Flox Intelligence is using AI to keep wildlife off tracks, roads and runways

Flox Intelligence is developing edge AI that can detect wildlife near trains, airports and roads and respond with species-specific deterrents before animals cause damage, delays or fatalities. The Stockholm startup, a finalist in TechCrunch Disrupt’s Startup Battlefield 200, says its system is already deployed in about 150 locations across North America, Europe and Australia and is now preparing a $10 million Series A to scale further.

The pitch is simple but ambitious: instead of relying on static fences, lights, sirens or culling, Flox wants infrastructure to recognize what kind of animal is nearby, understand how it is behaving, and choose a response that is more likely to move it away safely.

For founder and chief executive Sara Nozkova, the idea grew out of a long career in transport and autonomous systems, and out of the realization that the built world still struggles when it meets the natural one. The problem is not niche. Wildlife collisions and intrusions cost airlines, railways and road operators billions in damage, delays and repairs, while also putting animals at risk.

Why wildlife and infrastructure keep colliding

Wild animals do not read timetables, and infrastructure rarely adapts quickly enough to changing conditions. That mismatch is expensive in aviation, disruptive on rail networks and dangerous on roads that cut through migration routes or wildlife corridors.

In the United States, the Federal Aviation Administration estimates that wildlife strikes have cost civil aviation an average of roughly $272 million a year over the long span from 1990 to 2025. In France, SNCF says animals interfere with more than 8,000 trains annually and are among the five biggest causes of delays. In Brittany alone, rail operators recorded 120 animal strikes in 2023, disrupting 660 services and causing more than 15,000 minutes of delays.

Those figures only cover a slice of the broader problem. Similar incidents can affect mine sites, electrical substations, agricultural land, airports and rail crossings anywhere animals and heavy equipment share space.

“I was convinced the answer was not a louder deterrent,” Nozkova said, arguing that the sector needed smarter responses rather than more noise.

Her view reflects a wider shift in industrial AI: the most useful systems are increasingly those that can interpret messy real-world environments, not just controlled factory floors. In Flox’s case, that means teaching machines to distinguish between species, assess behavior and decide how to react in a way that is more effective than a one-size-fits-all alarm.

What Flox Intelligence actually does

Flox builds battery-powered edge devices that are mounted on poles, trees and other structures near problem areas. When a motion sensor wakes the unit, a camera captures the scene and on-device AI models determine whether an animal is present and which species it appears to be.

If the system identifies a target, it emits a deterrent sound designed for that species. The company says the device also varies its behavior over time so animals do not simply get used to the same signal and ignore it.

The edge-first design matters. Rather than sending everything to the cloud, Flox processes most of the data on-site, which can reduce latency, lower connectivity requirements and make the system more practical in remote locations where reliable broadband may not be available.

How does Flox avoid habituation?

It does so by changing the way it responds. Flox says the platform is designed to learn from field experience and adjust its deterrent patterns so wildlife does not adapt too quickly to repeated sounds or identical cues.

That approach is important because many traditional deterrents fail over time. Animals can become accustomed to sirens, flashing lights or repetitive noises, which reduces their effectiveness and can force operators to escalate to more intrusive or harmful measures.

Nozkova’s background helps explain the product choice. After more than a decade in transportation and autonomous vehicles, she saw firsthand how systems built for roads, rails and warehouses could be highly reliable inside engineered spaces and far less effective once nature entered the equation.

Nozkova said the lesson from transportation and autonomy was that technology can be highly capable within fixed, human-designed environments, but can quickly lose its advantage when it encounters the complexity of the outdoors.

How big is Flox’s deployment today?

Flox says about 150 of its devices are live today across North America, Europe and Australia. The units have been placed in a range of settings where animal movement creates risk or disruption, including airfields, airport perimeters, road and rail corridors, fence ends, wildlife crossings, national parks, agricultural sites, urban areas, mines and critical infrastructure such as electrical substations.

That spread matters because the company is not trying to solve a single-use problem. Instead, it is aiming at a common infrastructure challenge: keeping animals away without compromising safety, operations or the animals themselves.

The startup also offers software that shows customers where devices are installed, what has been detected on the ground and what response each unit has triggered. The platform can generate compliance reports and support biodiversity monitoring, giving operators a view beyond simple deterrence.

Longer term, Flox wants the software stack to work not just on fixed devices but on mobile equipment as well, including trains, cars and drones. That would expand the company from a wildlife-deterrent product into a broader “nature intelligence” layer for outdoor machines.

Inside Flox’s business model

Flox sells the hardware up front and then charges a recurring annual fee for software and AI services after the first year. The company also works with original equipment manufacturers that pay to embed its software in vehicles and rail systems. One example is train maker Alstom, which is listed by the startup as a partner.

The model is designed to create a hardware-plus-subscription business rather than a one-time product sale. That gives Flox a path to recurring revenue while also embedding its technology into the systems operators already use.

Nozkova said the company expects to end 2026 with about $1.5 million in revenue. For a deep-tech startup in a specialised niche, that is modest but notable, especially given the complexity of real-world deployments and the need to prove reliability in varied geographies and species environments.

What is the company raising now?

Flox is opening a $10 million Series A round. The money is expected to fund a move from roughly 150 deployed devices to several thousand, while also extending the core system beyond its current wildlife use cases.

The company has so far raised about $5 million across pre-seed and seed financings. Its seed round was led by Unconventional Ventures and Tilia Impact Ventures, with participation from GENIUS NY, Norrsken, Satgana and angel investors.

The fundraising push comes at a moment when more investors are looking for practical AI companies with clear operational value and real-world data advantages. Flox fits that mold: it combines computer vision, embedded hardware, environmental sensing and recurring software revenue around a problem that is expensive, visible and recurring.

Why this AI startup says its moat is data

Competition is emerging, but Flox argues that its edge is not just the product itself. It is also the field data it has accumulated across different terrains, species and infrastructure settings.

That data, the company says, gives it what it describes as ground truth about what actually persuades a moose, a bird or a bear to move away safely. In a category where effectiveness is measured in incidents avoided, real-world evidence may matter more than a flashy demo.

Competitors include Robin Radar’s Scarecrow, which uses species-specific distress calls and records activity, and iCHASE, which identifies birds with computer vision and aims a moving laser at them, adjusting the response as needed.

Those products suggest the market is still forming. Some systems focus on birds, some on radar, some on lasers and some on auditory deterrence. Flox is trying to build a broader platform that can adapt across species and use cases, not just one type of animal or one type of site.

What makes Flox different from older deterrence tools?

It uses AI to choose the response, not just a fixed alarm. That difference matters because the best deterrent can depend on the species, the environment and the animal’s behavior at that moment.

Traditional approaches often rely on static defenses:

• fences and barriers
• sirens and loudspeakers
• flashing lights
• falconry or trained birds of prey
• culling or removal

Those methods can work in some settings, but they are often costly, difficult to maintain or ethically fraught. They can also fail when conditions change, wildlife learns to ignore them or the barrier itself becomes impractical to maintain over long distances.

Flox’s pitch is that perception and adaptation can do better than brute force. By identifying animals locally and responding in a targeted way, the startup hopes to reduce both infrastructure disruption and harm to wildlife.

How does the startup fit into the broader AI market?

Flox sits at the intersection of physical AI, edge computing and industrial automation. The company’s thesis is that the next major layer of machine perception will not only help robots navigate warehouses or self-driving systems interpret roads; it will also help outdoor systems deal with living, unpredictable environments.

That broader idea is important because much of the current AI investment conversation focuses on text, software agents and data center compute. Flox points to a different frontier: AI systems that must operate when the world is not neatly labeled, controlled or static.

For outdoor machines, that means understanding weather, terrain, motion patterns and animal behavior. It also means the technology must be robust enough for remote deployment, low-power operation and compliance reporting, all while remaining affordable for operators that may need dozens or hundreds of units spread over large areas.

Nozkova argues that machines working outside — from trains and trucks to drones and tractors — will increasingly need to understand the natural environment, not just the human-built one.

That is the larger commercial bet behind Flox’s wildlife-deterrent system: if a company can help infrastructure sense and respond to nature, it may be able to expand into a much wider class of outdoor automation problems.

Timeline: how Flox evolved from research to startup

The company’s path from lab research to commercial deployment has been gradual, reflecting the challenges of building hardware and AI for the physical world.

Why investors may pay attention

Flox has a few qualities that make it stand out in a crowded AI startup market. First, it solves a tangible operational pain point with measurable costs. Second, its product is tied to real-world deployment rather than a demo environment. Third, it accumulates proprietary data as it expands, which could strengthen both the model and the moat.

There is also a sustainability angle. If the company reduces collisions and disturbances without requiring more fences or more aggressive methods, it could help operators balance safety, animal welfare and compliance requirements more effectively.

Still, the business will need to prove it can scale hardware, maintain accuracy across species and geographies, and win long procurement cycles with railways, airports and infrastructure operators. Those customers usually buy slowly, test heavily and demand reliability over novelty.

Even so, the market is real, the problem is expensive and the use case is easy to understand. A train that is delayed, an aircraft that must divert, or a road closed because of wildlife can quickly justify the cost of a preventative system.

What happens next for Flox?

The company’s near-term priorities are expansion, product generalization and fundraise execution. If the Series A succeeds, Flox says it wants to move from hundreds to thousands of devices and broaden the software beyond wildlife deterrence into more outdoor machine-intelligence applications.

That would bring the company closer to its broader vision: a perception stack for the outdoor world, where machines can sense, classify and respond to living environments instead of treating them as noise.

For now, the startup is making a practical bet that the right combination of edge AI, species-specific responses and field data can keep animals safer and keep infrastructure moving. In an industry where delays are expensive and collisions are dangerous, that may be enough to earn attention from operators and investors alike.

And for a founder who watched transportation systems struggle against nature, that may also be the most important point: the outdoors is not a bug in the system. It is the environment the next generation of AI will have to learn to read.