Flox Intelligence's edge AI devices detect wildlife and drive them off rail lines and runways

Flox Intelligence, a startup that emerged from Sweden's KTH University, builds battery-powered edge devices that identify animals near railways, runways and other critical sites using on-device AI, then drive them away with sounds that keep changing. Around 150 of the devices are already operating across North America, Europe and Australia, according to a Hacks.gr report republished on dev.to.

How the devices work

Each unit is mounted on a pole or a tree along the infrastructure being protected. A motion sensor wakes the device when it detects activity, which switches on a camera. The animal is then identified locally on the hardware itself, with no cloud round-trip, and the device emits a sound chosen to scare off that particular animal.

The central design choice is variation. The system rotates the audio signals it plays so animals cannot get used to one repeated cue — the failure mode that undermines conventional deterrents such as loud sirens and fixed fencing. Alongside the hardware, Flox offers a software platform for monitoring device operation and producing reports.

The company was founded in 2020 and launched out of KTH in Stockholm. Its founder, Sara Nozkova, had previously spent 13 years working in transportation and autonomous vehicles, and the system was developed in collaboration with KTH's artificial intelligence laboratory.

Where it is being used

Current deployments sit at airports, along rail corridors, in national parks and at other critical infrastructure sites on three continents. Flox is also working with train manufacturers, including Alstom, to integrate its software directly into vehicles, which would carry the deterrent from fixed wayside hardware onto moving trains.

Business model and funding

Flox sells the devices for an upfront fee and charges an annual subscription for the software after the first year. According to Hacks.gr, the company expects to close 2026 with revenue of roughly $1.5 million, and the report names Robin Radar, Scarecrow and iCHASE as other companies operating in the same space.

The startup has raised about $5 million to date from investors including Unconventional Ventures and Tilia Impact Ventures. It is a finalist in the Startup Battlefield 200 competition at TechCrunch Disrupt in San Francisco and has opened a new $10 million funding round, with the stated aim of growing its active fleet from 150 devices to several thousand.

Why it matters

Animals straying onto runways and rail lines are a long-running safety and cost problem for transport operators, and the usual countermeasures — fences, static sirens — tend to lose effectiveness as wildlife adapts. Flox is a compact case study in applied edge AI: a vision model runs inference on battery-operated hardware out in the field, where connectivity cannot be assumed, and the model's output is a physical action rather than a dashboard metric.

The habituation-resistant rotation of sounds is what makes the AI genuinely load-bearing rather than decorative, since classification only helps if the response stays effective over time. The commercial signals are worth watching too: a hardware-plus-subscription model, an integration deal with a major rolling-stock manufacturer, and a raise intended to scale deployments from 150 units to several thousand will test whether this niche of applied AI can grow beyond pilot-sized installations.