The Galician technology company Visual Trans, specialized in software for freight forwarders, freight agents and international logistics operators, has announced its participation in the XXII Customs Forum to be held in the Portuguese capital. At the event, the company will present its line of agentic artificial intelligence solutions: systems capable of executing complete tasks autonomously, which the company calls virtual employees or digital workers.
The presentation is part of one of the featured conferences of the meeting, which brings together more than 350 foreign trade professionals from November 12 to 14 in Lisbon in the first Iberian edition of the forum.
For almost three decades of experience, Visual Trans has accompanied the sector, bringing each innovation closer to its products to improve the efficiency of companies in the logistics sector. Today, agentic artificial intelligence represents another step. The difference with respect to generative AI – which has dominated the last two years – is one of nature, not power. A conversational assistant works in turns: someone asks, the system answers. A virtual employee receives an objective and a configuration and, from there, he or she chains the necessary actions to achieve it on his or her own. The company, which first implemented them in its own processes, defends a gradual adoption model with the person as a point of control: “we automate tasks from start to finish, but maintaining all human capital,” comments Santiago González, head of vCustoms product development at Visual Trans.
The human being supervises and validates
For Visual Trans, the value of people who work in the logistics sector is crucial, providing precision and responsibility for each shipment or declaration. And now, agentic artificial intelligence is the next episode: A virtual employee receives an objective and a configuration and, from there, chains on his or her own the necessary actions to fulfill it. Until now, “artificial intelligence was something that the professional decided to consult,” explains González. “A virtual employee is not consulted. He is always entrusted with the human in supervision and validation tasks. And that changes the conversation completely, because it stops being an individual decision and becomes an organizational decision.”
In practice, this translates into tasks that appear resolved without anyone having started them that morning: a square bank reconciliation, a file registered based on the documentation received by mail, a tariff query resolved with its documented justification.
Tangible results in the logistics sector
These types of solutions are already underway in several clients in the logistics sector – and outside it – with truly striking results: since May of this year, virtual employees implemented in clients have carried out the equivalent of 3,500 hours of administrative work. In the month of October, an important logistics and customs group has already carried out 98% of its bank reconciliation with an incidence rate of less than 2%. At customs, since the end of September, it has filed in its file 6,148 of the 6,162 clearance documents that arrived. And when the documents don’t add up, let someone know.
Regarding data quality, in just two months virtual employees detected around 33,000 errors in shipments that were reviewed and corrected by people, freeing up more than 200 hours of work.
The uncertainty that accompanies change
The logistics sector is a chain of processes that is repeated thousands of times a day. For merchandise to reach its destination, documentation must be prepared, forms filled out, data verified, deliveries coordinated, incidents resolved… Today, with the help of virtual employees in charge of these tedious tasks, people can dedicate themselves fully to tasks with high added value. “What you gain is the margin to look further, provide better service or attract more clients,” González clarifies.
The company’s approach rests on three limits that accompany each virtual employee. It operates within an explicit parameterization that defines what tasks it executes, how often, and over what sources. Leave a trace of each decision, so that it can be reconstructed months later when verified. And he stops to ask when something doesn’t fit, instead of continuing.
Each virtual employee operates within an explicit parameterization that defines what tasks he or she executes, how often, and over what sources.
“In foreign trade operations, an error is expensive, and the responsibility for a declaration always falls on a specific person,” says González, who concludes that “deciding at what point in an automated process the system stops so that someone can validate it is not a computer decision, it is a professional decision. Technology can suggest, prepare and execute, but the signature and responsibility remains – and will continue – to be human.”
