Engineering the Next Species of Robotic Coworker

Hey, everyone. My name is Mary Frame. I'm the senior director of product for Spot at Boston Dynamics. And welcome to today's webinar. We're we're thrilled to have you here. Today, we're going to be discussing how companies and industries can prepare for the next wave of robotic innovation and automation. And you know, as part of the conversation, we'll be discussing ways that institutions can prepare for a world where classic specialized robots and general purpose robots and humans all work together. And I'm very lucky today to be joined by Sharon and Amanda. Hi everyone. I'm Sharon Oluma. I am a product manager on the warehouse robotics team here at Boston Dynamics. Hi, everyone. My name is Amanda Ganano. I'm the associate director of product for Orbit. Orbit is Boston Dynamics' software platform for all of our robots here at Boston Dynamics. I thought it would be great to get started here today to talk about Boston Dynamics robots. We're framing this as a way to introduce new species, but we already have some species out in the world that are working with people. Sharon, do you wanna go first and talk about Stretch? Yeah. So Stretch is an autonomous mobile industrial case handling robot today. Stretch is working at customer warehouses around the world unloading floor loaded cases from trailers. Soon, Stretch, will also be doing performing mobile palletization of cases, not just for applications like inbound, receiving, but also building palletized orders for outbound order fulfillment as well. Yeah. And Spot is our agile mobile quadruped. So it's a four legged robot. I like to say it has, four legs and one heart. It's got three main, use cases that we see in the market. The first is industrial. So in this use case, our customers are using the robot fully autonomously. So there's no remote control. The robot learns its job and repeats its job over and over again. It's dynamic. It has athletic intelligence. It'll avoid obstacles. But really what it's doing in these industrial use cases is collecting enormous amounts of data about a site. And that's data that your teams need to do their job better. So it's a tool to help humans. And it does things like it tells you if there's a puddle on the ground. It tells you if there's a piece of equipment that's overheating. It uses all of its state of the art sensors to build a virtual representation of the world and what's in it. So that's industrial spot. That's our first major use case. The second major use case is in tactical response. So that's those are situations where you have very high stakes manipulation needs. So these are our first responders who are using this robot. They are remote controlling it. So they're driving the robot into dangerous situations like EOD, explosive ordnance detection situations, or clandestine labs, or disaster areas. And they're using the robot to get situational awareness on what's going on, and in some cases intervene if there is something dangerous there. And then the third use case of Spot is for research. Now, what research has looked like in robotics in the last few years has really changed. So the first round of these research style customers, they were primarily academic, pushing the limits on what a robot can do, adding new capabilities to the robot. But now we're seeing more and more of an innovator class that's focused on adding agentic AI tools into the robot, connecting it to agentic systems, which is going to get that robot even further along the path toward becoming general purpose. So those are our three main use cases for Spot. And I would add that both Spot and Stretch use Orbit, both to exchange that data between our customers' external systems like WMS and MES, and the robots themselves. We actually just launched or announced our performance dashboards in Orbit for the stretch trailer unloading application. But when we launch new applications like case picking and order building, we will leverage Orbit even more for that work orchestration and coordination, as well as exchanging that robot status information and telemetry data and alert information to the people, like Mary mentioned, who may need to act on it. Yeah. That's right. Robots are great, but you actually need to have that robot brain that is controlling everything that those robots do. So Boston Dynamics is known for the hardware, but really what powers everything is the software in orbit. Yeah. And I think across both hardware and software, it's an incredible moment in robotics, in research, in the industry, and where we're headed. Now, we've talked about spot and stretch. We have thousands of robots deployed out in the world. But we're at this moment now where we're looking at the next generation of robots and how we integrate AI. But we've been doing this for a really long time here at Boston I m x. Right? AI is not just a thing that we're like, oh, we're gonna think about it now that everyone else is talking about it or it's in the public lexicon. So I I would think it would be great to talk a little bit about how we're using AI today with our robots and how it's been kind of part of our research process to start. Yeah. So in the Spot world, we we started using AI years ago on the behavior side. You know, so specifically around making the robot better at mobility and locomotion. Hey, Spot. If you see that wall, don't run into it. You know, that was really the beginning of getting AI into robotics. But to your point, we're in a totally different world now. And what that means is we've got the hardware now that enables general purpose robotics, and the software has come up to match. And what you really need, the recipe for a general purpose robot, in in my opinion, recipe for a general purpose robot is a robot that can go anywhere, not just anywhere a person can go, but places that are dangerous Inaccessible to humans. A robot that can understand its surroundings. So understand the context of its world. Look at the world and say, okay, I understand how to navigate safely in this space that's built for humans. And then finally, the third piece, of course, is interact with the world. Manipulate anything. Be able to do an end to end task simply, easily, straightforward. And that's really when people think about robots participating in the world, being a a moving from an industrial era of robotics to a service era of robotics. You need that whole story. And that for me is where AI is really coming in. Go anywhere, we've pretty much nailed that. The robot bodies are doing a great job with that. That understand surroundings and manipulate anything piece, that's where AI is becoming extremely powerful. Our partnership with Google DeepMind is gonna really help on both on that manipulation side, and probably on the understanding surroundings side too. So we're we're really building this incredibly rich software based interface that allows us to put more robots into the world. I would say the same is definitely true in the warehousing logistics space with Stretch, when it comes to that world understanding and that situational awareness that we need to build such that we can make very confident manipulation or handling decisions, you know. And when we think about what it takes for a person to do something that we might consider pretty trivial in terms of the cognitive demand unloading boxes from a trailer. I've developed such a a greater appreciation for not just the human form, but also the human mind when it comes to all of the decision making that happens just subconsciously when humans and people are deciding what is the right box to pick next. What is the best way for me to grasp that box or multiple boxes if that's the more efficient thing to do. And it's really hard to embody that logic in a robot. And so we're taking images of these pic scenes of these worlds to inform our situational awareness, but there sometimes are, hidden understandings that we can't see from those initial images that we take, and only once we go and attempt to grasp something and try to manipulate it do we learn, maybe that wasn't actually the right way to approach it. Maybe I should try a different approach next time for the same sort of scenario. So using these machine learning techniques and AI data driven approaches, we're taking not just those, that perception data, that visual data, but also that situational data that we observe as the robots are actually trying to and also hopefully successfully performing that work in these increasingly more complex environments. Right? You might think about the first mile whereas having those neatly organized shipment containers of boxes and the more times that those materials are sorted throughout middle mile facilities, the higher diversity of objects need to be handled, and the higher the disarray. And we want to enable the same predictable performance in those highly disorganized environments as we're already seeing in those neatly stacked organized environments that we see today. And that's pretty similar for Spot too. You know, it's walking around these facilities as diverse as a steel foundry, and a pet food manufacturing facility, and an automotive plant, and a brewery. These all are extremely different. There are some things that are conserved. A palette is a palette is a palette. But there are gonna be so many things that it's gonna encounter as it moves through these facilities that are different, and that build on the semantic context of the world the robot is in. So the more data that Spot and Stretch collect about their world, the more disarray and entropy gets thrown at them, the better they get when you start to introduce those AI tools. But most importantly, the more useful data goes to the people who need it. Absolutely. And you also talked about how, you know, we've learned through the Spot deployments and scaling for the last several years about how to build good user experiences. Right? Not just in the way that they see and interpret data, but how they're physically interacting with the robots as well. And you know, Spot has been out in the world the longest. Stretch has been on the market for a few years now. And our incredible UX team applied learnings from that user experience and how easy it is for customers and operators to drive Spot to applying the same sort of experience in Stretch. And when our customers visit us, we encourage them to drive Spot or to see how intuitive and easy it is to not just drive, stretch, and operate it in our labs, in our test environments, but also see that apply to how we our deployment teams train their operators within a matter of hours Yeah. To confidently handle these robots in their production facilities. Yeah. I think that's the amazing thing about having these different robots out in the world that have really different form functions. Right? Yeah. Spot and stretch look incredibly different. But, the ways in which we develop them, getting high quality data, bringing it back so the robots can keep learning more and more over time, and sharing that intelligence not just for the human operators, but also as a research team. Right? So that we can bring lessons of how did it work to put Spot out in the world back to how it works to bring Stretch out to more and more people too. We're really sharing that across the teams, and not just for Spot and Stretch who are there today, but also for Atlas, thinking about what it's gonna take to deploy a robot that is a humanoid that looks even more different than Spot and Stretch today. Yeah. One of the other things we've learned from Spot is that Spot is a very powerful platform that is not only very capable in the applications that it supports today, but we have partners, integrators who are you know, building custom payloads specialized applications. We're thinking about how we might approach that in stretch, right? Whether it's swappable grippers and end effectors to perform other types of maybe more specialized industrial applications where different types of dexterity are needed, but you know, could Stretch also become a platform one day for integrators. Once Atlas enters the world, it will have a set of applications that you know, we've specifically designed to perform well and Atlas will continue to learn and become proficient, if not master additional applications. But we also want to democratize the ability for customers to build their own applications on top of their robots. And I think that applies across our entire portfolio. I think it's fair to say that Stretch, Spot, and Atlas are all general purpose robots. That is really what they're designed to be. And at Boston Dynamics, we now have this extremely rich history. We've learned a lot about, you know, what it takes to get more general purpose robots out into the world. And to do that, you know, we have this sort of proven track record of reusing the most cutting edge technologies of the day. We don't need to, you know, build everything from first principles. We want to use technology across the robots and the platforms. So when we talk about spot and stretch, learning from each other, using the same robot brain in orbit, this is all setting the foundation for customers who are ultimately going to deploy Atlas. Because Atlas is going to be using that same Orbit based robot brain. Yeah. Absolutely. We were just talking about how customers can think about preparing for Atlas and humanoids. And we know that's a big part of what customers are thinking about. And it's it's really very public. Right? All the exciting steps that we're taking towards getting towards humanoids and these general purpose robots with really intelligent brains. Something that comes up over and over again with our customers and that has been thrown out in the industry for a while now is this buzzword of digital twin. And it is a loaded term because it has such promise to it. Right? What the possibilities of what it could be. But for us here at Boston Dynamics, just simply, we think about this as a virtual representation of the world. A virtual representation of the world that can help people do things like simulate people, processes, and also products. Right? How they actually bring products to life in an environment. And what's exciting about that is robots can be key enablers to building these virtual worlds. And those are the kinds of applications that are going to make it possible to deploy not just Atlas, but many different robots at scale. And think about how different types of automation really play together in a broader ecosystem. And that's really exciting and something that we're thinking about for how Spot and Stretch play into those worlds too. Yeah. You know, from my perspective, we've seen this movie before with Spot. There's there's a great big hype cycle about, wow, we're gonna have these incredible robots that are gonna be able to do anything, any job, any task. We're gonna tell them what to do and they'll do them. We've all seen the videos, these incredible AI enabled videos of robots doing amazing things in industry and in the house. But the reality is, the only way to get to that future and that outcome is to invest now in that robot infrastructure, that robot brain that gives the robots that context, that understands surroundings. So, when I think of, to your point, that loaded term of digital twin, it's actually what is the robot experiencing and seeing, and how does that actually bring value to humans? Yep. So in the case of industry, and I think and this is where we're starting for both Atlas and Stretch and Spot. In the case of industry, that means like a software defined factory. You've got a lot of intelligent devices within your industrial facility. All of those can can and should be speaking to each other. That includes the robots in your facility. And on the Spot side, you know, what we can do is actually build that into a location based virtual representation. So as Spot's moving through a facility, it's collecting these three hundred and sixty degree panos as it goes. And it can collect them with our it's visual cameras, it's thermal cameras, but it's building this incredible view of the facility in in two dimensions. You can even build it in three dimensions with a laser scanner. So it's as it's building this representation of the world, when we add that semantic layer of AI based understanding saying, hey, I recognize that this is a vacuum pump. I recognize that this is a fire extinguisher, and I'm a robot, so I know the location of it too. That then tying into that virtual world gives you a map and a system that can build a software defined factory. Now, that's a very small piece of what a true virtual world is. You know, we would need to add to that dramatically to make a robot that can function in every part of the world. But heck, we can make a robot function in a factory, in a software defined factory, in a virtual world and complete the loop between virtual to physical to virtual to physical. Yep. And what I love about where we're starting is that software defined world can be useful for the robot to navigate, to do tasks, to manipulate the world. That's what's going to get us farther and farther along this path towards AI. But, it also is immediately applicable and useful to humans. Yes. And, that's what's really important about making a robot technology sticky and useful. Right? Yep. It has to do things that both work for the robot, but also work for humans. Having that kind of come complete circle out of the box makes it a lot less daunting of a task for customers who are thinking about how do I even get started in the software defined world. Right? We have people across really different stages of this transformation journey. Some who have full blown virtual worlds already built out that have robots adding to it and have teams and data scientists working on it day in and day out to optimize how they run their processes and build products. And we have other people who are equally as successful who are just starting and don't have anything yet. And so, we have to make sure that when you're thinking about adopting new technology, we have both of these customers in mind when we go and we deploy our robots. We wanna be able to deliver an end to end solution out of the box that can help you get started, but also tie into this larger vision. One of the things I'm really excited about as these data informed virtual worlds become more and more ubiquitous ubiquitous is the way that it's going to change how people interact with their facilities Yeah. And their workforces. Right? You used to have to be physically located where your machinery was, where your material handling equipment was to understand how it was performing, what impact it's having on your operation, where the bottlenecks are. You could be sitting at home or in you know an office nowhere near that facility and now have that deeper contextual understanding of not just metrics on dashboards, but performance in the context of the physical layout and design of your buildings. We're also not so naive to think that Boston Dynamics robots will be the only types of robots operating in these customer ecosystems. Right? And so, just as important as it is to receive that telemetry data from perimeter logic controllers, from IOT devices. It will be increasingly more important and essential to have that data exchange and communication with other types of robots so that if a customer is using that digital world as a single pane of glass into their operation, they don't have one pane of glass for one vendor and one for another. They're seeing this aggregated system and an orchestra of different types of robot agents and vehicles and AMRs and forklifts working all together in service of maximizing operational throughput And safety and efficiency for these operational contexts. Yeah. I think that's great. And, I think for us, we think about this all the time because our robots are built as a platform. Right? We have open APIs for our robots and for Orbit so that you can connect these systems together. And we want them to be. We want our data to be going out into customer systems and to be coming back in to make the robots more intelligent. Now, we think about this and we've built our products in a way that make it easy for customers to do this. But that doesn't mean that they're always ready to start to integrate these different systems. Right? I think one very very practical thing that anyone can do to get ready for automation is to have a clear data strategy. Yes. And it feels like we've been talking about this for a long time. I'm gonna say another buzzword. I'm like the buzzword person today. But, industry four point o or industry five point o, you pick pick an industry. Digital transformation? Yeah. Exactly. But, it all ends up coming back to data. It's been years and years we've been talking about data, but you can go to any conference and the majority of what people will talk about is, I have this amazing AI. To get this amazing AI, I had to clean up all my data. I had to make sure my systems can operate together. And it's not uncommon that we'll show up to a customer site and they'll say, we wanna get this data into our other systems. We wanna make sure that if Spot finds an anomaly or Stretch is unloading boxes, we want that operational metrics going towards our systems. But we're not ready yet. Like, we haven't agreed on our data strategy as an organization. That's only gonna become more and more important as we see more automation start to come up with this AI. So I think that before anything else, customers really need to think about how this data is gonna play together, not just for robotics, but for all the systems across. And, those data flows also extend to your IT and networking Right? We have a mission to deliver automation without infrastructure or as little infrastructure lift as possible for customers to realize the benefits of our automation solutions. And you know, what we generally mean by that is that you don't have to change your site layouts and your physical infrastructure because these robots can go anywhere. But you might think about based on all of the data that needs to flow between these robots and other external systems or, devices, what bandwidth is needed to accommodate that and make that happen? Is your networking infrastructure set up to not only support that, but also other automation you might integrate into your facility? So we're having those types of conversations with our customers now, so that they can be as prepared as possible to as seamlessly as possible, integrate these solutions within their existing site networks. It's really all about change management. Know, there is this piece of, you gotta prepare for a robot future. You gotta prepare to feed the AI engines the data that they need in the future to actually take advantage of these tools. What do you need to prepare for on-site to be ready for a robot like Atlas to come on-site? And there's the data strategy piece, extremely important. Data strategy and management. There's the networking piece, extremely important. There's the piece of preparing your workers, your teammates for a future where they have robot tools that they can work with. You know, what does that look like? On the Spot side, you know, there's a few small things we do. Because it is a pretty surprising thing to suddenly see walk into your workplace. If you've never seen a walking robot before and Spot suddenly shows up, that can be a surprise. So, you know, we have this whole program that we'll run through with our customers to prepare their people to work with a robot tool. And that includes things as silly as bringing the robot into the cafeteria every day, taking selfies, naming contests for the robot. But it also means educating the teammates on what the robot is and is not, And how the robot can help them do their job. So I think this is also a piece that is really important when you're preparing a site or an environment for robots to come in and be useful tools. So there's there's a lot of pieces you can do to lay the groundwork and enable really successful deployments of humanoids in the future. But it's easy to do it today, and it actually has an enormous amount of value today for your human teammates, for the people who are on-site, and for your data collection as an organization. And that very much applies to the safety aspects of change management too. You think about Spot entering the scene or crossing your path while you're going about your work. Well, now think about a three thousand pound behemoth like stretch doing that same thing. And Stretch is, you know, when you talk about what our robots are and what they're not, Stretch is an industrial mobile robot. We have to not take that lightly. It's not a force limiting cobot arm That is able to interact as closely or intimately with people as some other cobots on the market are. And we need to make sure that these people understand that deeply and understand not just what the robot's doing for them and their organization, but how to safely, work alongside that robot as another coworker, at their site. So we, you know, take the highest standards of safety as we're designing how these systems are actually going to integrate into these very, crowded sometimes environments with so many moving parts and and people to make sure that, yes, we're we're we're gaining the efficiencies that we think we can with these robotics, but we're also doing that as safely as possible and that people understand that safety comes before anything else. Yep. And, the piece that you both talked about, I mean, there's of course the experience of what does it mean to bring a robot in and how can we help our customers give our their employees the best experience possible. But then, once it is up and working and people feel comfortable, the real success happens when it becomes part of the everyday workflow. Yes. It's a coworker. I love going on-site to people working with Spot and they say, oh yeah, I rely on Spot to do this task every day. And I just know it does it. And I don't even see it move and get up and down every anymore. Right? It's just a part of my every day to see Spot walking down the hallway. Or same with Stretch. Right? You start to rely on, hey, this exact kind of operational efficiency that this robot is gaining means that we can do other things with our workflows. And that is a really important part about bringing these robots in. Right? It's not just replacing everything, but hey, how are we working together to think more broadly about our operations and how we now work with this new type of automation? I love that because you know, whereas there are these you know, stigmas of robots coming for people's jobs, the real conversation that we're having with customers are about how their associates are excited to work with robots. Right? And and it's turning out to be more of an attractor than something that is scaring people away because people are excited to up become upskilled. And, people tell me all the time that they don't want to perform the grueling task of lifting fifty pound boxes repeatedly throughout a ten hour shift. Right? If the robots can handle those remedial tasks, people can think about, you know, how best to optimize performance in that robotic system. We very much need those people to be successful, with our robotics solutions. And so when customers are thinking about their change management campaigns within their own organizations, I think that is a key point to highlight is that this presents many more opportunities than it certainly does take any away. Absolutely. And and I think all of this kinda comes back to the point of a software defined factory. Because what you're getting into with all of this is you're trying to optimize work. And that also means what tasks are robots really good at doing. Because robots are great at tasks, not at whole jobs. And then where do the human teammates optimize their time? The time that is best done on jobs that humans are best at. Robots are never going to be as good as humans at certain things. So being able to, again, optimize the facility through automation, through a software defined factory, you are getting that much closer to the future of work, what that looks like, and and a workforce that is empowered, enabled, and happy at their jobs. Absolutely. I've heard of more than one circumstance where Spot was taken off the job to go to a career fair. And where people and where's Spot? Yes. And someone said, oh, he took it to a recruiting fair. Yes. Which is great. Yeah. We love to hear that. That's right. We I think that's an incredible story. In a lot of my conversations with customers, you know, people saw the CES reveal of Atlas. The world saw, I guess, the CES reveal of Atlas. And it's such a remarkable robot. We're everyone at Boston Dynamics, we're so excited about it. And the world's really excited about it too. And that means that, you know, sometimes I'll hear from customers things like, well, why don't I just wait until Atlas is available sort of to get started with with robots. And I think that's a really important topic, you know, for us to just get out there and and talk about. It's a really good question to to contemplate. Know, there's a presumption that maybe one day Atlas will be able to do any of the tasks or sometimes jobs that robots like Spot and Stretch are doing today. And that may be the case. And so it's a really good decision point as an organization is considering what is the best type of automation for the problem that they are solving. And there are going to be some tasks that you could choose between stretch or Atlas or spot and Atlas. Especially if you plan to flex robots between processes just in the same way that, people might start one task in the morning and move over to some other task in the afternoon, that could definitely be a consideration. That is something that is a capability that is unlocked with this next generation of general purpose robots. But I really firmly believe that there will always be certain tasks where robots like Spot and Stretch which are still general purpose or multipurpose are the better robots for the job. Know Atlas could grab a stool over so that it can reach higher to the top of a shipping container. Right? Technically that's probably possible. Atlas could crouch down on the floor maybe to get a reading on a gauge or a motor underneath some conveyance equipment. But there's a lot of you know what we might consider wasted movement, or work that it takes to make that happen. Stretch is was specifically designed for efficient case handling. And so for those types of processes, especially where customers have, really high performance targets for throughput, efficiency may win out over the general generality of that robot to be, redeployed to other jobs and tasks in the building. In addition to some of those form factors playing a role in that is also, like we've been talking about, the data that it takes for general purpose robots to not only become proficient in a certain task, but also master that task. It I think of it a lot of ways in the same way as I think about when I go to the doctor, you know, I'll see my primary care general practitioner who has really high reasoning and adaptability to kind of examine what's going on with you and maybe give you a diagnosis. But what it would take for that same doctor to perform open heart surgery or, interpret an x-ray, would presumably take a lot more training and more time than a surgeon or a radiologist who performs that same job every single day and has a much better defined set of tasks to perform. So, I think between the the data size and and time to train these general purpose robots in addition to their physical forms, that will be a really important set of considerations as customers are deciding which is the right robot or system for the job. And frankly, the spot side, I'm actually not interested really in playing in the same pool as humanoids. You know, what I want to lean into are the places where that form factor and that type of data collection is advantageous for the task. To exactly your point, there are gonna be different form factors of robots that are better at different things. And science fiction has certainly showed that over the years. That's probably the reality of the future. A general purpose robot is still gonna have some degree of customization that is needed to do those those really niche tasks. So so, yeah. I I agree with you a hundred percent on first of the the sort of the first factor of saying, hey, why don't I wait for Atlas? First factor is, well, you don't actually have to. You know, there are robots today that do the tasks that you care about and give you value today. But I think there's a much an even bigger picture too, which is that path to Atlas. And I think, you know, Amanda, you're probably best suited to talk about all of the infrastructure that we wanna kinda get in place ahead of time in order to give you the best chance at value with Atlas. Yeah. And I When I talk to customers about this, it comes back to this system integration and software defined world and how you get all of these things communicating and working together. AI is incredible. Robots can understand the context of their world more and more, and that's wonderful. But, what I say to people is, look, if Atlas or Spot or Stretch is performing a task in front of them, they can understand and interpret the world. But if all of a sudden you have a material shipment that's happening in a different area that you need Spot to inspect before Stretch can move it and then Atlas can unload, well, they're not magic enough to know what's happening somewhere else. Right? Those systems have to be integrated with Orbit and with your other AMRs or all the other infrastructure that's happening in your facility so that information can be passing back and forth so that robots can do the jobs they need to do. And you don't have to wait for Atlas to get started on those integrations. You can do that right now. That's how we work with Spot and Stretch customers today to make sure they're ready for that fully integrated software defined factory. So build your software defined factory, and then understand what are the best workers to fit into the different pieces of that factory. And Spot is a worker, Stretch is a worker, Atlas will be. But having that infrastructure, having that data strategy, having that networking infrastructure, having the change management piece all sorted out ahead of Atlas coming to your facility, that's what's going to give you the best shot of success with a humanoid, of any humanoid. And by the way, by the time Atlas is eventually able to complete these jobs or perform these tasks as well if not better than the robots that are doing them today or the the the manual processes that are in place today, you probably will have already gotten your return on investing in Spot or Stretch. I know ROI is a huge driver for Yes. Making certain purchase decisions around integrating automation into the real world. So when we think about the the risk reward of of waiting for something like Atlas and and general humanoids and that promise to become true in the real world, the risk is fairly low. And if you see a win to be had in a strong business case with Spot or Stretch, and by the way, our teams are really good at helping customers confirm and validate those business cases. You know, go for it. Get that win. And you can also realize the benefits of Atlas once it arrives and is integrated into those same data platforms and operating systems as these robots which you can buy today already are. You're really preparing for an automation ecosystem. You know, what we're really talking about, what's got all the the product teams at Boston Dynamics really excited, is about an ultra connected future. Absolutely. It's about multi robot environments where robots share intelligence with both their human teammates and each other. It's actually physical AI operating in the real world. And I think that the nature of work and the nature of how we interact with the world is going to really change in the next few years. It's very exciting. Is. Good times. Incredibly exciting. So we've really enjoyed chatting today, but I also know there's been a lot of questions. So we are gonna take a quick one minute break, and then we will be back to answer your most burning questions. Welcome back everyone. We are ready to answer some of your questions. One of the most popular questions that people are asking is what do most customers work towards or how do they see an ROI for Spot and Stretch today? Yeah. And that's a great question. And it's kind of the what we lead with when we start talking about Spot for industrial use cases. For the tactical response customers, you know, those ones who are doing the explosives detection and first responders, the ROI is is pretty obvious. It's keep a person away from harm. Don't have to have a person put on the bomb suit. And for industrial, that piece exists too. Keep a person out of harm's way. You know, send the robot in to do the really dangerous checking of of equipment machinery. But when we talk about ROI for Spot, it is hard number savings. So you're looking at repair versus replace. So as Spot is going through your facility, it's gathering intel, it's saying, hey, this piece of equipment looks like it's gonna have a problem. So repairing that versus replacing it, typically a big cost savings. Unplanned downtime. You know, some of our customers, semiconductors, downtime costs them enormous amounts of money. So that's a really big bottom line piece. And then you also have pure energy savings, actually. So a lot of our industrial customers, they've got sustainability targets. So, hey, by twenty thirty, we have to reduce our energy consumption by twenty thirty percent. One of the best ways to do that is to patch up air and steam leaks at a lot of these facilities. And that's something that Spot is great at detecting are air leaks, steam leaks, gas leaks. So when you're talking about those bottom line savings, you're looking at an ROI for our industrial spot kit of around eighteen months. Definitely under two years, but around eighteen months for most of our use cases. Yeah. For for stretch, you know, we're for our commercial trailer unloading application around that three year industry benchmark for ROI in the warehousing logistics industry. Of course, that depends on, you know, customers ensuring that the robot's as utilized as it possibly can be. And when we have conversations about that utilization, how can we keep stretch as busy as possible? It sometimes leads to the concept of even how customers structure their work differently. The concept of a shift starts to change because Where you have people scheduled to work in eight or or ten hour shifts, a robot isn't gonna get tired, robot isn't going to need to go get lunch or coffee. And so, you could have stretch working around the clock so long as you have the volume to feed the robot, and that's really where customers are seeing that ROI come in is being able to use the same fleet of robotic resources to work for throughout an entire days where you would have to have two or more crews of people come in to do the very same job. I think ROI models tend to really focus on the productivity aspect of the work and kind of the the direct efficiencies with how fast these jobs are performed, how much throughput is achieved when it comes to material handling processes. But there are also really important conversations when it comes to some of the less quantifiable aspects of return. Certainly the safety aspect plays a big The reduction in and avoidance of injuries and musculoskeletal issues for people who are performing these ergonomically stressful jobs before is absolutely a benefit that should be captured. In some cases, customers have employed entire teams in dealing with and correcting some of the accuracy issues that are caused by human error. Right? Which we hope to not see so much of in a robotic process. Robots are not perfect, we make mistakes too sometimes, but hopefully less often, right? So you're talking about the performance aspect, the safety aspect, accuracy, and and there are probably other several indirect areas that your organization stands to to benefit in by making that switch over. And for me, the biggest indirect, outside of safety of course, which is true for for SPOT too, but for me, the biggest indirect value that we see is the data. Yep. You know, that is not something that's easily quantifiable even today. But we know that you're gonna need an enormous amount of data in order to do anything with optimization in the future. So, the collection of data from both Stretch and Spot is a contributor to the value of the robot even if we don't include it in an ROI calculation. That's so true. You know, I think customers are finding new ways of optimizing their businesses that they didn't originally think of when they were deploying these robots. With not only discrete data, but also image data and perception information, We are augmenting their databases and giving them visibility to things that they didn't have before. For example, in in logistics and warehousing, sometimes there are disputes about things like damages that happen to products as they move throughout facilities. And when you don't have that ground truth data, that proof to empower you in handling those disputes, there is some big money that's lost in the process. And so customers have access to the images that our robots are taking as they're handling these sorts of materials and valuable goods in a lot of cases. And they can come to those types of disputes or situations empowered with that data that is a lot less arguable than just an assumption. An auditable data record we call it. Yeah. Yep. And high too. And very high quality. Repeatable. Yes. That's the other thing that I think sets this data apart from just the general gathering more and more data in your facility. A robot is going to take the same picture in the same place with the same high quality cameras every time. And that makes it a really strong case. Yeah. And as more robots and more automated systems are integrated within these same facilities, you start to build that end to end picture Of the life cycle of your goods that are moved throughout your buildings, and and that helps you tell the story of what's really going on. Yes. So you've talked a little bit about hard ROI for spot and stretch, But what about people who are asking about, how do I just get started? How do I start making a business case for why I should bring in robots today and get started? And, you know, we talked about this previously a little bit. But certainly, making sure that you have a data policy, making sure that you have networking that can support fleets of robots, and then also, you know, working directly with your robot provider's teams to understand what you should expect of the robot. What should your ROI be? What does a site acceptance test look like? How are you going to support this fleet through its life cycle? Do you want to be involved with the maintenance of your fleet? Or do you want to have a service plan to make sure that your fleet is always, you know, working well? You know, the benefit of working with a company like Boston Dynamics, biased as I am, is that you have those that support and those professional services end to end that can help you with every piece of that. They can help you put together something about data, a policy about data. They can help you put together what the networking needs to be. But then they can also help you install those robots on-site, get to an acceptance test, hit those milestones that the people who are running the company need to see in order to make that program level decisions. So that that's my take on getting a site ready is understand what you need, work directly with your robot provider, and and then then start to to chip away at the things that are gonna get you that foundation. And you know what what it really starts with is talking to us. In order for us to work with our customers to find the best solutions for their business, we need to intimately understand their business. We don't take a cookie cutter approach and assume that one type of implementation or deployment is gonna work exactly for one customer or even one site at one customer the same as it would another. And so, we want to understand if it's if it's logistics, what type of materials are you moving? What do they look like? What kind of variation exists? You know, we oftentimes, will take a look at, customers item master data and run that through kind of eligibility analysis to project, hey, we think the robot would do really well with this type of stuff. Let's maybe do some testing or experimentation just to make sure we're confident about this set of product as well. And you know, these we're not so sure about, so maybe there is a collaborative workflow that makes sense when it comes to handling these types of goods, or that's actually a feature that's on our roadmap right now. So it may not be available now, but in a year it will be. And as we're talking about a deployment and scaling plan, let's align those rollout plans accordingly with when those features and and capabilities will be realized. So it it starts by reaching out to our sales teams and our product teams. We want to come to your sites. We want to see what it's like to work in those spaces, not just how these particular processes or jobs are performed, but even adjacent processes that are upstream or downstream for them. What are the integration points and the handoffs between them? What's something that given our, you know, robotics expertise and technical expertise, do we maybe see that hasn't come up in conversations yet, but we think would actually be pretty challenging For our robots to operate successfully in these environments. So I think it really starts with that developing that deep understanding of our customers, their motivations for automation. Is it is it safety that you're looking to improve? Is it performance? Is it both? Is it a general sense of predictability about your business? And based on that, and an understanding of your actual production environments, what do we think is the right path forward? Yeah. What does success look like to you? Yes. Yeah. I saw I saw a good one come through. What are you most excited about next in robotics? I mean, I think for me, what's most exciting about robotics in the coming years is what we've been talking about. Right? I love robots, but I love the software piece. I'm obviously biased thinking through Orbit. But that promise of interoperability and robots working together and not just thinking about one robot, one task, but how do you think about your broader business decisions and logic and outcomes you're trying to get to. Right? I I build widget x and I need to output it in y time. And in order to do that, I have to make sure that these processes work in this way and these people work well here and automation works well here. I love that we're getting closer and closer to a world where you can see all of that in one digital place and really optimize because that's the exciting thing about how organizations are going to become more and more competitive. Is understanding how they can really optimize. And, maybe that's not quite next year, but we're we're really getting closer and closer to that world today. And, I'm just so excited to see it come to life. I would very much echo that. And, you know, one of the other things I'm really excited about as these advanced robotics become even more ubiquitous and having a greater presence in our day to day working environments is all of the new ideas that will just emerge for how to either make better use of existing deployed robotic resources where maybe the the you initially deployed Spot or Stretch for an initial purpose, but you found a bunch of other use cases just by observing how the robots work and add value to to your businesses. And some of the the best and and crowd favorite features and capabilities have come through direct feedback from our customers. Once they've started, testing and integrating these robots within their ecosystem. So, you know, I think we're gonna see sort of this exponential growth in just the quantity and, unique types of ideas that will come back to our research and development teams and product teams about, you know, how can we really maximize the potential of these new coworkers. Building on what you've both said already. For me, it's the humanity piece actually that I'm most excited about in robotics. Because these tools, which is what they are, they are tools, are going to remove barriers for humanity. You know, our owners and collaborators, Hyundai, they've they've used this term partnering human progress. And I I actually it really resonates with me because the truth of the matter is the closer you get to general purpose robots, the more that we can remove barriers for people. And whether they're barriers in how people work, how people interact with the world, you know, these are tools that are ultimately designed to benefit humanity. That's what I'm most excited about. Think the future is very bright. Really good general purpose robot should be able to make work and life easier and better, and let people focus on the things that they are passionate about, on the skills that they have on on on expanding their horizons, expanding their skill sets into things they didn't necessarily have time for before. So I I think, I really do believe in this partnering human progress toward this brighter future for humanity. Now, our time has already come to an end, most unfortunately. This is all we have time for today. But I will just quickly hit one point that we keep seeing in the chat, which is about procuring robots. Know, how how can you buy Spot, Stretch, Atlas? You know, both Spot and Stretch, go to the Boston Dynamics website. There are portals where you can contact our sales team and and get through to us too directly. That's right. And then on the Atlas side, there are we're still quite early, of course, in the development. We're not fully commercialized yet. If you have interest in Atlas, there is a form you can fill out on our website, which is gonna get you a follow-up from the teams that are that are building out that program. So so stay tuned everyone. We've got a lot of exciting things coming from Boston Dynamics in the next twelve months. And we really appreciate you being here today.