# Building Argentina's Largest Oilfield Drone Program Across Vaca Muerta - Event: NestGen ’26 - Published: October 2, 2026 - Length: 24:39 - Watch: https://nestgen.org/library/watch/pXzGZjRqR2M/ > 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. ## About this session Oil field operators scaling across Vaca Muerta, one of the world's largest shale formations, need a way to extend safety and security monitoring, from methane detection to intrusion response, across a rapidly growing footprint without adding headcount at every well pad. In this NestGen session, 3K SAS will share how they built Argentina's largest oilfield drone program across Vaca Muerta, connecting real-time methane and intrusion detection through autonomous docks, and what energy teams should think through when scaling similar programs across expanding oil field operations. ## Transcript **[0:00]** Well, thank you very much Dick for the introduction, welcome everyone. We are proud to be here to tell our story. This is a bit of a story about how we went from the office to the field and then from the field back to the office again. We are [clears throat] a company from Neuquén in the Republic of Argentina. We operate throughout Argentina. We are Santiago and Sergio. And we're going to tell you where we come from, what we're doing, and what we're going to do. As I was saying, we are from Neuquén. Aerotech was founded in 2013, conducting photogrammetric surveys using drones when there was still no **[1:05]** sustained development of drones in Argentina. In fact, we brought the first fixed-wing drone to Argentina, reducing the costs of field survey work because, well, until that time it was done with satellites, which was very expensive, and also with people in the field. The areas in which we work are primarily gas and oil, but also mining and environmental management. The area where we are, next please, Dick. Where we operate next again. It's Vaca Vaca Muerta. A region in the province of Neuquén, the next slide, please, Dick. Okay. Uh, it's a very large region, **[2:09]** uh, compared to, for example, New York City, it 's 40 times larger, compared to Buenos Aires it's 150 times larger. It is a region that has its complexities due to the distances, clearly, but also due to the type of terrain we have there. And so, uh, for us to operate there, uh, in those deposits that exist in that area, is a very, very big challenge. The image shows representations of New York City and Buenos Aires. Obviously, they're not in that place, but it's a way of representing its large size . We, uh, the path we travel is from the first drones that existed here in the region, uh, to the present, uh, when we are operating fully autonomous drones in a territory as large as that. But Santi is **[3:16]** going to tell you now, how we do it. Well. Good afternoon, thank you all for being here. Thank you very much, Dick, thank you very much, Sergio. Um, I'm going a little more into the topic of in the industry, traditional inspection relies on something called routers. Uh, what are the routers? These are teams that travel throughout the entire oil fields every day, looking for things like spills, equipment failures, anything new they see, they send it by email, by WhatsApp, they communicate with an engineer in the offices. Many times the engineers or the area manager end up going to see the assets with their own eyes. For example, in our experience it's 100 km there, 100 km back. As Sergio says, the Vaca Muerta area is a very large site, with many assets, which must be kept under constant monitoring to know what is happening in the field. This **[4:16]** is what the tour operators do, or what the tour operator of 2014 did, since we were in the industry, until now. Well, there are critical events that I can't give much detail about, but we need to be there to see it. Streaming live in under 5 minutes, which is our general experience. That's impossible by way of way, so the inspection frequency was defined by the people and vehicles available, it wasn't what the asset needed, for example. Well, there in the slide we see a typical animal of Argentina that goes back to the years uh in the years 2014-2015, we had a contract in which uh when we were just starting we had to count by sight how many guanacos, which are a typical animal of Argentina, how many guanacos there were in the field. And we counted 10,000. So that's the beginning of technology. Next. Please. **[5:17]** The first change to the field tour was our field staff using a drone. It was a van, two people and a drone, but it was still the same 60 to 100 km from Neuquén, where our offices are, to Vaca Muerta. That changed; it was arriving, flying, returning, and starting another stage that, well, depending on the circumstances, depending on the survey, required processing, and the information, being able to send the information, sometimes took days to reach the client or the engineer. So the data was better than what the runner sees at the start. But it was still the same trip, the same crew, and the same information, though not the same, but it arrived several days later. Next. Here is an example capture; it is the same thermal location and color on the same flight. And then **[6:19]** another different problem appeared, which is that each mission has to be designed, named, programmed and maintained. When we talk about thousands and thousands of assets. Em. It's complicated to get to the point where you have to go to each asset, survey it, and send the information. We are still dealing with the same problem. There we understand that replacing the truck with a drone did not create autonomy, it was a bottleneck that was moved. From flying to planning. And that [clears throat] is what we had to figure out now. We solved it in our next phase, which is next, slide please. That's the next step. They were remotely operated autonomous drones. And for that, the drone has to stay in the field; we call that a dock station. The team you see in the slide is our own team on-site on the day of an installation. And. To explain a little, installing a dock is like a small construction project; it **[7:20]** needs to have its base, it needs to have its anchor. Energy, connectivity. They have docks that have RTK so the drone knows where it is. And this also entails security, putting a security fence around the dock. Em. Having cameras is great, the dock has its own camera. And it also involves the launch of the first safety flight. On the day of installation, a safety flight is also carried out to ensure that everything is working and that the system can be placed in its final location. The only time we need to see the box again is during maintenance. Now we can leave the team on the field . There are two types of deployment. Where there are facilities nearby, the device connects to an electrical network and to fiber optics or 4G, for example. Where there is nothing, what has happened to us. Hey, the doc is using a satellite link. It can be fitted with solar panels and a backup generator. **[8:26]** And weeks go by again without anyone visiting him. It only needs to be visited for maintenance. Where each document goes ends up being a design decision. From the land and especially from the energy available at that location. The assets within that coverage radius also depend on that location . We work within a coverage radius of 6 km. Depending on the coverage ranges we have, that's where we'll place a doc station. Next please. With the docs installed, the model changes in nature; it is no longer just a drone. It ends up being a remote operations layer. Our pilots don't fly by hand. They supervise the operations. An operator, as you can see in the image, with two monitors, map, live video, telemetry, alerts. Our operators have several documents in their charge. Uh, we usually have one **[9:29]** operator and five docs, five drones in charge. When there is an incident, when there is an event, when there is some kind of location that we have to see, we have to arrive in between 5 and 30 minutes from when the drone leaves until it arrives at the place, 30 minutes because it ends up being a loading time that the equipment may have. Well , with that video we could already work with the information and send it to you live, not just post-processed, sending it after download. In live video, we could work with, for example, an area manager, an engineer, someone who needs to see a spill, for example, who can't wait until the next day to see it. Next slide. Now the key question would be, how do we tell the drone what it has to do? Because a pilot with five drones cannot fly by hand. We work with thousands of wells in **[10:33]** hundreds of locations, so more roads, for example, more power lines, batteries. Recording all those missions would have taken us months of work because there are so many assets. It was a long road, we started at the beginning flying by hand, then missions created one by one with the docs. The first lever, so to speak, was to reuse and apply stacking those missions. After that there was a recording stage, but generating the missions was effective in generating them in a way that was from a georeferenced database. All the wells, roads, power lines, batteries, all the information we can obtain from various resources, because roads, for example, can be obtained from a map. That giant database tells us exactly where each mission needs to be pointed in three **[11:35]** dimensions. Today, two artificial intelligence agents work together on that giant database. The mission agent reads the coordinates, calculates how the drone has to look at the asset, the geometry, the height, the camera sections, and generates the missions en masse, according to those we define, of course, for each asset. And we have another agent who plans the day's operations. What task does each pilot perform, with which document, and in what order? The pilots receive the plan from one and execute the missions from the other, speaking in terms of agent one and agent two. The system is not perfect; the database doesn't know what's in the location. For example, if a tower appears, an asset appears, something appears in the location that the database cannot recognize. That's what you see in the field, that's why the pilots have to stay active, they **[12:37]** have to keep watching the screens and if they see anything in the field, it gives an alert inside our system and the agents change the missions to a safe altitude, for example. The person who created the missions no longer exists there, they left our workflow and now there are review processes. And that's what makes it possible for a pilot to operate five drones at once. Okay, next, please. With the same infrastructure, we solve different needs. For wells, we patrol the wellhead; for thermal, the same location or the same asset in infrared; for power lines, each pole; for roads and spills, if there is an event, we have to be on site, etc., etc., etc. It's the same document, but it's the same room, it's the same mission logic, what changes is what we do with the data and where that information goes. **[13:39]** Hey Sergio, could you explain a little more about analytics? Next steps. Yes, next, please. Good. So, what Santi said, basically what we do to achieve these results. In Argentina we like to compare everything to football, to football, because we are very fanatical. So, we chose this example to talk about how far a drone is capable of flying and surveying . We chose a soccer field. A drone, as Santiago said a little while ago, has a coverage radius of 6 km. That represents 15,800 football fields, which, in the image we are seeing right now , is practically one pixel. That 's a lot of football field that the drone can survey. **[14:44]** Why is this important? Well, because what used to take us 60, 100, 200 km of human travel in a vehicle, today only takes us 5 minutes of time. So, to get from the office to the field, all we have to do is press a button. I'm putting it simply, but there's a lot of complexity behind it. There's a team of people working, but not working much more efficiently than at the beginning. So, next, please, Dip. Um, where before we had two people and a vehicle to go to the field and relieve with the drone, uh, and do some work in the field, today we have, for each person, for each pilot who works, at least, uh, five drones flying. That can be done at the same time. With a handful of **[15:46]** these minimal teams, we can survey, for example, 1,000 or more locations, which in this case could be any type of location, in any field. Um, next, please, uh, Dip. Today, um, we define the frequency with which we visit an asset, um, it is not defined by the availability of the crew or the routers that Santiago mentioned earlier, um, but rather it is defined by the need of the asset. The asset also defines how many times we repeat the same action, what response we give to an incident, and with whom we share the data. In other words, if someone, uh, on a company team, needs to see the work we're doing, well, we can share it and they **[16:50]** can look at it. Next, please, Dip. Um, this is an ecosystem. In this ecosystem, in addition to the people who work to, uh, make it work, uh, connectivity, the weather, the geometry of the missions, how we name and manage our databases and the trust that we and artificial intelligence can give uh play a very important role uh Next please, Diego. In addition to the static drones that can be used in the field and the way we are currently defining missions, we have a challenge which is to further improve the way we work. **[17:53]** What does this mean? Well, if an asset needs us to inspect it once in a while, that is , once a week or once a month, we can equip ourselves with a mobile unit on a vehicle, as shown in the photo, and go out into the field to take off from wherever we want. We still have the ability to manage it from an office, to be far from the field and to share the videos and everything we want to see from the office. The truck just goes along and the drone works all its magic. And at the same time, the next step is not to think about the missions ourselves, about what we have to go and visit, but to let the asset think for itself. What does the asset need? What do we need? **[18:58]** So, the asset decides it needs to be inspected? So, uh, that's the next step. Improve the complexity of the system so that it is self-sufficient in some way. That has different layers of action, which are, for example, artificial intelligence, which we are not dealing with in this conference, but which is something that we will have in mind in the future. It's mobility upon mobility, intelligence upon intelligence, and all focused on the asset. Well, with this we conclude our talk. We are happy to answer any questions you may have, and we greatly appreciate your presence. **[20:55]** Um, the operational infrastructure works autonomously, uh the operations are automatic, always supervised with a pilot in charge and we are enabled and authorized by uh ANAC, ah so yes, it does allow it. I get the impression that it has Em, well, in this case I don't know about other sites in Latin America, uh Argentina particularly uh we have a lot of contact with ANAC to be able to use these operations, to be able to set up these operations. I personally believe that other places in Latin America can follow suit, **[21:58]** but I couldn't be that specific. I think that's what it means. **[23:01]** I can answer it in Spanish, even though it 's in English. Do you want to answer it in English? Do you want to answer it? Yes. In Spanish. Okay, let me see if I understand the question correctly. Um, I think it refers to a question of, uh, " Where does the information go?" Safety, right? Or if you go to a head office. That depends on the platform that each company decides to use. In our case, we have platforms that are certified and capable of storing data in secure data centers prepared for this purpose, because all the data we work with is extremely confidential and cannot withstand any other type of processing. Uh, the place is the cloud **[24:04]** in general terms, but I can't reveal more than that. Thank you so much. Bye bye.