Drones and AI against fire: PYR ONE aims to put out wildfires before they exist, but independent data is lacking

By Radio Agricultura · October 24, 2026.

Cristián Martínez, commercial director of Grupo Belator, presented the PYR ONE platform on the program El Agro TV. It combines drones and artificial intelligence to detect and contain fire outbreaks in their earliest stages, before they turn into a full-blown wildfire.

The key figure, according to Martínez himself, comes from a test: a drone detected an event 3.1 kilometers away in 48 seconds, with exact coordinates. Adding the containment drone's flight time and the drop, the entire process takes no more than 4 minutes. These are company figures; the article does not mention independent verification or the test conditions (wind, size of the outbreak, type of vegetation).

Belator's argument is about time. Martínez argues that typical response times are too long and that, “at 50 km/h”, the area burned in just a few minutes is enormous (the article does not clarify what that speed refers to). He defines a “critical window”: from the moment the outbreak starts until the fire's energy exceeds the capacity of basic means. According to him, by the time CONAF (Chile's National Forestry Corporation) and the fire department step in, the wildfire is already fully under way. PYR ONE aims to act just before that point. The executive also said that, even if they manage to bring the fire under control, they notify the fire department to watch for a possible flare-up.

Context: broadly speaking, the logic of early detection is sound. In a wildfire, every minute gained at the start is worth far more than the resources needed later. The combination of sensors, computer vision and autonomous drones fits a broader trend toward automating the surveillance of large territories, where a permanent human presence is unfeasible.

Our take. This is an AI application that deserves attention precisely because it is unglamorous: it does not promise superintelligence, but rather gaining minutes on a physical, measurable problem. If confirmed, it is the kind of use that creates clear social value and that, in the long run, brings us closer to a world where disasters are managed proactively rather than reactively.

But it is worth keeping our feet on the ground. A test at 3.1 km is a demonstration, not a real fire season. Key data are missing: false alarm rate, performance in smoke, wind or difficult terrain, battery endurance, a drone's water or retardant payload, and cost per protected hectare. A containment drone is unlikely to put out an outbreak that has already gained intensity; its usefulness lies precisely in that first phase, which makes detection reliability critical.

There is also a governance question: a system that decides to act on a fire autonomously must be integrated with official emergency services, with clear protocols on airspace and liability. The fact that Belator notifies the fire department after each intervention points in that direction, but it will be deployment at scale, not a demonstration on a TV program, that determines whether PYR ONE is a useful complementary tool or just another promise.

In the short term, the sensible approach is to see it as a complement for private properties and communities, not as a substitute for firefighting crews or traditional prevention. What will matter most in the coming months are open trials and field results.