The cost of a collision
Wildlife strikes cost US civil aviation an estimated average of about $272 million a year from 1990 to 2025, according to the Federal Aviation Administration. French rail operator SNCF says animals affect more than 8,000 trains annually and rank among the five leading causes of delays. In Brittany, 120 animal collisions in 2023 disrupted 660 trains and caused more than 15,000 minutes of delays.
Flox founder Sara Nozkova told TechCrunch she did not think the answer was to make deterrent sounds louder. The company instead tries to change the signal according to the animal and its behavior.
Before Flox, Nozkova spent 13 years working in transportation and autonomous vehicles. She saw systems built for controlled settings struggle when conditions became natural and changeable. That led her to join the AI lab at KTH Royal Institute of Technology in Stockholm and work on a system that responds to nearby animals.
Flox spun out of KTH in 2020 and became a Startup Battlefield 200 finalist at TechCrunch Disrupt. This year’s conference is scheduled for October 13–15 at Moscone West in San Francisco.
Detection, then deterrence
Flox’s battery-powered devices process data locally rather than in the cloud. A motion sensor wakes a device and activates its camera; on-device models identify nearby animals, then select a sound intended to deter that species. Nozkova says the system also learns from experience and changes its signals so animals do not become accustomed to repeated sounds.
Customers can track device locations and activity through Flox’s software platform, which also generates reports for regulatory compliance and other uses, including biodiversity monitoring. The company sells the hardware upfront, then charges an annual fee for software and AI after the first year. Equipment makers pay an annual fee to install the software on their machines; train manufacturer Alstom is among Flox’s partners.
The devices are used in several kinds of settings:
Nozkova says about 150 devices are deployed across North America, Europe and Australia. Flox expects to finish 2026 with about $1.5 million in revenue.
A wider ambition, with a proof gap
Flox is not alone in animal deterrence. Robin Radar’s Scarecrow uses species-specific alarm calls to disperse birds and records what happens. iCHASE uses computer vision to locate birds and direct a moving laser at them, adjusting its response when needed. Nozkova says Flox’s advantage is its broader coverage and the field data it has accumulated on what safely moves animals such as moose and grizzly bears.
The larger pitch is that physical AI must work beyond factories, in outdoor environments that keep changing. Nozkova argues that trains, trucks, drones, tractors and outdoor robots will need to understand nature, and that few companies are focused on this layer of perception.
I think the expansion plan is more ambitious than the current deployment numbers can yet validate. Flox has raised about $5 million across pre-seed and seed rounds, with the seed round led by Unconventional Ventures and Tilia Impact Ventures; GENIUS NY, Norrsken, Satgana and angel investors also participated across rounds. It is now raising a $10 million Series A, most of which it plans to use to grow from about 150 devices to several thousand and develop applications beyond wildlife protection.
What I’d want to know is how reliably the system changes animal behavior across sites and species, and how much of that performance can be transferred to machines operating in the field. The funding plan assumes Flox can scale both its hardware footprint and its software’s scope; the harder test is whether the data gathered in one setting makes the next one safer.
Source: techcrunch.com
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X