Ian James: Where Robotics Meet Data and Analytics

Speaker A: Hello and welcome to the Business of Data podcast, brought to you by Corinium Global Intelligence. In this podcast we talk to senior executives, thought leaders and experts from a range of industries and departments within large and small organizations across the globe. They share stories and experiences that shape their passion for data and analytics inform the future of our industry.

Speaker B: Welcome to the Business of Data podcast. In today's episode we're diving into robotics at the intersection of data analytics and AI. Precision robotics have been used for many years in industries like manufacturing, for example in automobile construction. But as the technology progresses, robotics are increasingly used in other industries like agriculture, healthcare, e commerce, and even in customer service. But how do robotic systems interact with and influence data and analytics systems and processes? And what are some of the associated challenges and opportunities? Well, our guest today is ideally placed to answer that question. Ian James is the head of engineering at Right Hand Robotics and has a rich background in robotics, data and AI engineering. So welcome to the podcast, Ian, and thank you for joining me today.

Speaker A: Hey Gareth, how you doing? Thanks so much for having me.

Speaker B: It's a pleasure. And I thought we could perhaps get started by uh, you telling us a little bit about Right Hand Robotics, where you guys sit in the market and what your focused on at the moment.

Speaker A: Yeah, absolutely. Excited to. So Right Hand Robotics is a uh, autonomous robotic peace picking solution. Um, and by that I mean uh, in warehousing and fulfillment or supply chain and logistics. I think the best analogy is when you order something online at some E commerce platform, the fulfillment of that order of two or three or however many items, there's like 30 steps in a chain from when you click purchase to it actually arriving at your door. And of those 30 steps, uh, fulfillment and supply chain logistics. It's a complicated process and there's a lot of human involvement and a lot of manual labor. Um, and as things start to scale across the world, uh, in that industry there's just not enough labor out there and there's too much goods moving to the system at too high a velocity for there not to be a huge sort uh, of automated component. Um, so we're an automation solution. There's a lot out there. In other parts of uh, supply chain we particularly focus on, as I said, robotic piece picking, uh, manipulation with objects, an intelligent system to be able to uh, basically do order fulfillment. Right. So the actual quarter sort of packing of your order, uh, um, in conjunction with the systems of the warehouse that have inventory, the warehouse, uh, management system, for example, um, we are kind of doing the boring, repetitive jobs, uh, uh, that there's not enough people out there to do if anyone's seen Amazon hiring at like 30 to $40 an hour for people to do uh, some of the most straightforward things for example and still not being able to find enough folks to manage their distribution centers. Um, but just those very basic things where to humans, you know, picking up and manipulating an object and being able to pack it uh, efficiently inside a box. For example we have a system that through ah, a computer vision and uh, compliant robotic gripper manipulation and um, sort of a lot of other intelligence around motion planning and object detection we're able to uh, basically recreate human object manipulation but do so in a really kind of fast and effective and efficient way and allow uh, for one thing we're really excited about, which is called uh, lights out picking or lights out distribution which is now the ability for you to click purchase on your order at 12pm and with no one in the warehouse. Uh, um, our solution is part of a sort of a chain of automation that allows that order to be fulfilled and ready to go on a truck at 3:00am without uh, any human involvement. So it's a really, really powerful platform. Uh, it's a really kind of exciting market that we're in. And um, yeah it's a, it's a cool piece of tech.

Speaker B: Yeah, absolutely. And you know I think all of us rely so much on you know, mail order for everything these days that you know this kind of automation is only more important. I actually had the opportunity several years ago to go into the warehouse of a major home retailer in the United Kingdom. And it's fascinating to watch these systems at work. They had a fully automated background, uh, sort of fulfillment um, area. I'm ah, sure a lot has developed since then. So um, you know, what are the um, what's currently happening in the world of robotics? What are some of the most exciting developments right now?

Speaker A: Yeah, yeah, it's a great question. Um, it's interesting when I, when I kind of came out of college and sort of got into the professional world, um, if you looked at robotics then I think the innovation was really pushing more and historically this has been true on the hardware side of things. How do you imitate lifelike motion either of humans or other organisms? Um, how do you make uh, uh, uh, robots more collaborative, uh, have them safely in the space around humans. We ah, traditionally think of um, robots particularly in automation sense as those arms that put doors on cars. Those things are not intelligent if you get in their way. I'm not sure you're going to see the light of day. These are very dangerous sort of um, applications in the sense they need to be gated off and guarded. So collaborative robots or cobots is really coming um, into industry terminology and kind of gaining a ton of traction right when I entered the professional world kind of in the early 2010s, um, and I think the big shift I've seen, there've been a lot of other pushes in the interim between then and now, but the biggest shift I've seen in the last couple of years I think is uh, really pushing the boundaries um, on the software end of things. Uh, so less focus. And this is I guess more from a commercial sense, but probably true from an academic sense as well. Because when you think about robotic innovation, there's what's happening in R and D and what's happening in companies that are trying to actually sell their wares and be commercially successful and how can they do that sustainably? And I really think on both ends of the spectrum it really has been to kind of say hey listen, we've developed a lot of really powerful robotic solutions, right? Whether it's uh, robotic arms or bipedal robots or AGVs or AMRs, autonomous mobile robots, um, let's not stop innovation there but let's put our focus more to how do we use these things, these hardware components and almost treat them as commoditized, uh, uh, and focus on manipulating them in more intelligent ways through software and really focusing the R and D and sort of the innovation around capabilities in the software space and how we can actually manipulate uh, uh, uh, this hardware, uh, as opposed to how can we make our arm faster or more precise, how can we make these AGVs, uh, sort of have tighter turns in the warehouse for example, right? There's always going to be companies and sort of uh, uh, researchers uh, that are focused on more of these hardware oriented problems. But I think largely that the shift I've seen is if you see where existing companies are pushing and where new robotic startups that are getting funded, what their focus is you really see, especially with the huge kind of explosion of AI coming into the forefront, um, you see it really being uh, uh, software driven innovation in the space as opposed to Boston Dynamics being the opposite. Example, when's the next Atlas video coming out? How fast is it running? Uh, as opposed to that being top of mind, it's hey, forget the robot in the image. You'll see a lot of great applications of gen AI are interesting ones where you're just telling it to find and pick up an object and it's able to figure that out. Uh, in that video, do you care a lot about the robotic arm or the end effector? Do you seem to pick it up? Not as much. But you, what you're wowed by or impressed by is how it's able to interpret that command, act upon it intelligently.

Speaker B: Yeah, I think that's fascinating. You know, in your role, as you're looking at the kind of, the development of the software side, you know, how do you, how do you approach that and what, what kind of hurdles are you trying to overcome or you know, what innovations are you trying to focus on as you're developing that hard, uh, that software?

Speaker A: Yeah, that's a great question. So I always tell our team it's, we're still making hardware innovations because we're not a software product, right? Like Right Hand's main product is called rightpick and that's a name for the fully integrated kind of piece picking solution, right? The vision system, uh, the actual sort of uh, um, patented unique gripper that we make which kind of imitates the compliant force of a hand, like your hand wrapping around a coffee mug, what have you. When we look at RightPick, it's ultimately a solution that interacts with the real world in a physical sense. So it can't ever be just pure software. Even if we're not developing a new gripper or making changes there, or altering the cameras we use for our vision system, which is not something that happens every month, for example, uh, we're still ultimately coding for a hardware based real world product. And so I think what our, what our attitude is is we're exploiting hardware innovation through software. So for example, you know, uh, in our system there's so many sensors, right? Like a camera is obviously a type of sensor. It's taking in visual data. There's so many other sensors, um, that kind of are components of the system, um, that are able to, you know, detect the presence of an item. Uh, they're able to detect certain aspects of the surrounding environment that inform and allow the system to operate successfully or operate more successfully. And for us I think the push is hey, we don't exploit these all as much as we could. How could we use our existing hardware more powerfully in a more innovative way? And we're unlocking that not by going into SolidWorks and designing a new component, right? We're exploiting that through software from high level down to the actual firmware running on the boards, uh, of these devices to be able to interpret that Data and sort of inform the movement of the bot or sort of the force by which we grip or grasp or interact with the real world, um, in a more kind of intelligent way. Uh, so that's really the mantra, I think, is hardware innovation, uh, exploiting hardware innovation via software. And then of course, going back to the focus of this podcast and a, ah, big reason I'm excited to talk to you is we have a global fleet of bots that are producing a ton of data and using that data to make the system more intelligent and do its job better and actually make the entire sort of supply chain for that organization, for our customers more effective. Um, that's another really important aspect of how we think about, uh, software. Software innovation at right hand.

Speaker B: Perfect. Yeah, and that's um, a great segue. Uh, let's talk about that a little bit. Um, there's a very large amount of data that's generated these days, um, in all aspects of business. But how did you approach the challenge of, um, leveraging this large volume of data, uh, that's generated by robotic operations?

Speaker A: Yeah, it's a great question. I think there's always a kind of a dual approach, uh, when you think about it, when you think about data. Right. It's the engineering challenge for us. We have all these very decentralized architecture because we have all these bots running in the field. They need to have compute and processing on site in order to power the bot. It's producing a ton of data. We need to get that back to some centralized location and have all the infrastructure and pipelines and processing capability, um, to be able to ingest, process and store this data and then a ton of different pieces of software and infrastructure to expose that, to be able to use it. So there's the technology side, but what I think I thought about and what we thought about first was, okay, before we even start thinking about that work, which we kind of know is necessary, uh, where's the value coming from? You know, like, how exactly is this going to affect our bottom line, affect our value, um, make our product more effective for our customers and solve, quite honestly, a lot of problems that we have internally, whether it's support or preventative maintenance or errors or exceptions that happen during the normal operation of uh, uh, one of our right to work cells. Um, and so, you know, having a clear kind of set of goals or an understanding that, hey, this is going to be a really powerful tool to solve our own problems, to solve customer problems, and then going one step further to create solutions and value that you Know, uh, would make us a lot more attractive, um, for current and future customers. That's kind of the baseline, um, uh, for accepting the challenge to leverage these large volumes of data from, from kind of an engineering standpoint. Um, so it was easy to establish that bar. I think when I joined the organization. Part of the reason I joined was to kind of pick up this effort. The funny story that I like to tell people is we have petabyte scale data and when I joined the company, it was a series of daisy chained external hard drives that went back to just one basic kind of server and you accessed it internally through the network. It wasn't in a data center and it wasn't in a climate controlled room and that room actually had a leaky roof. We were in an old building. So this was kind of as basic and uh, almost ineffective as you can get. And the transformation from there in about a year and a half to a completely cloud native scalable platform that was handling ingesting data from our global fleet, processing it and then making it available by a series of APIs in uh, different formats to allow anyone to build anything on top of that and to solve interesting problems or to create value that didn't exist. Um, and I think that's probably another part of the challenge is you should have goals that you're going towards. But the idea behind, you know, leveraging your data is you don't know all the amazing ways that smart people that you hire are going to be able to use this. Right. So how do you ensure that there's a foundational platform or a foundational way to access this, that anyone, whether they're technical or not, can start to dig into this and solve problems or create new ideas, uh, that add value for you.

Speaker B: Thank you. Yeah, that's fascinating. I love those stories of, um, the kind of infrastructures that people come into sometimes.

Speaker A: Um, I'm no longer surprised, Gareth. I'm no longer surprised. I have seen the craziest things. Now you just smile and you get excited about how much change you can drive and how much value it's going to add. It's pretty exciting.

Speaker B: Absolutely, yeah. Um, coming back to what you were talking about earlier on, in terms of the optimization, um, uh, that can be achieved with autonomous piece picking robots, um, how do they interact with, um, data systems, uh, to actually optimize that? Um, I'm just wondering if you have any real world examples of these kinds of applications.

Speaker A: Yeah, no, it's a great question. Um, so it's interesting because our customers, everyone's Sort of fascinated by technology and intelligent robots. And so I think that's kind of like a baseline, uh, reality. But ultimately we're not selling our customers cool technology and ultimately they don't care. Right? We're selling them fulfillment. Right? And when I say we're selling them fulfillment, we're really selling them for the data benchmarks. We're saying, hey, we are going to meet these certain KPIs, these certain metrics related to order fishing, right? If you present us with this much inventory, we're going to uh, hit these benchmarks for how successful we are because like humans, uh, uh, uh, uh, these systems aren't perfect. We're going to make some mistakes. We're going to fail to pick certain items or it might drop. And these are on very small sub single, uh, digit percentage, um, rates. But at the same time when you're picking tens of thousands of items a day, uh, that's still something you can quantify. So in selling fulfillment and in selling metrics that we're going to hit, I think ultimately, um, the people that interact and measure our system from a customer end all they care about and all they're looking at is optimizing the data. Right? And the data is for them is operationally, how well is this warehouse doing?

Speaker B: Right?

Speaker A: Like how well are we, you know, uh, completing our objective, uh, of shipping out this many units or fulfilling this many orders, right? So I think, you know, in sharing success stories, sort of all our customers that use us and have scaled with us, you know, a notable example being, you know, we announced back in February, uh, of a really big partnership with Staples, right? And I think they're the uh, second or close to the first, um, largest E commerce operation for kind of general office merchandise in particular. So talk about metrics and talk about scale. They are very curious about it. And our partnership with them I think really opened our eyes to that to a degree that we hadn't seen, but also proved that our technology was kind of willing, willingness, was ready and able to handle that. Right. Um, so you know, I think as far as how it integrates with their systems to kind of optimize efficiency, right? The very sort of premise of autonomous piece picking in a warehouse implies that hey, you know, you don't need to worry about labor shortages. These things can run sort of more than just eight hour shifts, more than one eight hour shift. For example. You can start to integrate them with systems like uh, asrss, that's automated storage and retrieval systems, um, for people out there that know those are Uh, a big one on the market that's really powerful is Autostore for example. They can integrate with those in a way that is incredibly natural. It's like plugging software together. If they have a central controlling system that can call up ASRs and say hey, we got to fulfill an order. There are these three items in the order. Bring these totes over to this robotic piece picking station. That's where the right pick is located. Right. And uh, from there I'm going to send a command to uh, right pick and say hey, you should be expecting these totes with these items. Pick this many of each, ah, place into this order, confirm that everything was there, um, and we're good to go. So kind of that sort of tracer bullet through the system of uh, you know again almost like software components is what they might remind people of. But they're actual, real functioning, you know, physical, physical machines uh, cooperating with each other, driven uh, entirely kind of by traditionally uh, like a warehouse execution or warehouse management system. Right. Um, that is just kind of the prime example of uh, data systems kind of efficiently working together. Right. Uh, I think some, some real world applications I told you about Staple, I think Staples. I think one of the really exciting things for us is um, Apple, it's an online pharmaceutical company in Scandinavia. Um, they were one of our first companies many years ago and they scaled with us. They have a 16, uh, bot installation and uh, like I said, I think earlier, it's a pretty much a lights out 247 operation where at any given time they're fulfilling orders autonomously. And these 16 bots are sort of picking items and fulfilling orders uh, around the clock. And we've had the privilege of visiting the site and seeing them in operation and it's just incredible. Right? You know what I mean? They're just surrounded by all this inventory that's moving through the system at a high rate. And um, you know it really is again it's kind of a model for you know, this is not just doing something that was done in the past. Like this is creating a new way, um, uh, for companies to be able to provide fulfillment for their products. Right. And it's, it's a pretty amazing thing to uh, see.

Speaker B: Yeah, it's fascinating and I think it's um, as I said at the beginning, you know, it seems that it's becoming more and more common in different kinds of industries and hearing those examples is really uh, enlightening. Um, another area where there's been a lot of development over the last couple of years is, um, in sort of AI being even more at the front of everyone's minds, uh, since the GPTs came onto the market. Uh, how do kind of AI and machine learning intersect with robotics in your work at Right Hand Robotics? And what are the main opportunities and risks associated with it?

Speaker A: Yeah, it's a great question. So, you know, at AI, uh, we don't have a product, um, so kind of just walking through the system, you know. So to be able to see and visually interact in the world, right, we rely on, um, computer vision, um, particularly on sort of, uh, ML techniques around computer vision that involve, in our case, you know, uh, CNN's convolutional neural networks that are able to sort of process, uh, uh, uh, visuals, uh, from the real world that are taken by our system and to be able to kind of make sense of that. So for us, the most important thing is, hey, if we're able, we're manipulating objects in the real world, we need to be able to detect what's an object, where is it in space, what are its contours and outlines, Both from a just kind of 2D standpoint. I'm seeing, I see a coffee mug on my table, but also. And this happens in the human body and just we take it for granted. But understanding that, hey, I keep going back to this, but my coffee cup, it's one foot away from me, and I need to extend my arm this much, right? So that's motion planning, for example. And you need to be able to sort of intelligently grasp this item, because I know a little bit about it based on what I can see and maybe what, uh, my memory is of it. In the case of us, our memory is the Warehouse system saying, hey, this is a coffee mug. These are its dimensions. These are its. This is its weight, for example. So I have an understanding of how to interact with that. Right? So none of that cycle starts without, you know, being able to see. Right? And that's, you know, a clear application of AI, specifically machine learning, in, uh, the vision space, to be able to segment, as we call it, and detect an item, uh, in a color image. Right. Um, so, you know, that's table stakes, right? Without that, we don't have. We don't have a business, um, and there's nothing we can do beyond that. I think in addition to our computer vision kind of capabilities, we employ AI and are, uh, sort of working to employ AI, uh, really just on the data level as well. We have a global fleet that completed over 50 million picks in the lifetime of. Right Hand robotics, that's a lot of information that you can learn a lot from. Whether it's incredibly basic things like, hey, you've asked me to pick this item a million times and you've told me it weighs a kilogram. And I have intelligent mass estimation because I have a robotic arm. And based on all the forces that are being applied, I can detect, uh, in that moment I can get a rough idea of how big is this thing, what's the mass of this thing that I'm picking up. We can see back customer and say, hey, we think you have the weight wrong. And how does that affect them? Um, well, guess what? They print shipping labels based on what they think a weight will be and they pay according to that. All of a sudden now with just a very basic analysis of the data. I don't even know if I necessarily call it AI, but it's very, perhaps classical AI to just be able to analyze and look for statistical anomalies based on, based on combing over the tens of thousands of data points we have for these millions and millions of uh, picks that we've done over the life cycle of the company. There are so many types of AI that we can apply. Um, uh, like I said, computer vision is a huge one. I think the ability to again, take all that data, uh, uh, uh, that's based on uh, sensor data that we get from the bot and the gripper combined with images and be able to understand when we're manipulating objects, what's the best way to interact in the real world with an object of this certain type or these dimensions, uh, what's the best way to interact and grasp, uh, notebooks or specific geometries like toothpaste tubes or pharmaceutical medicine, uh, boxes, for example. This is all information. And this goes to the learning aspect, which is a big part of why people employ AI, right? They want the system to do pretty well, um, at first and continue to get smarter and smarter and smarter. That's the promise. That's the dream. Right. And that's kind of the entire premise, uh, uh, for right hand robotics. Right. Even in the computer vision aspect. Right. Like our ability to detect and segment objects in the real world continues to get better. Why? Because they're deployed in the real world and we're constantly learning where we're doing well and where we're doing not so well and sort of feeding that back in, in sort of a curated fashion, uh, into our models to be able to train more powerful ones over time that are able to more intelligently, uh, separate and Understand, um, what it's seeing in the real world. Right. So yeah, I mean we could make this the whole podcast because there's so much we're doing and there's so many opportunities. Uh, but ultimately this easy task for humans of picking up things and manipulating and grasping objects in the real world, it requires a lot of intelligence that's totally subconscious for us. And in order to make that practical and commercially viable, we uh, have to have systems that employ a ton of AI in order to make that uh, possible.

Speaker B: Yeah, I think it's fascinating. And the thing you were saying about being able to detect the weight of an object and then the potential consequences of that and maybe be able to save money a way that you weren't expecting, it's a really compelling example of how it can actually affect the bottom line of a business. Um, absolutely. So just to, um, come back to some of the data elements, um,

Speaker A: what

Speaker B: are some of the challenges associated with data management and analytics in robotics that data leaders should be aware of?

Speaker A: Yeah, that's a good question. I think there are some challenges that you face across the board. Outside of robotics, I think we share that understanding where the signal is, um, in a petabyte scale data, it's not all signal, a lot of it is noise. And noise in our sense would mean it's just data that probably doesn't have a lot of long term value. Um, but you don't know what you don't know and you don't understand what you need, particularly for really powerful and complex systems. So you document and store as much as you can, um, so that you can go back in time and potentially recreate or reanalyze things in interesting and novel ways. The challenge of managing that data at scale over time, just like in any other application, I think specific to robotics, um, it goes back to the aspect of having a connection with the real world. We're not just storing numerical data, for example, or what we would think of traditionally as numerical data that would fit kind of nicely in a relational database row. Right. We have tons of sort of um, sensor and telemetric data, uh, that otherwise wouldn't be present in software systems. We obviously have tons of visual data, right. Um, both still images and video data that I think it's just very rich and much more complex and yeah, just much more difficult to deal with from an analysis side and really kind of gain, uh, the right insight that you would want to gain from this data. You can imagine, hey, if we're recording videos, uh, uh, uh, of what are happening in our robotic work cells for a global fleet. There's a lot of interesting things that you could mine from that. But doing that is not easy. The techniques and understanding how to actually extract the signal from that. It's not as straightforward as hey, I want to analyze uh, going to sort of ah, online advertising space. I want to analyze the best way to get a click through rate. That's not easy either. There are a lot of people trying to figure that out and that's a continuously evolving problem. But at the same time I think the dynamics of how to get and move forward with that data, it's a little more straightforward. I would say the traditional problems you would face in any large scale data system are there. And then I think robotics add the little twist of we have really high volume data, um, and that high volume data can come uh, uh again in types of data whether they be visual or telemetry data for the bot that are a little more non standard and require a little bit different processing and different sort of analysis uh, than you typically might get from say a financial or sort of ah, a pure software ah, company like a SaaS company that's analyzing again retention rate or um, product usage or anything of that nature.

Speaker B: Yeah, it's fascinating. I wonder whether uh, AI could potentially have a role there as well, analyzing some of that unstructured data and helping to come up with those uh, insights. Um, well the time's absolutely been flying past. I just got a couple more questions for you. Let's look to the future a little bit. Um, from an engineering perspective, um, what kinds of trends or innovations do you foresee shaping the future of robotics in the next five to 10 years?

Speaker A: It's a great question. I hesitate to speak to that simply because I have so many great colleagues who are still uh, still uh, in the lab, they're still in academia and those individuals really are pushing the boundaries free of kind of the commercial limitations that many of us had where we have to ship a viable solution that's adding value to the customer now or in six months as opposed to five to ten years. Um, but I think what I can tell you from being on the ground is to reemphasize, I think exploiting innovation in robotics through software. I think that's going to be a continued trend and a continued theme for uh, a while. I don't know if it's the next five to 10 years, but it's definitely the next couple of years. Um huh, for a lot of reasons I think uh, uh, hardware development and R and D, um, is incredibly expensive, it's incredibly uh, uh, um, resource intensive. And uh, as companies try to solve problems, not just in fulfillment supply chain but across the board with robotics, right, at the end of the day they need to, you know, there needs to be an appetite from an investment standpoint. There needs to be sort of a viable business model, uh, for paying back or sort of justifying that kind of R and D investment. Right. And you know, software is cheaper than hardware. It's a mantra that you hear again and again and it is definitely true. Right. So you know, the themes that we're seeing as far as the software, First Robotics, right, And treating the hardware more as a commodity, um, I think that helps to bridge the gap between what we have today and where we still need to go. Right? Because you know, I always say this to people, you know, robotics as an actual commercial solution, uh, people might think, yep, it's here, it's been here for a while. It's still a frontier market, right? Um, you know, a lot of verticals, a lot of markets, a lot of industries are still figuring out how do we use this technology reliably, consistently and at scale. Right? And there's still a journey there. Uh, I think the market uh, uh, has uh, to go kind of across the board. I think again in certain industries it's a lot better than others. Automation has really kind of heavily adopted robotics is an important part of its business model. But at the end of the day it's like even in automation we're still overcoming challenges of, of legacy sort of setups and infrastructure that were built in the 50s and the 60s and the 70s that are really uh, ah, well accustomed to new technology like what we have. Right. So going back to your question, I think from an engineering standpoint, uh, a software first mindset is going to be huge. And I think robotics is at the point now where um, that frontier market status is almost at an inflection point. And there's going to be a lot of focus on, okay, we have all these technologies, we have all these innovations. Again, how do we apply them in a sustainable and economically viable way in order for us to um, stay around in order for this to not just be uh, a uh, cool interesting thing that we might see at some warehouse or another, or a cool thing we might see. There's a lot of medical robotic applications that we might see one off in a certain hospital that we maybe have visited. But how do we make this sort of uh, uh, go into mass adoption mode? Right? And I think, you know, I Think, I think for that to happen there needs to be sort uh, of a lot of innovation on the front of, you know, making these solutions and making these technologies uh, work effectively. Right. And work effectively in a safe way in order for them to, you know, deliver on the value that customers have to spend in order to implement them. This isn't cheap. These systems are uh, uh, uh, incredibly expensive. To reflect both how powerful they are and I think to reflect again the investment that was made in order to develop such kind of smart and intelligent solutions. So really a drive towards practicality with software. While it might sound a little bit boring, I think that's really, um, I think that's really kind of be an important theme you're going to see with a lot of uh, exciting robotics companies. Boston Dynamics, I brought them up. I have a lot of close friends and colleagues that I've worked through the years that ended up there. Everyone loves Atlas, which is a bipedal robot that runs around. What is their commercial focus? Um, it's a robot that unloads boxes off of a truck. Even for them, it's taking this incredible technology that they've developed in the control space, in the software space, AI space, and applying it towards solving a real world problem where it might not be as exciting and as sexy, for lack of better words, but it's going to revolutionize how fast we move things and how fast uh, the economy operates. So it will be incredibly disruptive. Just won't make us as interesting YouTube videos. Yeah.

Speaker B: In terms of delivering value in a kind of safe way, um, fast food production. What do you think? Am I likely to get my uh, Big Mac served up by a robot anytime soon?

Speaker A: That's so interesting. Especially in the Boston area, we had quite a few, a spate of startups that were basically, they were fully automated, quick serve restaurants, throwing out names. If you think of a sweet grill or a Chipotle, like these bowl based companies, and not maybe akin to the fast food like sandwich and side kind of thing, but like these bowl based companies. Right. Because what do you do, right? You just drop a bunch of ingredients in a bowl and you package it up. I um, think like with other areas where robotics made a splash, right. I think it's a lot easier to make a splash and get people excited than it is to have staying power in an industry. Because at the end of the day, is it exciting and cool to see a robot make your sweet green salad bowl? Sure. Is it good for the bottom line of sweet green? If the answer is no, one of the things that I forgot to say in your last question was people always ask, what's your biggest competitor? It's not other cool, interesting robots. You know what our biggest competitor is? Not automating. Not having a robot. Right. So, like, you know, I'm glad you brought up this question because that's a point I love to make. Like, hey, what's the future of this industry? What's the future of robotics? And really, it's being better than the alternative of nothing. And nothing is really just the status quo. Right. So, you know, particularly in the fast food or quick serve space, um, I think it's really exciting. I think that we will get there. I think it makes a ton of sense. Right? Because then you can open these kind of almost like micro fulfillment centers for food where you can have a tiny little outpost for your company that's making like 10 or 20 bowls every couple of hours as opposed to pumping out hundreds, maybe like a larger sweet green or a larger Chipotle or something of that nature. Um, and it's not going away and we've seen it and it's happening. The question is, how is it going to be viable commercially? Because there's technology being effective and working and being cool, and then there's that technology actually making sense, uh, uh, from a financial standpoint. And we have to bridge that gap. As an engineer, I want to bridge that gap because I want this cool technology around us in our everyday life and we're experiencing that all the time. And then I think from a societal standpoint, you lose out on all this advancement, unfortunately, if you can't make it viable. But hey, that benchmark of being viable, that's been a good one. That has driven innovation for hundreds of years. So it exists for a reason and I'm confident it's going to get us there.

Speaker B: That's awesome. Well, we've got a lot to look forward to. Um, uh, just before I wrap up. Yeah. I'd like to say thank you, Ian, so much for your time today. It's a fascinating discussion, uh, for our listeners, just some final thoughts. Um, if we've got data and analytics leaders out there, uh, who are considering implementing a robotic solution, what should they think about? What would your key takeaways be from this or your advice to them about what they should think about?

Speaker A: Yeah. Then for the question from the standpoint of would you say, uh, uh, uh, people in my space or customers or just in general? I think with data infrastructure? Because I could go a lot of ways with this. So I want to make sure I'm giving you what makes the most sense.

Speaker B: Yeah. For data and analytics leaders out there, uh, who are maybe considering implementing a robotic solution in their organization.

Speaker A: Gotcha. Uh, gotcha. Excellent. That's my favorite question to answer because those are our customers and our potential customers. You know, like I said, I think any company that's selling robotics to solve automation or to solve a manufacturing, which is kind of like a, like a sister industry, I guess, uh, to what we would do, right? We are selling them or they are purchasing or real estate, they're purchasing fulfillment, they're purchasing metrics, right? So they can't judge what's going to give them a positive roi, what's going to return their investment on these interesting and cool systems if they don't have a really strong command of their data. Um, in supply chain logistics, in uh, warehousing, that's pretty much the bread and butter, right? Like you are operating on, always on the edge of ensuring that your metrics and that your KPIs hit the mark. That is your kind of, uh, guiding force and your driving light, right? So they tend to have a really strong understanding of their data, which is great. But I think the key is understanding how to marry what you need and what you expect. M from a KPI standpoint with how are these new robotic systems going to give you an edge there? How are they going to improve those metrics, right? You know, how are they going to allow you to potentially solve some of the bottlenecks or some of the issues that you have right now in interesting and creative ways that wouldn't happen given the current configuration or the layout, um, of how your, how your warehouse or how your fulfillment solution operates, right? So I think, you know, kind of to dis still that the key advice is really know your numbers, right? And understand how to push the boundaries of your KPIs and of your metrics, um, by doing things a little bit differently, right? Because I think introducing automation, um, it almost, it's like it changes the denominator, right? It changes sort of the benchmark or the coefficient by which you're kind of figuring out like, are things flowing and working well, right? And ultimately you want to reach that same goal of I want to fulfill a million orders in a day, right? But how you get there and what you measure in order to get there and what sort of factors you can play with, it changes a lot, right? Like we talked about, suddenly you're not constrained by time. Suddenly you are able to pick things. 24. Excuse me, fulfill orders 24, seven, potentially. Right. If your automation is set up, um, so the metrics of how fast you're doing it, suddenly that changes. Suddenly you don't need to think about the rate at which you're getting things done in two eight hour shifts. The same if you have 24 hours to work with. Right. So in a sense that metric or that piece of data you're less concerned about or perhaps the goalpost gets a little closer, it gets a little bit easier. Um, uh, whereas now when you think about resiliency and order consistency and order integrity, those are things now you're much more focused on. Right. Um, because again, it's like, you know, you don't have human verification. You're relying on a system that we would argue is actually, uh, uh, a lot easier to work with. Right. Because, uh, it always tells you the truth and uh, it only knows as much as we can sort of, uh, you know, we can sort of prime it to understand, um, and it always has a confidence score for how well it did and you can access all its decisions. Um, and you know, there's going to be total transparency and truth there. Right. So, uh, yeah, that's kind of how I would approach it. Right. Is know your numbers and then understand how the equation might change when robots all around your warehouse and throughout your supply chain are kind of, kind, uh, of upending how you think about getting things done, but ultimately giving you a lot more flexibility and making things, you know, a lot more scalable.

Speaker B: Well, thank you so much, Ian. Again, it's a fascinating look on how, uh, you know, business operations and uh, warehousing and all sorts of other things can be affected by robotics. So thank you for joining me and I'd like to say thank you, uh, to everyone who tuned in too.

Speaker A: Awesome. Thank you so much, Gareth. It was a pleasure. We hope you enjoyed the episode. Be sure to subscribe to the Business of Data podcast wherever you're currently listening. And keep up with us on socials. And for more content, visit us on, um, businessofdata.com be well and thank you for listening.