Leaders from OpenAI, Olam Agri and GoComet joined senior supply chain executives to examine how AI can take on more of the operational work behind global trade.
SINGAPORE, Sept. 24, 2026 /PRNewswire/ — Last week at Odyssey, its flagship gathering of the leaders who run global supply chains, GoComet launched Nova, an AI-native execution layer for enterprise logistics. The Singapore edition was attended by over 150 supply chain and logistics leaders from global enterprises, including Shell, BHP, Bayer, L’Oréal and DP World.
With speakers from OpenAI and Olam Agri, the day turned on one question: what happens when software stops reporting the work of global trade and starts doing it?
$70 billion of trade a day, run manually as pressure mounts
Global merchandise trade reached about US$25 trillion in 2025, according to UN Trade and Development, roughly US$70 billion of physical goods crossing borders every day. The machinery that moves them has grown highly sophisticated; the coordination that connects it has not.
Much of it still runs on people chasing status, comparing systems, validating documents and reconciling charges by email, spreadsheet and phone. That coordination layer, GoComet argues, is the largest piece of enterprise work still waiting to be automated. And the pressure on it is rising.
Tariffs shift, routes move, suppliers slip and capacity tightens, and supply chains are expected to respond faster each time. The hard part is no longer making the decision; it is carrying it through the thousands of manual actions that follow.
“The strategy can change quickly,” said Chitransh Sahai, Co-founder and CEO of GoComet. “The difficult part is carrying that change through the operation. Every order, partner, booking, document, shipment and payment then has to move with it. Today, people are still carrying too much of that work manually.”
Nova: from software that reports to software that acts
That is the shift Odyssey set out to examine: software that no longer just records and reports work, but begins to carry it out. In a keynote, Anthony Russell of OpenAI set out how AI is shifting software from systems that record and report work to systems that act on it, and how enterprises should start measuring and redesigning workflows around that now.
Vivek Agarwal of Olam Agri brought the operator’s view — how a $30bn agribusiness across 30+ countries is automating freight procurement without forcing every business unit onto one template, because agentic AI can adapt to how each already works.
GoComet showcased Nova, an AI-native execution layer for enterprise logistics built on more than eight years of cross-border logistics context: over 50 million shipments, 500-plus enterprises and more than 70 countries.
“Nova is built on 10+ years of cross-border logistics context, then grounded in how each enterprise actually operates,” Sahai said.
On stage, GoComet ran Nova across a single freight journey from planning to payment, carrying that context from one step to the next rather than rebuilding it at each handoff. GoComet showed how Nova monitors an inbox around the clock, catches a document error and opens and closes a service ticket. Attendees got a live look at how it pulls and compares contract-versus-spot quotes under a cost guardrail and books the shipment; with a human sign-off where it matters.
“The point is not to put more AI in front of a logistics team,” Sahai said. “The point is to remove the coordination between a decision and its outcome.“
Useful AI in logistics, the company argues, needs more than a model; it needs the context behind each decision: the shipment, rate, carrier, document, commitment, rule and approval. Nova carries this context — documents, rates, shipping records, tracking, invoices — across every step, instead of rebuilding it at each handoff.
That is the first step toward what GoComet calls the Invisible Hand of Global Trade, an operating model in which software quietly handles the coordination beneath trade, so that trade just happens.
The end state is a cleaner division of labor. AI does the execution: it finds the information, reconciles it, updates the systems, follows up with the missing party, and closes the routine loop. Humans do the deciding: negotiating a critical rate, managing a stranded shipment, weighing the trade-offs and handling the exceptions no model should own alone.
About GoComet
GoComet is the AI-Native Platform for Global Trade, helping enterprises take control of freight operations from Planning to Payment. Built on more than eight years of cross-border logistics context, GoComet has managed 50M+ shipments and serves 500+ enterprises across 70+ countries.
Nova, GoComet’s AI-native execution layer for enterprise logistics, works across each enterprise’s processes, systems, partners, rules and approvals to help reduce manual coordination, control freight cost and improve execution.
GoComet’s long-term vision is to build the Invisible Hand of Global Trade, where technology handles more of the operational complexity underneath global trade while people focus on judgement, relationships and growth.
GoComet is headquartered in Singapore, with teams across India, the United States, Dubai, Japan and Southeast Asia.
To learn more, visit www.gocomet.com
