I recall a vivid childhood memory from age eight, when our train to Mumbai abruptly stopped with a loud jolt. After an extensive eight-hour delay, someone observed a mangled tail unexpectedly hanging from a trackside branch as we resumed our journey.
It transpired that the train had collided with a wild bison at high speed, causing severe engine damage that necessitated towing for repairs. For many years, I believed such incidents were confined to regions like rural India, but I later discovered that conflicts between wildlife and human infrastructure are a global issue.
According to FAA estimates, wildlife encounters have incurred an average annual cost of approximately $272 million for U.S. civil aviation between 1990 and 2025. French railway operator SNCF reports that nomadic animals impact over 8,000 trains nationwide annually, ranking among the top five causes of operational delays. Specifically in Brittany, 120 trains struck animals in 2023 alone, leading to the disruption of 660 services and over 15,000 minutes of accumulated delays.
Witnessing the extensive damage inflicted upon both infrastructure and wildlife prompted Sara Nozkova (shown above on the left) to create a novel wildlife deterrence system. This system deviates from conventional methods like sirens, lights, fencing, trained falcons, and culling typically employed to deter animals from critical infrastructure such as roads, railways, and airports.
“I became certain that a more potent deterrent was not the solution,” she shared with TechCrunch.
Having dedicated 13 years to transportation and autonomous vehicle development, Nozkova observed how swiftly engineered systems designed for controlled environments, such as roads, can falter when natural elements are introduced.
“While we’ve enabled machines to excel, their brilliance is often confined to the environments we design for them,” she explained. “Think of cities, factory floors, or warehouses—places with static layouts and identifiable objects. However, venturing into natural settings, I witnessed these very machines rapidly lose their effectiveness. This is because nature is inherently intricate and dynamic.”
Consequently, she joined the AI lab at KTH Royal Institute of Technology in Stockholm, where she began developing a system capable of adapting its deterrence methods according to the behavior and species of nearby animals.
This research culminated in Flox Intelligence, a company that emerged from KTH in 2020 and has been selected as a finalist for this year’s Startup Battlefield 200 at TechCrunch Disrupt. (The firm presented its pitch to a group of VC judges on Tuesday at Disrupt, an event held at Moscone West in San Francisco until Thursday, October 15.)
Flox produces edge devices that conduct data processing locally, bypassing cloud-based operations. Similar to autonomous vehicles, these devices incorporate perception systems to monitor their environment, but are specifically trained to recognize wildlife behavior. Upon detecting movement via a motion sensor, the device activates a camera which utilizes onboard models to identify nearby animals. Subsequently, it emits a sound tailored to the specific species to deter them.
Nozkova states that the system also learns adaptively, modifying its responses to prevent animals from becoming accustomed to recurrent signals.
Currently, the company provides battery-powered devices suitable for installation on poles, trees, and other fixtures. Additionally, it offers a software platform that gives clients visibility into device deployment locations and real-time ground activity. This platform is also capable of generating reports for regulatory adherence and applications such as biodiversity tracking. The ambitious long-range objective is to develop the software to be sufficiently portable for deployment on various outdoor machinery, including trains, automobiles, and drones.
Nozkova reports that approximately 150 Flox devices are currently operational across North America, Europe, and Australia. These deployments span various locations, including airfields and airport boundaries; road and rail pathways (particularly at fence terminations and intersections within wildlife corridors); national parks; farming areas; metropolitan regions; mining operations; and vital infrastructure sites like electrical substations.
The company sells its edge units with an upfront payment, followed by an annual subscription fee for software and AI services after the initial year. Flox has also established OEM partnerships where companies, such as train manufacturer Alstom, pay an annual fee to integrate its software onto vehicles like trains.
Nozkova stated that the company anticipates achieving revenues of approximately $1.5 million by the close of 2026.
While Flox’s product and methodology are distinctive, the startup operates within a competitive landscape. For instance, Robin Radar’s Scarecrow employs species-specific distress calls to deter birds and logs their activities. Similarly, iCHASE offers a system that uses computer vision to detect birds and directs a moving laser at them, adapting its actions as necessary.
Nozkova asserts that the company’s competitive advantage lies in the comprehensive scope of its system and the extensive field data Flox has amassed, which she describes as “verified insights into what genuinely prompts a moose or a grizzly bear to relocate safely.”
“Physical AI is transitioning from structured factory environments to the intricate, dynamic outdoors,” Nozkova comments. “Every machine functioning externally—be it trains, trucks, drones, tractors, or robots—will require an understanding of nature. This ‘nature intelligence’ represents the forthcoming essential layer in the perception stack, and currently, very few are developing it.”
To date, the company has secured approximately $5 million through its pre-seed and seed funding rounds, with Unconventional Ventures and Tilia Impact Ventures spearheading the seed round. Nozkova indicated that co-investors participating in these rounds comprise GENIUS NY, Norrsken, Satgana, and various angel investors.
Flox Intelligence is currently initiating a $10 million Series A funding round. The primary allocation of these funds will be directed towards scaling the deployment of devices from approximately 150 to several thousand units, as well as expanding the core system’s capabilities beyond its existing applications in wildlife management.