# AI-Powered Detection with Docked Drones in High-Stakes Environments - Event: NestGen ’25 - Published: March 7, 2025 - Length: 27:36 - Watch: https://nestgen.org/library/watch/IktZARz07J4/ > The transcript of a recorded NestGen session, for reading or for use with an AI tool. It comes from the video's > captions; automatic captions can mishear names and terms. ## Transcript **[0:20]** Hello and welcome everyone to another exciting session at Nesten 2025. I'm Pri, your session moderator, and today we'll be talking about drones that not only observe, but think before they act. Imagine a security system where threats don't escape detection, where an eye in the sky is more than just a camera. It's an always-on guardian powered by AI. That is the reality we will explore today with the brilliant speaker Juan Gorgulan, CEO of UAV LATAM. Huan has led the way in AI-powered detection and autonomous security, ensuring that high-risk environments remain safe without relying solely on the human eye. Welcome Juan. But before we get down to business, here are some basic rules to make the session smooth and interactive. Use the question and answer box to ask any questions. Send them at any time and we will respond towards the end of the session. The session is recorded, so if you blink and miss something, don't worry because you can watch it again later. The participation **[1:21]** makes it really fun. Tell us where you're joining from. Let us know if you have any questions for Huan and the UA Latam team and we'll look into them. And finally, publish your key findings on LinkedIn or Twitter with the hashtag NestGen. Who knows, you might even get a surprise. I would also like to thank our NestGen 2025 sponsors. They have truly helped make this event a success and we greatly appreciate it. So yes, Huan, I'll give you the floor. Welcome, Juan. Thank you so much. It is an honor for me to be here and share with you the opportunities we are seeing in the market when you combine these three important things now. You combine docking drone technology, you combine artificial intelligence, and you combine FlytBase software and now some FlytBase hardware as well, which I'll explain later. That's definitely what **[2:23]** I'll share with you. Next, please. And yes, I am following the previous rules. Well, I'm not going to advertise just to put this into perspective of who we are. We are a company that operates in Latin America. Our business model is drone as a service. For us, providing a solution is more important than just the drone itself. We focus on large-scale inspections carried out by large clients, and also something we are seeing that is very, very significant today, especially in the Latin American region , is the security or safety and defense market. As I mentioned, we operate in the Latin American region, from Mexico to Chile. In **[3:24]** each of these countries we have local offices, we have local talent, we have pilots, and we also have the legal entity. But as you can imagine, and if you were to ask any drone company in the world, what are the main challenges? And this also applies to us; it's regulation. We have very close contact with the regulatory agents in each of the countries in which we were ordered to operate, because, generally, as entrepreneurs, as a company, we would like to go 100 miles per hour, but the regulator tells you: "Slow down, do it, make sure, go 20 miles per hour." That's the battle we have. We all want to go to 100, the regulators want to go to 20. So , we make concessions on some of those things, but we respect all the rules that each **[4:30]** company proposes. And, unfortunately, as I said, the difference between Europe and North America is that each of the Latin American countries wants to have its own regulations. That makes things a little more difficult, and to conclude this presentation, I believe that what you can see on the screen is something that gives us the credentials to be a major operator in the region. Well, without further ado, I will try to explain two success stories that I will share with you. The first one is about mining and the second one is about electric companies. Regarding mining, we will focus on a large mining industry. In Latin America, copper is a large mining industry and, generally, the entire copper industry has what are called open-pit mines. In other words, their modus operandi is **[5:35]** to carry out two or three explosions per week, and this is their main operation. And then, of course, blasting safety is very important, because there are between three and four thousand people in each of the mines . So it's not just two or three people, it's a lot of people moving around. Vehicles, trucks, people, and lots of people around. The client said that one of his main challenges is that he does n't want any accidents involving people when we do the blasting. Could you help us? And then we said yes, we could do that. So next. So we addressed the three main concerns this client had. First, anger and security, real-time monitoring and **[6:35]** improvement of Homer's operations . I'll explain it to you. You can read it here on the screen. But I'll explain what happened in real time . The customer call said : "A1, I would like you to be at the top of the blast five minutes before the blast, and I would like you to , uh, have a real-time feed to our commander's control center. And with artificial intelligence, please check that there is nobody within two or three kilometers , and please do that with artificial intelligence." So that's what we do. We asked our partners at FlytBase to provide that technology. We used a document in that case because it was some time ago. Do we use a dog? First, we used a mattress, uh, 30t, and provided the solution, and it was very successful. But you know what? We were very happy, and then the client said, "Hey Juan, you didn't understand, so what?" And he said, "Yes , what happened is that I don't want **[7:38]** anyone around." So that's where the solution with the dog and the FlytBase came from, and the client was very happy with the solution. Next, I can give all the technicians the map of how we did it. As you can see, starting from the left, we have the docking stations; the docking stations are connected to the drone. The drone has the map of all the blessing areas and, of course, as you know, all the mines in the middle of nowhere. So we have Ellen Musk to thank. We may not like everything Ellen Musk does, but Starink is definitely something we have to thank Ellen Musk for. We have this star link. So we can operate anywhere in the world with this solution and we provide a feed, and you can see in the image on the right that with deep learning and machine learning we provide all these AI tools so that people were very, very happy. Next, we are very pleased with the solutions we offer and, **[8:44]** again, I will present it later, but this is how we do it first. And this is something that, however sophisticated and interesting all the tools may be, you first have to understand the client. This is where it all begins, and then you start thinking, "Okay, this is what the client needs; how are we going to personalize the services?" It can be with equipment and software that we have, but it can include software and equipment that we don't have. That's why we have to customize the solution, and then we collect the data, process it, and this is where all the tools we provide for everything come in, for all the training of the AI ​​ model, and then we provide all the solutions to people so they can take some action. Here's how we define the problem with a client. What is the purpose of this task? Perform the detection correctly, and then what is the objective of **[9:51]** detecting people, vehicles, etc. And then we apply this in this environment. You can read it here. But the main message here, as highlighted, is that, as I mentioned, even with all the sophisticated tools at our disposal, we must start with a customer. This is the starting point, and then, once you've defined with a client, you work backwards. You use your tools, you use someone else's tool, but as long as you set the solution based on the client. Below are the first stories of Susi that take place in the mining industry. The client is very happy and we expect a sort of rush order for several of those mines around the world. And then, the second one is also from a major state-owned electricity distribution company. As you may know, these companies have **[10:55]** several hundred substations across the country. Some of them are in the city, others in remote areas. But basically, the next scenario is that most of these substations are already connected to Scattera. So, all the instruments are connected to the scatter, but these people from the commander's office, who agree that they have them, said they would also like to have an image because the temperature is important to them. So, again, this is the scheme we have in this environment. In this case, we are now using the doc two and then we have the signal for the drone and then we use the camera, and then with the camera we also make the flights with the standard RGB. And for them too, temperature is very important because, as you know, in substations the temperature of the cable insulators **[12:02]** is very important. That's why we do all these flights using this special camera, and on the right is what you see, the solution. So, in that case, the next client was very happy with the solution, and again, that's how we define what we're going to do, the spot inspection, what the objective is, and the applicability . This is how we get these two very good success stories. Then, and this is what really excites us, we see that the people at the FlytBase are one step ahead of us. They've thought about it a lot. Okay, these guys are doing the work, but what's the main challenge? And I can tell you this as someone who started working on the Internet with 300- euro models, because now younger people think **[13:04]** that gigabyte bandwidth is not enough. But in all cases, bandwidth is always scarce. So, in this case, with this, the people at Fly Bay said, "Okay, guys, you've got the dog in C2." But normally, when you send the drone's information in real time to the command center or the cloud, you're sending all the raw data across the entire bandwidth, and as you can see, as you can know, with these 4K or 8K cameras, because in one hour you can send a terabyte. You produce a terabyte of information in one hour. That's a lot of information to send from one workstation, even for a gigabyte. That's a lot of information. So these guys designed a small computer box that uses an Nvidia, a very high-end Nvidia, with which you can locally process your AI model. That way, whatever you send back to your commander's **[14:10]** control center , only the essence remains. You don't send the bones, you don't send the grid, you only send the essence. And this is very important when you do that implementation. So this is something that I really think is a game-changer for those of us who work in this field, in this arena. Next, please. And then again, you can read it here, but the main message is simply do it and do it now. You might not do it perfectly, but it's time to implement it. And these are the scenarios in which you can apply this flight-based AI technology to launch services. You know, you can read it here. But basically, if you ask me what the simplest goals of all this are, I'd say that security is the simplest goal. Our **[15:10]** good friends at FlytBase are giving us some great tools to address this issue, including maps, inspections, and emergency procedures. And I can also share with you, for example, if in both scenarios we are installing UPS, because even if you don't have electricity on the dock , with the installation, if you install UPS for the dog and Internet, you have video 24 hours a day, 7 days a week. For example, in a substation, you might have a problem with a fire. Hopefully it won't happen, but sometimes it does with a fire and you can deploy your drone because it will be powered by the UPS. These are the types of examples we improve when we create these models. Red, please. And we're ending up with that again. Defending security is the simplest objective in the market. However, **[16:16]** that doesn't mean that all other sectors are important. Yes, they are important and generate significant revenue. But my advice would be to focus on security and the defense network, please. And then, of course, there are some of the more obvious decisions about cost reduction and time efficiency. But for these high-tech environments, the last two are very, very important. Security is fundamental and, of course, scalability, and this is where FlytBase plays a significant role. You might think most people can handle one or two dogs, but when you're implementing, say in the case of an electric distribution company , they're going to have a hundred dogs moving the software. So, how are you going to handle this? This is where our friend from FlytBase makes **[17:20]** the software very easy to use and manage. Next, please. And to conclude, this is something that some of our people wrote. And again, this is actually the most important part of this whole quote you can read; it's the last word now. You have to do it now because this is the great opportunity we have . Following. And that's what I hope to share with you, what we did, and do n't hesitate to talk to me, talk to the people at FlytBase. We have an important partnership with them and we are on our way to conquering all of Latin America with their help. Thank you very much and I'll be happy to answer your questions, and please bear in mind that I know enough to be dangerous, but not enough to be **[18:21]** competent. So don't be too tough with the questions. Forward. Very good, Juan, I think it has been a great and fantastic conversation. Thank you for explaining the science behind AI-powered detection . And for those who want to know more about how UAB Latam can help them, go to their website, talk to the team, and yes, someone will get in touch with you. And yes, Juan, I think we can answer some questions because I've received some in the questions and answers tab. So let me go ahead and make one. How has UAV LATAM developed ROS 2 technology? In that case , could you connect with FlytBase? Could you repeat the question? Has UAV LATAM developed ROS 2 technology? In that case, could you connect with FlytBase? Yes, definitely. The way we connect with FlytBase is through FlytBase software. So, **[19:22]** they provide the APIs, and then you take your software gurus and then use the APIs again based on the client's needs. So, you start with a customer need somewhere on the business or marketing side, translate it into simple terms for the technical staff to understand, and then connect that with FlytBase. Okay. The next question is: are there specific industries or sectors where you see the safety of drones with AI technology having the greatest impact? As I mentioned, security is and extends horizontally to all sectors. So again, industry, mining, for example: in each of these sectors, there is always a big problem. I can tell you that in agriculture, especially, **[20:24]** people steal because there are huge fields; They steal the product, they steal the fruit, they steal, you know. And then, in mining, you have, unfortunately, illegal mining. So, with this tool, you can provide evidence of these security problems you have . And as I mentioned, in electricity distribution and telecommunications, and I could go on and on, but the whole application is related to security. And of course, the defense also comes into play because I can tell you, you know, something that will perhaps be for the next presentation (I'm making my invitation now), but with the experience we have in penitentiaries, because bad people, they are bad people, but they are very intelligent and they have money and they are also using drones to go to places and **[21:27]** be in places or transport themselves to places they should n't be. So that's my answer to that question. Okay. One more. How did you manage to accelerate the adoption of the documents with your client in the mining industry? Well, basically that's what they mentioned because I'm not saying anything new, but if you go and ask a big mining company, I hope they do n't tell you that the union people are threatening me. But if you ask any large mining company in the world where they will be in five years' time, you will see in their green book that they plan to have no one in mining. The goal is to have no one in mining. They will control all mining remotely from a very nice building in the city center. This is where all this automation plays an important **[22:34]** role. So I think when you go into mining, it shouldn't be a problem for them to understand that with this technology you don't need people in the field to manage these drones. It's a bit difficult, but hey, technology is advancing. And yes, some jobs will be eliminated, but others will be created. Let's say they're going to eliminate some of these pilots, but those smart guys are going to start working on AI and similar things. So yes, it's going to be an evolution of this, but it's going to be very easy for some people in mining to convince them about these canine solutions. And of course, they're going to say, "Hey, Juan, or a pre-, how are we going to handle these 10 or 20 dogs?" Well, that's what they say: "Hey, call us at pre or call Juan. We have a solution **[23:35]** for you." Yes, great answer, Juan. I think the most important thing about that answer was that we won't have people at the mining sites and that we'll prefer to have drones . So it was something very powerful. One more question, J Huan. Someone asks if you are currently limited by the range, endurance, and payload of the current generation of drones, for example, to protect large-scale facilities. Yes. Yes. Of course, it's the same problem we have when we have a phone that only lasts until 6 pm. Clear. For some reason, battery technology isn't keeping up with us . But on the other hand, this solution with dogs is not what we are talking about; most of these solutions are not half a million dollar systems, but something around 30,000 or 40,000 and something, and you can afford to install two or three for this type of environment. I mean, even 30,000 or 40,000 is a lot of money, but **[24:41]** compared to the security, if you compare it to the installation of five or ten people or the installation of all these cameras, the trade-off is significant. So you can install several bases to compensate for this lack of resistance we have today until someone invents this battery that everyone is waiting for and that will allow our phones and drones to last longer . Okay, thanks. One last question. If anyone has something important to say , this is the right time. Okay. What are the misconceptions that companies have when they first consider AI-powered drone security ? Well, as I said, we'll address that one. Hey, but I've got my people out on the field, you know, hey, we've got this watchman , you know, doing that. But if you go, and I can also tell you that it will be for most of the people in our audience, **[25:47]** drunk companies, not union members, but if you talk to security companies, they will tell you: "Hey, my clients tell me they don't need more people, they need more technology." This is where I would give my advice (and I'm not going to share my grandmother's recipe here), but knock on the doors of these security companies, I.e., Pro G4S, the big ones, and tell them: "Hey, I have this solution." You will be very welcome because you are providing technology rather than just people with all the limitations we know about people. Well, but don't get me wrong, these people are going to have to change and become employees of FlytBase or similar companies so they can be retrained. Definitely. Yes, I think so... Thank you, Juan. The answers were great and, once **[26:48]** again, it was an amazing conversation. I'm sure everyone... You know, we'll come away from this session with some important key takeaways, and I think that's all from us. Thanks to everyone who joined the session. Thank you for sending your questions. I hope we have answered all of them, and if not, please contact UV LAM. We will have... We will have someone who will respond on their end as well. And until then, keep innovating, keep flying, and we'll see you in the next session. Thank you. Well, it's a pleasure. A pleasure. Thank you. Thank you.