Flox Intelligence, a Swedish spin-off from Stockholm’s KTH Royal Institute of Technology, is opening a $10 million Series A — roughly €8.7 million — to scale the edge-AI hardware it uses to keep wildlife away from runways, railway lines and highways. The round is not closed yet, but around 150 devices are already installed across North America, Europe and Australia, and the company is pitching investors on a problem with a measurable price tag.
Flox was founded in 2020, keeps its R&D in Stockholm and now runs its headquarters in New York State.
- $10M raise: Flox is opening a $10 million Series A, still unclosed, to scale from 150 devices to several thousand.
- Edge processing: Battery-powered units identify animals with on-device AI, then play species-specific sounds, varying them so wildlife does not habituate.
- Measured demand: The FAA logged 24,447 US civil aviation wildlife strikes in 2025, up 9.3% year over year.
Cameras, on-device models and species-specific sound
The unit, called Edge, is a battery-powered autonomous device that processes everything locally rather than in the cloud — a design dictated by the remote sites where it operates. Mounted on poles, trees or other structures, a motion sensor wakes it, a camera captures the scene, and on-device models identify the animal before the device emits a sound matched to that species.
The differentiator is variation: responses change over time so animals do not simply grow used to the noise, the flaw that undermines conventional acoustic scarecrows.
“I was convinced the answer wasn’t a louder deterrent,” said co-founder and CEO Sara Nozkova, who spent 13 years in the transport and autonomous vehicle sectors.
Customers also get a software platform showing device locations and detections, plus automated regulatory compliance and biodiversity reports. Flox says Edge has logged more than 60,000 wildlife events across 27 species — figures that have not been independently audited.
Why airports anchor the business case
The FAA’s latest wildlife strike report, published in August 2026, counted 24,447 strikes in US civil aviation during 2025, up 9.3% year over year, with birds accounting for 94.4% of them. The report estimates roughly 74,372 hours of aircraft downtime and $391 million in losses, while cautioning that those totals may sit at the high end because of how costs are reported. Between 1990 and 2025, the agency puts average annual costs at about $272 million.
Flox already has reference customers: Gerald R. Ford International Airport and the US Department of Agriculture. It previously tested autonomous drones to clear wildlife at the Swedish airports of Malmö, Kiruna and Umeå, and prepared a demonstration at a San Mateo County airfield in California.
Beyond the runway
Devices also sit along road and rail corridors, in national parks, on farms, at mines and at electrical substations. With Alstom, Flox has an OEM-style agreement under which the manufacturer pays an annual fee to integrate its software into vehicles. Nozkova frames the wider opportunity as “physical AI” moving out of the factory, and argues those machines will need a layer of “nature intelligence” that “almost nobody is building.”
Funding, revenue model and rivals
Flox has raised roughly $5 million across pre-seed and seed funding. It took a nearly $1 million seed round in summer 2025 led by Unconventional Ventures with Norrsken Accelerate, a $3 million Seed+ extension in February 2026 led by Tilia Impact Ventures, and a $1 million prize from the GENIUS NY accelerator in May 2026, which requires it to operate in central New York for at least a year. Nozkova said Syracuse would become the global headquarters.
The model pairs an upfront hardware sale with an annual software and AI subscription from the second year. Flox expects to close 2026 with about $1.5 million in revenue. It is also one of five finalists in the Startup Battlefield at Disrupt 2026 in San Francisco, where the winner collects $100,000 equity-free.
Competition is real. Robin Radar‘s Scarecrow system emits alarm calls tuned to individual bird species, while iCHASE pairs computer vision with a mobile laser. Flox positions itself as multi-species, mammals included, built for remote low-connectivity sites, and claims to learn from experience. Nozkova calls it “the ground truth of what makes a moose or a brown bear move safely.”
The case for Flox rests on something many AI startups lack: an expensive, regulator-quantified problem, and customers already paying for weaker solutions. The harder question is scale. Moving from 150 units to several thousand across runways, rail corridors and mines is a hardware challenge as much as a modelling one, and the efficacy figures remain self-reported. Independent before-and-after strike data from a single airport would settle the argument. Until that arrives, the thesis is credible — but unproven.
Sources & References