Could AI Become the Operating System of Global Trade?

From a corner office high above Singapore’s skyline, Scott Sherwood, chief technology officer at Kpler, can see some of the ships that help drive the global economy. Tankers creep toward port, bulk carriers haul grain, and liquified natural gas (LNG) vessels slice through the haze.

But Sherwood isn’t watching the water. He’s thinking about the next command those ships might receive—not from a trader’s phone call, but from a machine that, according to Kpler, has already done the thinking.

Kpler, a technology company that’s spent a decade tracking global trade with what it calls near-surgical precision, is now trying to go further. Kpler’s new system, Copilot, aims to act less like software and more like a colleague, one that could interpret market shifts, anticipate outcomes, and suggest what to do next.

In a business where a single cargo diversion could ripple through entire economies, Kpler knows that Copilot’s goal—to turn trading itself into a dialogue between human intuition and machine reasoning—is ambitious.

Trading Unplugged from the Spreadsheet

Kpler says it built its reputation by collecting data on everything that moves—tankers, terminals, emissions, and storage levels. Its clients then use that intelligence to map flows of oil, gas, and grain around the planet. But Sherwood argues that reporting data is no longer enough, and that the advantage now is in understanding it.

“This industry has always run on instinct, spreadsheets, and endless email chains,” he says. “We’re threading that instinct into an operating layer that can advise you directly.”

Kpler says that if you ask Copilot whether to reroute diesel from Fujairah in the United Arab Emirates to Rotterdam in the Netherlands, it can weigh the tariffs, weather, political risk, and compliance exposure before suggesting a course of action. What once could have taken hours of analysis may unfold in minutes, according to Kpler—not because the human element has vanished, but because the company says machines have finally learned to speak their language with the most accurate data.

A Machine That Reasons

Kpler says that behind Copilot’s interface is a web of models trained on a decade of proprietary data: satellite feeds, customs filings, and port call records. Kpler says its strength lies not only in access to data, but in the context it can build from long-term observation of how real trade behaves, such as how cargoes shift when droughts hit or sanctions land.

Sherwood describes its process as “retrieve, reason, act.” The system draws on what AI researchers call contextual grounding, combining structured data with the unpredictable rhythm of physical trade. Its goal isn’t to replace the trader’s logic, but to extend it—to help spot the unseen correlations, the price distortions that could hint at something deeper.

“It’s not summarizing anymore,” Sherwood says. “It’s thinking—and it never sleeps.”

Kpler’s engineers say Copilot can already detect subtle anomalies in grain flows or tanker routes that might take a human analyst days to surface. The machine doesn’t always explain why it sees what it sees, but in markets that move billions, Kpler says that even a hint could be enough for decision-makers to act.

Beyond Dashboards, Toward Autonomy

For veteran traders, the idea of a machine weighing in on million-dollar decisions could feel unnerving. But Kpler says Copilot is designed to amplify, not automate, human judgment.

“The human is still the strategist,” Sherwood says. “The machine just compresses the time to do the analysis by identifying patterns and filtering out the noise.”

Still, the company is experimenting with an autopilot mode, where Copilot could autonomously redirect shipments, trigger smart contracts, or rebalance portfolios within strict human-set limits. It’s not full autonomy, says Kpler, but it edges close to it—a kind of supervised agency.

A trader involved in early trials described it to Kpler as “a second pair of eyes that never blinks.” The challenge, he apparently said, isn’t trusting the system—it’s learning to ask better questions.

Kpler says its vision echoes a broader trend across global trading desks: the rise of AI agents that can act on live data. Firms from Geneva to Houston are experimenting with LLM-powered copilots to accelerate risk analysis and logistics planning, the company says. What makes its effort distinct, according to Kpler, is its focus on the physical—on ships and cargoes.

Decisions With Geopolitical Weight

Unlike algorithms that decide what ad you see or what stock to buy, Kpler says that Copilot operates in a world where a single decision could affect national economies.

“We’re talking about sovereign-level consequences,” Sherwood says. “One suggestion from Copilot could move fifty million dollars in cargo—or shift a country’s energy balance.”

For example, a rerouted LNG cargo could change power generation schedules in Europe, or a grain diversion could reshape food supply in North Africa, according to Kpler. Each decision is a tiny lever on global stability—and Kpler says that machines are beginning to touch those levers.

A Future That Talks Back

The trading terminal of the future might not be a dashboard at all, according to Kpler. Instead, it could be a conversation. Sherwood imagines a world where traders talk to their systems the way pilots talk to copilots: with oversight, trust, and constant feedback. While the ethics and intent would remain human, the execution would become machine, with data being the shared language.

“The next trading platform isn’t a screen full of charts,” Sherwood says. “It’s a voice that talks back.”

That vision is still aspirational. Kpler says that AI in commodity markets remains tightly governed, human-in-the-loop, and constrained by compliance and audit trails. But the company believes the industry is moving in that direction. The spreadsheets that once defined trading may be giving way to dialogue—faster, more dynamic, and, according to Kpler, perhaps more transparent.

The Human Loop

Commodity trading has always been a paradox, according to Kpler: a world of intuition and relationships powered by data measured to six decimal places. Kpler says its Copilot system sits at that intersection. It is not designed to eliminate the human gut—but instead aims to give it a wider field of vision.

“We’re teaching the system to see the world the way a trader does,” says Sherwood. “And maybe, one day, to show the trader what they’ve been missing.”

Click here to get early access to Copilot.