When I was eight, I remember our train to Mumbai suddenly stopped with a loud crash. About eight hours later, when the train started moving again, someone pointed to a torn-off tail hanging from a branch near the tracks.
Apparently, the train had collided at full speed with a wild buffalo, severely damaging the engine and forcing it to be towed away for repairs. For a long time, I thought such incidents only occurred in places like rural India, but several years ago I realized that wildlife and infrastructure are incompatible anywhere on Earth.
The U.S. Federal Aviation Administration (FAA) estimates that wildlife strikes have cost U.S. civil aviation an average of $272 million annually between 1990 and 2025. In France, railway operator SNCF reports that wandering animals affect over 8,000 trains nationwide each year, ranking as one of its top five causes of delays. In Brittany alone, 120 trains collided with animals in 2023, disrupting 660 services and causing more than 15,000 minutes of delays.
After reading about the damage to infrastructure and wildlife, Sara Nozkova (left in the image above) was inspired to develop a wildlife deterrent system different from traditional methods such as alarms, lights, fences, falcons, and culling, designed to keep animals away from roads, railways, airports, and other infrastructure.
“I’m sure the answer isn’t a louder deterrent,” she told TechCrunch.
After 13 years working in transportation and autonomous driving, Nozkova has learned that these systems rapidly fail when nature enters the frameworks we’ve established for navigating roads and other controlled environments.
“We’ve taught machines to be very smart, but only within the environments we’ve designed for them,” she said. “Cities, production lines, warehouses—fixed geometries, known objects. Step into nature, and I’ve seen the same machines quickly go blind. Because nature is complex and always alive.”
She therefore joined the AI laboratory at the KTH Royal Institute of Technology in Stockholm to research a system that could adjust repellent responses based on the behavior—and even the species—of nearby animals.
This initiative later evolved into Flox Intelligence. The company was spun out from KTH in 2020 and was selected as one of the finalists for this year’s TechCrunch Disrupt Startup Battlefield 200. (The company pitched to a panel of venture capital judges at Disrupt on Tuesday; this year’s Disrupt is being held at Moscone West in San Francisco and runs through Thursday, October 15.)
Flox manufactures edge devices that process data locally rather than in the cloud. Like autonomous vehicles, these devices use perception systems to sense their surroundings, but they are trained to recognize wildlife behavior. Motion sensors wake the device when movement is detected, activating cameras that use local models to identify nearby animals. The device then emits species-specific sounds to deter them.
Nozkova said the system also learns from experience and adjusts its responses to prevent animals from becoming habituated to repeated signals.
The company currently sells battery-powered devices that can be mounted on utility poles, trees, and other structures. It also provides a software platform that shows customers where the devices are deployed and what is happening on-site. The platform can also generate reports for regulatory compliance and applications such as biodiversity monitoring. The company’s longer-term goal is to make the software portable enough to run on machines used in outdoor environments, including trains, cars, and drones.
Nozkova said that approximately 150 Flox devices are currently in operation across North America, Europe, and Australia, deployed at locations including airport runways and surrounding areas; highways and railway corridors—particularly at the ends of fences and intersections along wildlife corridors; national parks; agricultural lands; urban areas; mines; and critical infrastructure such as power substations.
The company initially sells edge hardware devices on a one-time basis, then charges an annual software and AI subscription fee starting in the first year. Flox also partners with several OEMs that pay annual fees to deploy its software on vehicles such as trains. Alstom, a train manufacturer, is one such partner.
Nozkova said the company expects revenue of approximately $1.5 million by the end of 2026.
Although Flox’s products and methods are distinctive, the startup is not without competition. Robin Radar’s Scarecrow also uses species-specific distress calls to deter birds and monitor activity; iCHASE sells a system that uses computer vision to detect birds and targets them with mobile lasers, adjusting its response as needed.
Nozkova said her moat lies in the breadth of the system and the data Flox collects on the ground, which she calls “the real evidence of what makes a moose or a grizzly bear move safely.”
“Physical AI is leaving factories and entering this chaotic, living outdoor world,” says Nozkova. “All machines operating outdoors—trains, trucks, drones, tractors, robots—need to understand nature, and natural intelligence is the next layer of the perception stack. Almost no one is doing this.”
To date, the company has raised approximately $5 million through pre-seed and seed rounds, with the seed round led by Unconventional Ventures and Tilia Impact Ventures. Nozkova said that GENIUS NY, Norrsken, Satgana, and angel investors also participated in these rounds.
The company is currently raising a $10 million Series A round, with the majority of funds intended to scale the number of deployed devices from approximately 150 to thousands, and to extend the underlying system beyond its current wildlife use cases.