interesting happened, and this little startup called Locust kicked off, and they sort of became that backfill for where as Amazon pulled Kivas out, it seemed like Locusts started moving in. It would be interesting to see if that plays out exactly as you you described, is could we actually have new companies that rise up as some of these bigger guys pull back in in-house into themselves?

32:50 So going back to how we started our conversation, are there still a ton of uh is there still some use cases out there that don't need physical AI right now, but do need the automation that you think is some low-hanging fruit that is out still out there?

33:05 Absolutely. I I believe there are. Um, you know, the uh when you have a what we call a repeatable process, we have the technology to automate it. You need two things. You need a repeatable process and volume. Because if you're gonna spend uh $100,000 on a robotic cell, you need to know that the volume going through that $100,000 being amortized and paid for within a reasonable amount of time.

33:33 So those still exist. We've also made major strides with autonomous mobile robots, AMRs. They're still young. There's still a lot of uh warehouse logistics that could use their benefit from them, and they're not physical AI. What makes AMRs really good is we've gotten stronger machine vision, and so it's making AMRs much more functional and and able to perform.

33:59 Also, machine vision is going a long way. Take a company like Path Robotics. Path Robotics was not using AI to perform their welding, they were using really smart algorithms with machine vision to get that done. Um they recently have now developed a physical AI model, uh, but it's young. I mean, it's it's it's not that old.

34:25 Uh, and so it's gonna further improve their their ability to manufacture or or or weld complex uh systems. So the point is, is there's still plenty of fruit out there. Um having said that though, the biggest thing is is there's so much. It's like there's so much more that we I think we just have to do.

34:50 We're not we're not gonna get the manufacturing renaissance solved with current technology. We've got to get physical AI developed.

35:01 So as we wrap up, if you were advising a CEO right now, and he has a factory and he's building different widgets, and he only has maybe tinkered with a robot. Maybe he had a welding robot at one time, or he had a pick and place robot on a CNC machine. What would be your recommendations for what he might want to look at or how he might want to proceed through some of these things?

35:34 Well, number one, go to some trade shows, go to automate. I highly recommend Automate is a fantastic place to have a hundred conversations in in a single day that might get you what you want, which is information. IMTS is coming up. That is another good place. It's not as good as Automate, but it's a good place to be able to talk to a lot of these uh robotic companies.

36:02 Um the big boys are gonna be there, like you know, Kuka, Fanic, Universal, but also uh you're gonna have some integrators there. I I think creative is there, mission uh design is there. Um so talk you need to have these conversations with with these integrators and with these robot companies, and then you're gonna see some startups coming out there that are advancing the technology pretty rapidly.

36:28 And a company like, for example, uh in bolt, um, which which FANEX showed at Automate, uh, pretty impressive, impressive stuff, uh, worth a look. They're they're gonna be at IMTS as well. Um, so, anyways, um information gathering is is is is is uh very much needed. Um you know, be careful about relying on Chat GPT to get that information for you because the technology is advancing fast and you don't know what's real and not real.

37:02 So having these converse direct conversations at trade shows really helps. Uh bringing sales staff in, I would bring integrators and and have them converse with you on what is real and what's not real. But remember, integrators are gonna have their own prejudices. So that's why you have to talk to uh multiples and not not just have relying on one.

37:26 We are gonna be in Las Vegas next year for Automate uh 2027. So definitely mark your calendars now for that one. So thanks for the shout-out on that, Bob, because we know Bob, you'll you'll definitely be there with us in Vegas uh next year. So Bob, I um I greatly appreciate you coming on the show today.

37:43 I know we're gonna have you back for future conversations around topics like this, but I think we we've given a lot of folks a lot of information on how to get started, how to miss some of those pitfalls out there, and also what's just coming down the pipeline. Is there anything you'd like to say before we uh close this out?

38:01 Every day I read about a new manufacturing story where they're rebuilding, reshoring, um, or just modernizing their facility every single day. But number two, every single day I read about a new physical AI solution uh that it makes the process of automating better. So it's happening in real time. And so keep an eye open.

38:29 Excellent closing thoughts. So, folks, thanks. We would we greatly thank Bob for coming on the show today. This has been the business of automation. I'm Aaron Prathur, Director of Market Intelligence here at A3, the Association for Advancing Automation. We'll see you next week.