By: Bertrand Chen

Few technologies have captured the imagination of boardrooms quite like artificial intelligence.

In recent months, companies have rapidly expanded AI budgets in pursuit of competitive advantage. Uber recently admitted it had exhausted its annual AI budget within the first four months of the year, while hyperscalers continue to commit hundreds of billions of dollars towards infrastructure. Across boardrooms, AI is no longer one thing. It is about investment, competition, geopolitics, open versus closed models, even layoffs. Much of today’s anxiety seems to trace back to AI in one form or another.

But beneath all of this sits a simpler question. When the investment has been made, who will be the ones realising its value? 

Many of us are already familiar with Generative AI. Whether it is generating emails or producing videos of Harry Potter in alternative universes, AI has largely acted as a highly capable assistant. The next phase is Agentic AI, where systems do not simply recommend actions but execute them end-to-end. This matters because execution changes the equation.

Agentic AI through the lens of global trade

Once an AI agent is authorised to order inventory, approve invoices, arrange financing, or trigger payments, the challenge is no longer intelligence. The challenge becomes trusted coordination.

Global trade offers a useful lens to understand this shift. A single international shipment can involve carriers, shippers, freight forwarders, terminals, customs authorities, insurers, and banks. Each participant operates different systems, follows different processes, and has different incentives. Despite this complexity, goods, data, and money must move synchronously.

Historically, this coordination has relied on manual processes, document checks, and institutional relationships. These are inefficient, but they work because they create trust.

Yet as AI takes on more of this work, automation alone does not resolve the underlying issue. Trade is not simply a data challenge, but a trust one. If participants cannot rely on the integrity of actions taken across the network, the system breaks down. In this context, an AI agent that triggers a payment is only as valuable as the trust behind that instruction.

Automation demands trusted infrastructure

This is not the first time the industry has faced such a challenge. Over the past decade, blockchain technology has travelled its own path through the hype cycle. Early expectations centred on disruption, but over time, its role has become clearer as a coordination infrastructure. AI can interpret information and make decisions. But it cannot, on its own, establish shared agreement between multiple parties. That requires a coordination layer.

In global trade, this coordination increasingly sits around the electronic bill of lading, serving as a legally recognised proof of title ownership for each shipment. This has been made possible because of blockchain, which guarantees verifiability and immutability, meeting the legal requirements that govern digital trade documents. In turn, it becomes a trusted data exchange vehicle that allows for supply chain participants who would otherwise not trust each other to collaborate.

With this as a foundation, the eBL becomes a ‘data container’ that links different systems associated with a shipment, from logistics to its financing and payment. And this is where the convergence becomes critical. As payments evolve, they are becoming programmable. Stablecoins, tokenised deposits, and CBDCs allow funds to move automatically once predefined conditions are met.

The value of AI lies in trust

If AI agents are to act in the real economy, those actions must connect directly to trusted payment rails. When an AI agent identifies that goods have arrived and triggers a payment, that instruction must be anchored to verifiable data, such as an authenticated eBL, and accepted by all parties involved.

For this reason, the future value of AI will not be determined solely by the sophistication of models. It will depend on the infrastructure that allows AI to act with trust. The internet did not transform commerce simply because websites improved. It transformed commerce because trusted transaction infrastructure emerged at scale. AI may follow a similar path.

For treasury and payments professionals, the implications are that as AI agents begin to participate in financing, settlement, and cross-border trade, the central question is no longer what AI can think, but what it can be trusted to do. The next chapter of digital transformation may therefore be less about intelligence itself and more about the systems that allow intelligence to execute safely in the real world.