Stockholm startup Flox Intelligence is selling AI-powered edge devices that spot animals near trains, airports and roads, work out what species they are, and scare them off with a sound picked for that animal. TechCrunch AI reports that the company is a finalist at this year’s Startup Battlefield 200 at TechCrunch Disrupt. It’s also opening a $10 million Series A to grow from about 150 deployed units to several thousand.
The problem is costly. According to TechCrunch AI, the FAA estimates that wildlife strikes cost U.S. civil aviation about $272 million a year on average between 1990 and 2025. In France, rail operator SNCF says wandering animals affect more than 8,000 trains a year. In Brittany alone, 120 trains hit animals in 2023, which disrupted 660 services and added more than 15,000 minutes of delays.
🦌 What Flox actually does
Founder Sara Nozkova spent 13 years in transportation and autonomous vehicles before joining the AI lab at KTH Royal Institute of Technology. Flox spun out of KTH in 2020. Her pitch is simple: “I was convinced the answer was not a louder deterrent.”
Here’s how the system works, step by step:
- Motion wakes it up. A motion sensor sets the device off when something moves nearby. That saves power on battery-operated units mounted on poles, trees and other structures.
- A camera identifies the animal on the device. Models running on the device classify the animal without sending anything to the cloud. That matters in remote corridors where connectivity is patchy or missing.
- It plays a sound that fits the species. Flox emits a sound matched to the animal it’s seen, rather than one generic siren that scares everything (or nothing).
- It changes its responses over time. Nozkova says the system learns from experience and mixes up its signals so animals don’t get used to them. Habituation is the big weakness of traditional deterrents, so this is the part to watch.
- A software layer for clients. A dashboard shows where devices are deployed and what’s happening on the ground. It also produces reports for regulatory compliance and biodiversity monitoring.
📍 Where it’s running and how it makes money
About 150 Flox devices are live in North America, Europe and Australia. Deployments include airfield perimeters, road and rail corridors (especially fence ends and wildlife crossings), national parks, farmland, mining sites, urban areas and electrical substations.
The business model has three layers:
- Hardware sold up front: customers buy the edge units outright.
- Annual subscription: a software and AI fee kicks in after the first year.
- OEM licensing: partners pay an annual fee to run Flox software on their own vehicles. Train maker Alstom is one of them.
Nozkova expects Flox to finish 2026 with about $1.5 million in revenue. So far it has raised around $5 million across pre-seed and seed rounds. Unconventional Ventures and Tilia Impact Ventures led the seed, with GENIUS NY, Norrsken, Satgana and angel investors also taking part.
⚔️ The competition
Flox isn’t the only one in this space. Robin Radar’s Scarecrow also plays species-specific distress calls to scatter birds and logs the activity. iCHASE uses computer vision to detect birds and targets them with a moving laser, adjusting as it goes.
Nozkova says Flox’s edge is the range of animals it handles and its field data, which she calls “ground truth on what actually makes a moose or a grizzly bear move safely.” That’s a fair claim to make. Bird control is a crowded niche, while large mammals on rail and road corridors get far less attention.
🧭 Why this matters beyond scaring off deer
What stands out is the bigger bet. Nozkova frames Flox as a perception company, not a deterrent company. “We have taught machines to be brilliant, but only inside the environments we engineered for them,” she said. Outside, in nature, she adds, “the same machines quickly go blind.”
The long-term plan is to make the software portable enough to run on trains, cars and drones. In her words, “nature intelligence is the next layer of the perception stack. Almost nobody is building it.”
That ties into one of the industry’s hottest themes. Physical AI is moving out of warehouses and onto open roads, farms and rail lines, and those systems will keep running into animals they weren’t trained to understand. If Flox’s field data holds up, licensing that perception layer to OEMs could be worth far more than the boxes on poles.
⚠️ Caveats
This is still an early-stage company. 150 devices is a small footprint, the revenue figures are Nozkova’s own projections, and TechCrunch AI’s report doesn’t include independent data on how well the deterrent works. The Series A will mostly fund scaling to thousands of units and expanding beyond wildlife use cases, and that’s when we’ll see whether the approach holds up at scale.
The full story, including Nozkova’s background and the Disrupt pitch, is in TechCrunch AI’s original report.