Flox Intelligence Builds Edge AI to Scare Animals From Tracks

Flox Intelligence has built an edge-computing device that identifies wild animals near railways, roads, and airports and plays a species-specific sound to move them off the infrastructure, according to TechCrunch. The company is a finalist in the Startup Battlefield 200 competition at TechCrunch Disrupt, running October 13-15 at Moscone West in San Francisco.

How does the device work?

The system runs entirely on-site rather than sending data to the cloud. A motion sensor wakes the unit when it detects movement nearby, which triggers a camera. On-device models then classify the animal in frame, and the device emits a sound calibrated to that species to drive it away, TechCrunch reported. Founder Sara Nozkova told the outlet the goal was behavioral precision rather than volume: "I was convinced the answer was not a louder deterrent."

Why do existing deterrents fall short, according to Flox?

Railways, airports, and road operators have historically relied on sirens, lights, fences, trained falcons, and culling to keep animals off infrastructure. Nozkova spent 13 years in transportation and autonomous-vehicle work before founding Flox and said she saw how perception systems built for controlled environments — "cities, factory lines, warehouses" with "fixed geometry, known objects" — failed once deployed outdoors. "Stepping outside into nature, I saw firsthand how the same machines quickly go blind," she told TechCrunch, "because nature is complex and always alive."

How large is the wildlife-strike problem Flox is targeting?

TechCrunch cited two regulatory and operator figures to frame the scale. The Federal Aviation Administration estimates wildlife strikes have cost U.S. civil aviation roughly $272 million a year on average between 1990 and 2025. In France, national rail operator SNCF says wandering animals affect more than 8,000 trains annually and rank among its five leading causes of delays; in the Brittany region alone, 120 trains struck animals in 2023, disrupting 660 services and generating more than 15,000 minutes of delays, according to SNCF figures reported by TechCrunch.

Where did the technology originate?

Flox spun out of the AI lab at KTH Royal Institute of Technology in Stockholm in 2020, after Nozkova began developing a perception system trained specifically on wildlife behavior rather than the static object sets used in autonomous-vehicle navigation. The company's pitch distinguishes its approach from legacy deterrents by making the response conditional: the device adjusts output based on which species it detects and what that animal is doing, rather than broadcasting a single generic alarm regardless of context.

What happens next?

Flox's appearance at Startup Battlefield 200 puts the company in front of investors and infrastructure operators evaluating edge-AI deterrent systems as an alternative to fencing and culling programs. TechCrunch's account does not specify current deployment sites, unit pricing, or regulatory approvals for use near airports or rail corridors, and those details were not included in the source material reviewed for this report.

The original TechCrunch report, including Nozkova's full account of the system's design, is available here.

For a Bay Area-baked gift, Stirred, Not Shaken ships in the U.S.