Flox Intelligence is building AI to keep wildlife away from trains, airports, and roads

A startup born out of wildlife damage to trains and airports is betting the fix isn’t louder sirens or more fences. It’s AI that can tell what animal is nearby, judge what it’s doing, and change its response on the spot.

That’s the pitch behind Flox Intelligence, a spinout from KTH Royal Institute of Technology in Stockholm. Sara Nozkova, who spent 13 years in transportation and autonomous vehicles, said she came away convinced that traditional deterrents were too blunt. Her team built on-edge devices that wake up on motion, use a camera and on-device models to identify an animal, and then play a species-specific sound to push it away. The system is also meant to avoid the classic problem of animals getting used to the same trick over and over.

The company sells battery-powered hardware that can go on poles, trees, and similar structures, plus software that shows where the devices are deployed and what they’re seeing. That software also handles reporting for regulatory needs and biodiversity monitoring. Flox says the longer-term plan is to make that software portable enough to run on machines that work outdoors, including trains, cars, and drones.

So far, Nozkova says, about 150 devices are live across North America, Europe, and Australia. They’re being used at airfields and airport perimeters, rail and road corridors, national parks, farms, urban areas, mining sites, and electrical substations. The company charges upfront for the hardware, then an annual software and AI subscription after the first year. It also has OEM deals, including one with train maker Alstom, that let partners pay yearly to run Flox software on vehicles.

The numbers behind the problem are ugly. The FAA estimates wildlife strikes cost U.S. civil aviation about $272 million a year on average between 1990 and 2025. SNCF says wandering animals affect more than 8,000 trains in France each year and are among its five biggest causes of delays. Flox is trying to sell a more adaptive answer than scarecrows for infrastructure, and it’s doing it with data gathered in the field, not just in a lab. The company has raised about $5 million so far, is opening a $10 million Series A, and expects to end 2026 with about $1.5 million in revenue.