Logtech Essentials: From Procurement to Payment

Special Report

COMMENTARY

By Bertrand Chen

AI gives the frontline a new voice in how technology gets built.

Much of the discussion around AI focuses on the technology itself, and the eye-watering sums being invested in the pursuit of artificial general intelligence( AGI).

But beyond this noise, the shipping industry risks overlooking the more practical question: What can actually be built with AI?

This matters more now because the barriers to experimentation and adoption are falling fast, thanks to a growing range of models changing the economics of what businesses can build with AI.

The AI market is often presented as a race among companies developing closed frontier models, such as OpenAI and Anthropic. These models offer extraordinary capabilities at a premium price, but they are no longer the only option.

Open-weight models— AI models that are publicly available— are expanding the offer. This year we have seen a drastic shift in usage toward Chinese-built models GLM, DeepSeek and Kimi, but this is not exclusively a Chinese story. European developers such as Mistral and American alternatives including Muse Spark and Nemotron also offer alternatives. In 2026, ocean carrier CMA CGM announced a major agreement to deploy Mistral AI across its operations.

For shipping, there is another practical reason to consider these models. In an industry where data crosses jurisdictions, questions of sovereignty, security and data control are central to AI adoption. No company can afford to have access to critical technology rescinded arbitrarily. Openweight models offer greater control over deployment while lowering the cost of experimentation. This does not mean every company should use only open-weight models. But as AI becomes cheaper and more accessible, the differentiator will increasingly be knowing what to build, and which model to use to build it.

This is where many vendor-led transformation programs go wrong. Traditionally, technology teams have been tasked with building solutions while business teams are consulted through requirement- gathering exercises. Yet, some of the people who understand what needs fixing best are those who encounter the problems every day. Every company has frontline workers that understand the challenging handoffs, repetitive processes, and operational frustrations that get in the way of doing their jobs. The problem is not necessarily that they lack ideas. It is that, historically, they have had limited ways to express what a better solution might look like. AI changes that.

People working closest to customers and operations can now turn ideas into basic prototypes, even with very little technical training. Instead of simply describing a problem or writing a lengthy requirements document, they can show what a potential solution might look like, test assumptions and make their ideas tangible. AI gives the frontline a new voice in how technology gets built.

This does not make IT redundant. A prototype vibe-coded in an afternoon is not an enterprise grade solution. It still requires data, governance, and serious engineering.

But what changes is the starting point. Frontline teams can make their ideas visible and test whether a proposed solution addresses their actual pain points before the organization commits significant time and investment. The prototype then becomes a much richer bridge between business and technology.

We saw this during a recent“ vibeathon” organized by the Global Shipping Business Network in Shanghai. Participants included experts from shipping lines, terminal operators, standards bodies and software vendors around the world, spanning commercial, operational and technology functions. The event was not driven primarily by AI vendors, but by people who understood the business.

One group began with little development experience. Using DeepSeek Harness with various openweight models, it created a three- dimensional rendering of a container vessel leveraging the Vessel Data eXchange( VDX) vessel profile standard introduced only the previous day. Another team explored transshipment operations and created a prototype decision tool combining information, a missed-connection risk engine, and an intuitive interface to explore different planning scenarios.

Again, neither was production-ready. What mattered was that operational knowledge could be turned into something tangible enough for business specialists, management and IT to inspect, debate and decide whether it justified further investment. Remarkably, the entire token consumption for 38 people prototyping an entire afternoon amounted to only $ 20.

Every company has people who care deeply about fixing the problems they encounter every day. AI now gives them a way to turn what they know into what could be built.

The role of leadership is not simply to hire more AI experts. It is to identify the people who understand the problems, expose them to these new tools and let them experiment.

Bertrand Chen is the CEO of the Global Business Shipping Network. email: bertrand @ gsbn. trade

38 Journal of Commerce | November 2, 2026 www. joc. com