Skypuzzler says its platform runs without AI and coordinates roughly 100 drone operators at the Port of Rotterdam. The company is selling auditability to the FAA, with low altitude rules open.
The Port of Rotterdam runs roughly 100 drone operators through a single coordination platform, according to Skypuzzler, the Copenhagen company that built the software. The system does not learn from flights. It solves for them. That is the technical choice Skypuzzler is now pitching to U.S. regulators as the FAA moves toward picking the architecture for low-altitude airspace.
Skypuzzler's product, what the company calls an Integrated Digital Air Traffic Control system, combines two layers of conflict resolution. One runs before takeoff to sequence routes and reserve corridors. The other runs in flight, comparing each drone's filed trajectory against every other drone in the same volume and rerouting when paths would otherwise cross. Both layers, Skou and Skypuzzler say, are written as deterministic algorithms rather than trained models. Chief executive Jesper Skou, a former air traffic controller in Danish and Swedish airspace, has framed the choice as one regulators can audit. DroneLife summarized the pitch as "math, not AI" in a recent profile.
The FAA is still drafting the rule that will determine how drones, cargo operators, air taxis, and inspection fleets share the airspace below 400 feet. That rule will set which kinds of conflict-resolution software the agency will accept, and on what evidence. The same week Skypuzzler's profile landed, the FCC moved on an adjacent question, proposing a sales ban on military-grade foreign drones. The two actions are not connected, but they share a posture: the United States is drawing the perimeter of who flies, with what hardware, under which technical rules.
United Airlines Ventures, the corporate venture arm of United Airlines, has made a market vote on which way the rule could go. The fund invested in Skypuzzler, DroneLife reported this month. The dollar amount was not disclosed. Skypuzzler has named Dallas as its first U.S. target, citing dense commercial demand and an existing partner network. A U.S. carrier with an active drone program is now backing an auditable platform over a learned one.
Skypuzzler says the Rotterdam platform coordinates cargo runs, oil-leak survey inspections, and ad-hoc requests across the port and adjacent facilities, working off telemetry the drones already transmit to existing drone traffic management infrastructure. No new hardware on the aircraft. The company has signed a memorandum of understanding with Thales, the French aerospace and defense group, to integrate with Thales' TopSky and AstraUTM products (Unmanned Airspace), and counts DSV, the Danish global logistics company, among its customers. The company closed a seed round with Rethink Ventures and the European Innovation Council Fund in August 2025, with additional backing from an EIC Accelerator grant (EIC).
A deterministic conflict-resolution system can be inspected the way an air traffic control radar can: line by line, against a known safety case. A trained model offers a different promise, that it can adapt to traffic patterns no human has seen, and a different problem, that its decisions are harder to defend in an accident review. Skypuzzler is betting the FAA will weight the first property more heavily than the second. An AI vendor could still win that bet, but only by wrapping a model in a certification surface that lets reviewers see the same line-by-line logic.
The Rotterdam workload is dense, but it is geographically bounded. A port, with a defined perimeter and a finite operator set, is a clean test case. The U.S. ask is messier: long-distance cargo corridors, ad-hoc emergency response, mixed-altitude operations near airports, and state-level rules that do not all align. The auditability argument that wins in Rotterdam still has to clear the noise of a national airspace.
The FAA's next published technical roadmap for drone traffic management, expected in the fall, will be the first read on whether the agency is leaning toward auditable algorithms, certifiable AI, or a defined interface that lets either type of system qualify. Skypuzzler will be pitching in. So will the AI vendors. The decision is not about which technology is smarter. It is about which one a regulator can stand behind when something goes wrong.