As artificial intelligence continues to transform digital processes in airfreight, a new frontier is emerging: robotic AI. Unlike traditional robotics, which rely on complex pre-coded instructions, robotic AI uses natural language processing and visual inputs to perform physical tasks—making it potentially revolutionary for cargo handling. Yet, as two industry leaders reveal, the path to automation on the warehouse floor remains uneven.
“Not many people talk about AI hardware,” said Matt Petot, founder and CEO of CargoAi. “Before you had robotic AI, you basically had to code the robot—move from here to there. But now… you actually don’t need to code the robot.” With large language models embedded, such as those powering ChatGPT, robotic systems can now understand commands like “pick up the box” or “place the package on the pallet” without explicit programming. “You can talk to it like you do with ChatGPT… and it can do it,” Petot explained. “They look human because they have arms and legs… and they can adapt to all environments.”
Though CargoAi itself is not developing robotic AI—its focus remains on AI-powered booking and pricing tools—Petot believes the technology is closer to the air cargo industry than many think. “They’re still in prototype mode right now, but far more advanced than what people believe… I think next year we will see a lot of them,” he said. “The cost will not be more than a car.”
24/7 labour—without the back pain
While the promise of robotic AI is clear—around-the-clock work, fewer injuries, and faster setup—Petot cautioned that performance is still limited by reliability and speed. “Right now it’s a bit slow… but it doesn’t get sick, it doesn’t get hurt,” he said. “You can break down, but at least you don’t hurt somebody physically.” He also pushed back on the idea that robotic AI requires prohibitive groundwork. “That’s the biggest misconception,” he said. “Before, you needed to have the data structure in the right way. Now… you can plug an AI on top of it, even if it’s unstructured.”
A reality check on the warehouse floor
But while the tech vision is compelling, Jeroen Giling, head of cargo for the Netherlands, Belgium and Finland at Swissport, delivered a starkly different assessment from the front lines. “In air cargo, what’s leading is the aircraft,” Giling said. “If you currently open a lower deck of a 747… the cargo is coming out the same way as it was in 1970.” Despite decades of digital progress in messaging and documentation, physical handling remains stubbornly unchanged.
“The build-up on the ULD still has to be done by somebody—with a forklift,” he said. “We are still using nets. Handling a net on a ULD is an activity we’ve been doing for 55 years, and nobody has invented an alternative.” Giling believes the lack of transformation isn’t due to unwillingness, but structural barriers: “Whatever solution you find has to comply with ICAO, IATA, FAA, EASA, Boeing and Airbus standards. That’s quite a task.”
While Giling acknowledged significant innovation in data exchange—”We made the way of processing the information more efficient”—he argued that physical processes have seen minimal disruption. “From a physical point of view, there is not so much innovation,” he said. “The message 20 years ago was this big. The message today is this big,” he added, stretching his hands apart. “So we had to digitalise… but it didn’t make the work itself smaller or simpler.”
Will robotic AI break through?
As pressure mounts to automate labour-intensive, high-risk tasks such as ULD build-up and breakdown, robotic AI may offer a long-term solution. But implementation will require more than just technical capability—it demands regulatory clearance, industry alignment, and above all, a shift in mindset.
“We’re not far,” Petot said optimistically. “It’s a matter of a few years, not ten.” Yet on the ground, the fundamentals remain unchanged. “We think we are very innovative,” Giling said. “But if you still use the same activity now compared to 55 years ago… how innovative have we really become?”