#283 Andrei Danescu: How Dexory Uses AI & Robotics for Warehouse Management

0:00If you think of a big space where you have a large number of pallets, because we can do the digital twin side of it, which is a 3D reconstruction of the space. we have a lot of volumetric information we have a lot of visual information so if you're a large warehouse that does for example e-commerce and you store the same goods in in 10 locations we can measure the quantity the space occupied and say okay you're actually utilizing 20 30 of the volume the cube allocated for this palette so if you optimize the way you store them and optimize the position in space and the position in the warehouse you can actually aggregate this 10 locations in three.

0:39We are focused exclusively on the intelligence layer in the warehouse space. Our robots go around the facility, they collect vast amounts of information and data. They literally created a 3D digital replica of that environment and we use this information to drive optimization, to provide end-to-end visibility, to remove operational errors. I'm Andrei, i'm one of the co-founders of tech story i have a technical background my background is in electronics in robotic systems so very very passionate about product very passionate about problem solving and i have a special passion for the kind of input we're bringing to the logistics industry and i'll tell you a little bit about why that is so at the start of my career i started working in motorsport i started working in formula one wow yeah it's it's it's been really really good fun so i was traveling around with the team and my job was literally looking at sensor data and being able to to make split second decisions and relate that to the team and and help us become faster become stronger and obviously win the races um now i don't know if you know much about f1 but But the world of F1 is very, very fast.

1:59There's many races that happen around the world. So there's a lot of logistics involved in this. So basically, we would travel from one country to another. And there were always a couple of glitches, a couple of hiccups. And I thought, you know, we have this incredible technology, incredible, super fast cars, very, very detailed information from the data. How can it be that logistics doesn't know exactly where everything is at every single point in time? So that kind of stuck with me for a while. And after I left F1, I decided that I want to build a company that uses autonomous robots. Robots have always been one of my big passions and one of my deep loves for technology.

2:42I wanted to use autonomous robots to digitize offline spaces. And a big application for logistics is literally to provide this level of visibility and to eliminate any kind of operational errors from warehouses, from the wider supply chain. So that's the combination of some of my background and how we got to the technology that we have today at Dexory, which is solving exactly those problems, eliminating any visibility gaps and eliminating any operational errors for logistics. Yeah. A couple of questions. Which team were you working on or did you work for several? So I used to work for a team called Force India, which is now Aston Martin.

3:23Yeah. Yeah. You know, I, along with a lot of people, didn't know anything about Formula One until the Netflix documentary or series came out. To be honest, it was before the change in management and before Netflix actually was a dying sport, believe it or not. Oh, is that right? Yes. So the audience was declining because there wasn't that level of engagement with the fans and everything. And now it's massively spun back out and it's huge everywhere. Yeah. Yeah. It really is remarkable. And your educational background was in engineering or in computer science or what? It was actually in robotics and basically electronics and control systems.

4:09Yeah. you know in Endevery's technology you're not are you building robots or are you building sensor systems to track robots so basically at Dexory we we're a full stack company so we build autonomous robots we're very proud to say that we have the world's tallest autonomous robots but most importantly part of being full stack we build every part of the technology from the robots, which are used as data capturing, through to the digital twin platform, which is a software platform, the product that our customers actually interact with and spend their time getting value from. Yeah, I've had Peter Chen on the podcast.

4:56He has a company with Peter Abiel out of Berkeley called, they changed the name of it. It's anyway, it used to be called embodied intelligence. It'll come to me, but, and they were focused on picking robots and, and the visual computer vision uh behind picking robots is uh is that the same as you guys or are you more focused on uh you know retrieving bins in a warehouse and no that's very very different um we are focused exclusively on the intelligence layer in in the warehouse space so our robots go around the facility they collect vast amounts of information and data they literally created a 3d digital replica of that environment and we use this information to drive optimization to provide end-to-end visibility to remove operational errors and to ensure that things like picking robots things like the the warehouse operator uh for operated forklifts things like the people that work in warehouses and actually do the physical action of picking and retrieving items can operate in the most efficient way.

6:29So the way we look at our technologies is a supercharger for the existing workforce because warehouses are difficult environments to operate in. So we strive to make them better. We strive to make people's jobs easier and provide this level of efficiency and make sure that we eliminate errors. So very much focused on the optimization data intelligence side rather than picking and end physical actions or interacting with the space in the warehouse right but you do produce robots so what are those robots doing so we produce robots like i said we have the the world's tallest autonomous robots they are going around the space going around the warehouse and they collect data they collect uh they take pictures of pallets they collect 3D scans of the warehouse.

7:16They transform or transpose the warehouse from the physical space into a 3D digital replica, enabling this level of optimization and this level of data processing on the physical space. I see. But you're not building robots to move physical objects in the warehouse. No, no. So it's all about collecting data and gathering information. Right. And is that for optimizing the human workers or is it also for optimizing robotic workers in the warehouse? So you can apply the technology to both. But I'm a big believer into solving the customer's problem in today's world. So rather than building a technology and then saying, okay, you know, in five years, 10 years time, when we're going to have a lot more automation in the workforce this is going to become even more relevant we want to focus on helping our customers and solving their problems they're pressing problems today and today the vast majority of what you find in warehouses are are people moving things around and doing the hard work doing the the hard jobs so it's very much making their lives easier and making their jobs easier and their environment better yeah that's interesting the um so you create this digital twin of the warehouse and can you give me an optimization example that you would use that digital twin to solve yeah so um if you think of a big space where you have um a large number of pallets because we we can do the the digital twin side of it which is a 3d reconstruction of the space we have a lot of volumetric information we have a lot of visual information so if you're a large warehouse that does for example e-commerce and you store the same goods in in 10 locations we can measure the quantity the space occupied and say okay you're actually utilizing 20 30 percent of the volume the cube allocated for this pallet so if you optimize the way you store them and optimize the position in space and the position in the warehouse you can actually aggregate these 10 locations in three and you make or four and you make space for another five six pallets which means you can put a lot more product through that environment which means you you make this transition from warehouses being a cost center into warehouses being a profit driver so you can actually do a lot more with the existing space just by optimizing the way you run the operation i see yeah uh and uh so uh uh sort of storage optimizing storage is uh one of the main uh things that you do uh what about uh traffic patterns and things like that in in the warehouse do you do you analyze that as well so uh we do and this is where it gets really interesting because um we can have a we can have a transition we can have parallel to using AI and machine learning technology for this so again going back to the way we acquire the data we have a digital twin we have an understanding of how goods are moving around the warehouse so the rate of depletion the speed of picking the way items are actually being collected and they move through the space and from there we can infer okay this is um this is an aisle with a lot of fast moving goods so there's going to be a lot of people there's going to be a lot of forklifts and machinery in the space so therefore there's always a high traffic and high congestion area so we can re-optimize the way the warehouse is being organized or the way the goods are being stored in order to distribute this high um highly populated or highly high traffic area across two or three or four different aisles massively alleviating the the way people are moving around this machinery and of course enhancing a faster peaking and better productivity for everyone operating in that space.

11:27Yeah and this is the core product is Dexery View is that right that is that the tall robot with all the sensors on it that glides around mapping the space? Sorry can you repeat the first part of the question I think we had a small is the core product product called dexery view yeah so that's the that's the core product and that's um the offer the product offering consists of the robots that acquire the information and dexery view is obviously the software platform where the digital twins are uh continuously updated and where customers can actually log in and interact with uh with the digital version of their warehouse yeah and then does uh does a warehouse manager experiment with the digital twin or is there an optimization algorithm that that immediately gives you two or three configurations uh and the manager can decide among them so at the moment we have algorithms that offer various types of optimization um so the the volumetric the the space utilization is an example and there's a number of other options that we offer and what we're building and what we have on the roadmap is exactly what you uh what you hinted to is the ability to run simulations the ability to analyze different scenarios of the warehouse and this is the kind of technology that uh we see a lot of customers being very excited about and what uh what we're looking to build and release over the next quarters into the product as well.

13:09Yeah, you said at the beginning that you want a product for today's use case, but warehouse robots, that space is moving fast. Are you integrated with any of the robot companies that are putting robots into, I mean, ABB, I think, is one. And, you know, there are a lot of them. I mean, Amazon has it bought that, I can't remember the name right now, but that robot company that it's renamed that has their flat robots that you stack pallets on and then they move them around. yeah so those are those are different different um different types of good movement solutions so for example you can have a goods to person uh system you can have incredible companies like like locust that have picking robots where they rely on the robot going in the proximity of a person and they put various uh products into the into the robot and then the robot carries on to another location or picking arms like you mentioned from abb at the moment we don't tend to integrate with the systems because they're relatively new into the space but we have built the right level of apis and we've built the system in a way that works with the latest and most up-to-date standards like vda 5050 and so on in order to make sure that in the future as the systems become more and more present in the market we're well positioned to be able to have this most pain-free integration with such uh such additional technologies yeah um and and how long do you think before there's a completely automated warehouse i mean one of the things its covariant was uh peter chen and peter abil's company uh and they're working with abb or were at least uh one of the things that fascinated me is one you can operate a warehouse 24 7 and two you could operate it in the dark because the the sensors don't necessarily need need light to operate is how long do you think before that is a reality where you know Dexery goes in and does the digital twin and then you give that information to a robot warehouse company and and it uses that information to guide its robots so that's that's definitely definitely an interesting question because um I think that biggest challenge or one of the biggest challenges we see in the logistics industry is that there's an ever-increasing um labor gap so the labor shortage is real so um whilst the opportunities there the industry is growing um obviously we're doing more and more as as humans as a species we're doing more and more that relies on or heavily relies on logistics we need to make sure that we first and foremost can thrive and create an efficient supply chain so that's one of the that's one of the big drivers for us is how do we supercharge the existing workforce to make sure that you can have the maximum impact with the relatively reduced number of people the reduced workforce that you have now so i think whilst dark warehouses are a reality now it really depends on the type of operation that you're looking to run so even in in highly automated environments what we find is there are tasks that are just not lending themselves to automation so it's a lot more complicated or it requires a level of technological sophistication that just doesn't have a direct application to make it commercially viable and what we're also finding is that customers become more and more demanding so they want to have a very versatile logistics partner So there's a very good space and very good place for enhancing the existing workforce.

17:38Like I said, this enhancement comes in many ways. So if you have a team of 10, 20, 50 people, if you're sending them around the warehouse trying to find the gaps that you have to store more goods, it's not maybe necessarily the best use of their time. if you can give them a completely optimized way of running their day-to-day operations you can achieve a level of efficiency that has not been seen before so I think the combination of a dark warehouse that can run 24 7 can can provide a level of optimization and efficiency in conjunction with a very powerful workforce is is what we're what we're looking at creating for the future and is definitely going to give us that robustness and flexibility to have a supply chain that is as independent to external factors as we can make it.

18:29Yeah. And the Dextry, it maps with the Dextry view or with the tall robot, maps the space. what are the algorithms then used to for optimization so when you look at when you look at breaking the technology down into some of the the component components or the subsystems so we use the autonomous robots they are fully autonomous by which I mean they use slam to be able to map the environment to be able to navigate we use a combination of lidars in camera technology to be able to understand and perceive the environment in which they operate. They then have the ability to do a 3D reconstruction of the space to be able to take pictures of pallet locations.

19:23So the human operator can then go into the digital twin and understand, okay, this is how my goods are actually looking in the warehouse. This is the environment in which I'm storing them. So for example, in an FMCG warehouse where you have perishable goods, you don't want to be storing them in inappropriate conditions. too hot or too humid and so on. So this is some of the core technology that enables us to map the spaces, to be able to collect the sensor data. And then index or review as you go into the digital twin, we have a fairly large number of technologies that drive this optimization.

19:59On one hand, we have different computer vision algorithms that enable us to extract a lot of information from images. On the other hand, we have machine learning systems that can identify, that can extract and segment various images, various areas of the warehouse. And then when you put these data sets together, we have a layer of AI, a layer of different types of AI technology from small language models into agentic AI and so on that can drive optimization. So it can tell you how to best organize certain elements, certain parts of warehouse can drive process optimization can tell you how to best organize process in the warehouse and and obviously the agentic ai layer will be able to run this continuous simulation will be able to run a lot of this um what-if scenarios and simulation scenarios and analysis in order to tell you what the outcome would be of of this of these actions or the proposed actions that the system could take yeah and and do you have metrics on on what kind of uh on what kind of improvements you can reach with this kind of optimization um yeah i mean a lot of it revolves around um error elimination so being able if you if you can run your operation with uh with no errors or no um of eliminate all the inefficiencies then you've pretty much achieved the holy grail.

21:30We have quite a few papers around the ROI and the speed to value that we drive for our customers. I think it's quite difficult to get into one unified metric because this technology is versatile and applies to a wide range of logistics operations. And various types of warehouses will tend to have various ways of running. You could have a high velocity operation where you have a lot of goods moving in and out. You could have a high volume where you have hundreds of thousands of pallets or a high value operation where it's very, very important to know at every single point in time what's going on because you're storing items that are incredibly valuable.

22:12Think of it as, I don't know, a pallet of iPhones is still one cubic meter, but it's completely different in terms of value to the same pallet of flour or rice or whatever else. sort of like goods we might have there. So different types of warehouses have different ways of extracting value. But like I said, it revolves around eliminating errors. It revolves around this orchestration of the operations in the most efficient way possible. Yeah. Can you give an example of an optimum? I mean, you talked about increasing, consolidating storage in a warehouse to create more space. But are there some other examples you can give of optimizations that maybe allow people to do more or fewer people to do more than they would otherwise?

23:17yeah so um i mean we can look at i think we touched quite a bit on 3pl or logistics style warehouses but we could look at warehouses that for example are serving um production lines so we work with a number of customers within automotive within various types of oems um and for for them for for this type of customers the operation needs to continuously flow so if you if if you imagine a large um automotive oem that would stop a production line that would cost millions of dollars in a very very short amount of time and those losses will start start accumulating over over days and so on so being able to ensure that the production line never stops that you never run out of materials it's it's a big it's a very very big cost um cost benefit um and this is an example to automotive, but for example, we have customers that produce electric motors, we have customers that produce furniture and so on.

24:22For them, equally important, they cannot stop the production line. They would start incurring a huge amount, huge bills of losses. So we would be able to find any product anywhere in the warehouse. So if before you would spend hours trying to locate these goods in the warehouse, because you still need to be able to produce the items and then obviously serve customers with Dexa review you have complete visibility over every single location you have in that facility so it means that you can actually go into Dexa review you can go into the platform look for whatever goods you are trying to find and immediately find where they are see a picture see when it was last scanned and basically be able to retrieve them ahead of any uh blockages or any kind of bottlenecks even happening yeah so so does dex review also track inventory of in real time so you always know your inventory levels for every sku yeah so it it tracks um it tracks the um quantity it's something that we call uh peak availability or peak face counting so we track the amount of product you have available for picking available for for taking to customers whether that customer is um you know the end customer if it's a B2C or whether that customer is a sub-assembly or sub-component manufacturing line within the larger plant.

25:49Yeah, actually, I'm not the most organized person in the world. I'm always losing stuff in the house. And I've often thought it would be wonderful if you had a system that would track every item every time it was moved. so uh you know you could look into a digital twin of your house and do a search and it would tell you right where the thing is um are there other applications for this beyond warehouses um yeah just just before I go there for example imagine the house would be a million square feet and you have a hundred thousand locations where you could be losing things it's very easy to see how this can how this can snowball in something really really inefficient very quickly um and any for example if we were to touch on on the latest tariffs and the the international trade if before you could say okay my pallet is worth ten thousand dollars and and i lost it final right off ten thousand dollars now with tariffs that palette could be twenty thirty thousand dollars suddenly the value has gone up a lot which means there's a much there's a much tighter focus on logistics excellence and operational excellence and and pretty much all our customers are seeing the the ROI of our technology and our product massively increased over probably the last two quarters because suddenly they started to have a very tight focus on how and and and where can they extract every bit of optimization from their operation so if we were to look for example for example, outside of warehousing, this has a very good effect at a wider industry level.

27:30So if you think that you're a logistics operator or you have multiple facilities and you have text review in all of them, the same would apply. You can literally log into one platform and you can see across all your location how they're progressing, how they're performing. You could have an operation that runs at 90 % efficiency and then you could have two other operations that run at 80 85 percent efficiency what dex review enables you to do is is virtually copy and paste intelligence you can look at why is this operation running at 90 95 percent efficiency what is the general setup and how does this compare in a unified platform against all my other facilities and that's incredibly powerful because it's not just about being able to collect the data but it's about being able to present this data in a unified way that enables you to compare and contrast across any facility you have pretty much anywhere in the world yeah that's fascinating

28:33and the i'm sorry i'll cut this out but there was a question i was going to ask

28:45on

28:49Yeah, I was going to ask about the speed that this operates. I mean, is... Well, let me... Yeah, ask something else. The optimization algorithms, are those proprietary? Are you using standard models from the research community? So, we use a combination. Of course, the models per se are, there's a lot of open research work that you can leverage. But what is really important to make the best use of this technology within the vertical space that you're operating in, the data becomes the most important asset. And to be able to acquire this data, we have obviously built the technology that enables us to do so.

29:42But we have huge amounts of data, which in turn allows us not just to best to create these bespoke models and to create the bespoke algorithms that we use in our optimizations, but also it enables us to train various models for this specific applications. Because, and I was recently listening to an open AI discussion and they were saying, look, there's like 10, 15 % of the world's information is available on the web. but the vast majority of it sits behind you know closed doors in industrial environments in in the in the real world applications in the let's say classical industries and that's where it you can drive a huge amount of value if you can access that data if you can start harnessing the insights and the power of the information you can really drive a huge impact for the industries and and and bring real benefit to the way we're actually doing things.

30:38Yeah. This optimization part, are there, I mean, outside of warehouses, for example, could you apply it to, I mean, is it specific to internal configurations? or you talk about the supply chain generally, maybe between sites or globally and moving product around, could it apply to a larger footprint like that? I mean, could you create a digital twin of your entire workflow from factory to customer and use the optimization algorithms on that? So I would say yes. It's not something that we're focusing on because we're very, very focused on the specific industry applications we have now. But as we evolve and expand the DexRateView platform, the vision we have is to build this into a warehouse operating system and then into a broader supply chain operating system.

31:52So you can have a good example to what we were discussing. You could have DexaReview present in a warehouse that serves a production line. You could have DexaReview present in warehouses that serve third-party logistics, or they even serve the retail stores and the retail centers. being able to close the loop and being able to provide this level of visibility and optimization across the wider supply chain and logistics industry means that we have the um we're best positioned i would say to be able to drive this um this level of efficiency across the wider industry and that that's very much where we're looking to take the technology is um a lot broader and a lot wider than the beating heart of the logistics industry which is the warehouse where we started yeah and you you formed the company in 2015 is that right yeah so uh me with with my other co-founders we started this business in 2015 we bootstrapped for a long period of time because as a full stack business there wasn't let's say that the robotics wasn't as exciting back then as it is today as an investment thesis for a lot of people.

33:06Yeah. And then how does somebody deploy this? The package includes the scanning robot and the platform. I mean, how long does it take for a company to get the system into their warehouse? So it's incredibly easy to get started and deploy the technology. So we're operating based on a subscription style of agreement. And we've created the technology with the end customer in mind. So we spend a lot of time working with the industry and understanding how people see robots coming into warehouses. because at the end of the day, a robotic system is a means to an end. It's the ability it gives you to collect the information, to collect the data.

34:01So we created a technology that requires no infrastructure changes. You literally don't have to make any adaptations or any changes to your environment. You don't have to put markers. You don't have to really do anything. We would come on site. We will install the robot. but it takes three, four days to be able to install and map the entire facility to build a digital twin. And within less than five days, you can actually see value from the product. You can see value from the technology. Obviously, that includes a training element because digital twins and the platform like we've created, it's not something that the industry has seen before.

34:38It's a very, very new product into the market. So the training is a very important part of it to make sure that our customers have the best experience with the product and also to make sure that we answer their pressing needs and we answer the pressing problems. Yeah. Well, when you began productizing this, did you work with any of the F1 teams? No, we didn't actually. They tend to have a very, very different style of operations, not necessarily something that would lend itself to this. right right uh and and you're located uh in london still um so we're based just we're based outside of london um kind of between london and oxford i would say probably about an hour's um drive from from heathrow airport yeah and excuse me where is the bulk of your market i mean it's obviously a global market and as you said fulfillment centers and the logistics industry is exploding with the shift to online commerce and all of that.

35:51Where is your largest market? So the markets of focus for us at the moment are Europe and America. I would say the larger market is the American market. And we're seeing very, very strong interest and we're seeing some incredible customers and incredible partners in the market that have really embraced the technology and have achieved incredible return on investment and incredible benefits from it. So if you were to ask me the question and I'm looking at the global ecosystem, I would say the american market is one of the it's basically the largest for us but it's also one that's incredibly fast moving and fast growing and fast adopting of of the technology yeah you said a couple of times that the the scanning robot uh is is the tallest robot is that how you put it uh on the market how tall is it and and if it's uh you know warehouses can be pretty high uh inside so how does it scan you know 30 40 feet up so the height of the product the current um the current version we have in the market is 46 feet oh wow it's a ground-based robot so it's got a telescopic tower on it and it scans it scans in one swift pass so um that's that's again so like it goes back to what we were discussing is we've built this product by leveraging the market and having very deep conversation and a lot of product discovery with customers it's a ground-based robot it enables it to operate under any conditions day or night with people around it it doesn't really matter and it scans continuously so it it moves on the side of the rack it moves on the side of the rack and collects the data but it can also scan this information in block stack environments where you don't really have racking and it's exactly the same principle being able to to leverage this the fact that it's a ground-based robot is very very tall we can scan well over 10 000 pallets an hour so we've we've had location we've had we've had customers and locations where we scan 12 13 000 pallets an hour so a very important point for our customers is the speed of scanning but more important than that the speed of transforming that scan data that raw data into insights for them that's really what uh what drives uh operational efficiency and it's really what drives excellence so we can scan the warehouse but how fast can you actually tell me what i need to do in order to drive improvements in my day-to-day operation.

38:35Yeah, and I can see that after an initial scan, the creation of a digital twin and the optimization, maybe you want to move these palettes over here and re-up space there and put something, that kind of configuration. But that happens once every so often. It's not an ongoing thing. does the robot continue to scan on a real-time basis to track inventory or alert if something is being misplaced according to the optimization plan yeah absolutely so the short answer is yes actually it scans at least once a day sometimes multiple times a day depending on the operation And the big thing about the optimization is that you drive this, you basically run this continuously.

39:36Because today you're selling a certain type and amount of product. And then as, for example, as the week progresses, customers want something else that changes and evolves continuously. You don't drive changes in the warehouse. You don't move the racks around and so on. but you could continuously optimize the way you lay down the space and where you stored the products to make sure that you have the best use of that facility. So the robots would scan continuously, eliminate any kind of inventory errors, track this real-time availability of product for picking, for making sure that everything is best utilized, and provide this information in real time.

40:18so you see this information pretty much at every single point of time in the day or night index a review you don't have to wait to the end of the scan you can see as the as the robot moves through the warehouse you see this information being pushed in the digital twin yeah so where is this going you know you talked about how warehouse robots and dark warehouses and that sort of thing are going to be common in the future. For Dexery, where are you guys looking to go? Are you just in the phase now of building market share with your product, or are you doing research on new products coming out? So I would say we're continuously developing new functionality for our products.

41:17The way we look at products is by combining DexReview, the digital twin, as well as the robotic systems and the autonomous robots to collect the data and information. I would say we have a very rich roadmap for where we are looking to take the technology. And we touched on a couple of aspects around AI. we touched on a couple of aspects around the agentic side of things because the way we see it is being able to create an ecosystem where you can run any type of product any type of goods any type of warehouse layout with any type of resources you have available so regardless of the amount of space you have or however way you want to equip that facility being able to deploy decks review means you can run, you can have the certainty that you run that operation with maximum efficiency.

42:07And I think that's one of the big, big drivers that we want to bring into the industry is this transformation that you don't have to know if you have an idea or if you want to run a type of product. You don't have to have past experience. You can leverage the fact that this technology is so widely adopted and then have that certainty that the way you're running your operation is going to be maximum efficiency is going to bring you the maximum benefit for that setup. Yeah. You were talking about the data that you've collected or that you collect in these scans. Is there a central data lake where all that data goes that continues to optimize the algorithms or train the algorithms is, I remember a number of warehouse robotics companies I've spoken to, you know, like a picking robot, its learnings will be fed back to the central AI system.

43:16And then all of the robotic arms attached to that system are updated with that learning is is there that kind of uh uh you know fleet learning going on yeah absolutely i mean this this level of um of distributed intelligence is something that um we've is one of the the core principles on which we build the system because that benefits pretty much anyone that uses the the dexa review platform and dexa risk technology so it's learning about running more efficiently is learning about scanning faster but also the perception system being able to extract more information from the the scans and the data we collect is then something that benefits everyone that leverages this technology to be able to drive efficiency in their business so it's a very very important point and we're operating a lot of processing directly on the robots directly on the device because we scan a huge amount of data we scan hundreds of gigs an hour there's no real infrastructure to be able to transmit all that information so we process the data we extract the insights and then we keep only the valuable parts of that data set for for further learning and further improvement so andre is there anything else that you want uh the audience to hear uh i'm i'm excited about the prospect of a fully automated uh global logistics system uh how far away do you think that is as i said with minimal human workers in warehouses uh dexterity view sort of providing the data and then a fleet state of robotic arms or or mobile robots you know moving things around in the warehouse is that five years away ten years away that's a very very good question I think it's a lot closer than than we would envision you look at some of the latest developments in in the in the field of of robotics in the field of automation but also in the field of uh intelligence and data analytics and i think that that's shaping up to be a very exciting future um and and as i was saying earlier i think it's really about how do we create the perfect symbiosis between um between humans and the machines that can actually enable us to do our jobs better faster easier i think once we once we start making this a reality we're going to have a very robust and resilient global supply network i don't like to call it supply chain because i think it's um it's about breaking the chains about creating this interconnected networks it's about um driving to to get to this seamless visibility so you can optimize end-to-end so we can be efficient and we can actually be a lot better and in the way we're doing things in the way we're pretty much moving everything around the world yeah and and finally what what sort of how big is the market it seems like it's massive and growing and and how much of that market are you serving are you still at the very beginning uh uh you know is is there sort of endless growth ahead of you i would um i would like to believe there is endless growth ahead of us of course i don't like to uh to look at the world with a with a limited or fixed mindset um i mean if even if you if you were to limit this technology and limit the product and applicability of what we do um to the warehousing space you know this is a massively growing industry and it's it's um it's accelerating and you look at some of the trends that we're seeing now with uh bringing manufacturing at the forefront of what we do and so on you can't bring manufacturing um and make make a flagship industry out of it without having a very very strong supply network around it so when you look at the the wider industry there's probably two three hundred thousand warehouses um as we know in the world right now where we're definitely just getting started um but i think you know there's applicability for for Dexa review as a technology to be a household name in pretty much every single one of these locations this is what we're looking to build this is where we're going towards warehouse operating systems and a wider logistics operating system means the ecosystem will be what's going to be powering the industry for for the foreseeable future