Faced with increasingly complex supply chains and a steady rise in tariff and non-tariff regulations, it is imperative that companies rethink their customs processes to ensure accurate, rapid and fully compliant management of their trade flows. Penalties for errors or omissions can be heavy, while international competitiveness is often based on the reliability and fluidity of trade.
At MyTower, a French publisher of TMS (Transportation Management System) and GTM (Global Trade Management) solutions, we are currently conducting R&D projects to explore the integration of artificial intelligence (AI) in customs and export control. We are convinced of the high value that these technologies could bring to companies, both in terms of compliance quality and risk and cost reduction. Use cases studied include automated declaration control, intelligent product classification and global, multilingual regulatory monitoring.
These reflections and R&D work are part of our approach to the global digitization of customs processes proposed by our GTM solutions, aimed at simplifying interactions between the various Supply Chain players, making information more reliable and reinforcing compliance. In a context where data is critical and authorities are increasingly demanding, AI can be a decisive lever, provided that strict principles of protection, sovereignty and information quality are respected.
I. Digitization and Transformation of Customs Processes
1.1 Modernizing for better flow control
Historically, customs management and export control operations have often relied on manual methods (reading PDF documents, email exchanges, telephone calls) or on internal IT systems with little communication. This fragmentation increases the risk of errors and lengthens processing times, particularly when last-minute adjustments are required or data needs to be cross-checked.
Today, TMS and GTM solutions already enable the digitization of processes and improved collaboration between the various links in the supply chain. The idea is to centralize information, avoid re-keying and provide clear traceability at every stage.
In this context, AI-assisted systems can go a step further in the approach to reducing process complexity and effort, while at the same time reducing the risk of non-compliance. For this to work, AI models need to be trained with qualitative datasets, including audited and verified historical transaction data from official sources.
However, it must be borne in mind that when it comes to customs compliance and export control, there is no such thing as "gray": a declaration can be right or wrong, and the consequences can be severe. Companies therefore need solid guarantees to ensure that technology, however promising, does not introduce biases or inaccuracies that could expose them to sanctions.
II. AI applications in customs and export control
2.1. Automated declarations and compliance checks
Among the avenues being explored is the automation of customs declaration control. In the current system, a Registered Customs Representative (RRC) submits declarations on behalf of a customer. Checking the consistency and accuracy of these declarations is often time-consuming, especially as the information comes from multiple sources: invoices,ERP data, internal or external databases, etc.
Artificial intelligence could be used to analyze and cross-reference all this information, detect regulatory or business inconsistencies (e.g., a customs code mismatched with the nature of the products, values not aligned between the declaration and ERP, or a legal entity error), and trigger alerts at the earliest opportunity to rectify errors. The aim is to reduce the risk of penalties by detecting any anomalies early on.
Customs authorities and other administrations in charge of controls have already begun to implement AI in risk analysis, enabling them to target controls.
2.2 Product classification with partial data
Customs classification of products is another major challenge. When an item is clearly defined in the ERP, with complete and structured attributes, the use of rules and decision trees is often sufficient. On the other hand, for less well-documented items - spare parts, samples, atypical products - the situation becomes more complicated, as snippets of information (descriptions, photos, technical specifications) have to be cross-referenced to determine the appropriate customs code.
In this area, we have chosen a hybrid approach that would combine human expertise (via stable business rules) for finished products and the analysis capabilities of artificial intelligence to suggest a probable customs code for poorly characterized products, without validating it autonomously. The user retains the right to confirm or deny the suggestion. The AI's training data would come from official repositories (articles already correctly classified, validated rulings, BTIs, explanatory notes, recognized government databases), to limit the risk of relying on unreliable sources (for example, a simple blog or platform whose content would not be verified).
This precision is crucial, as product classification is the fundamental basis for determining the measures applicable to import and export, and for determining origin: it is essential to rely on high-quality content. R&D projects are therefore aimed at providing intelligent assistance to operators, rather than blind automation, to make the process safer and faster.
2.3. Multilingual regulatory watch
The international regulatory landscape is constantly changing. Updates may concern tariff barriers, export control requirements, product bans, or the obligation to provide specific documents. To cope with this complexity, we are working on algorithms to automatically search, collect and process textual information sources related to the subject of compliance, from official press releases, specialized publications and even professional social networks, in order to detect changes and new constraints for international trade.
This proactive monitoring would enable companies to receive targeted alerts as soon as a new regulation is promulgated or a modification is announced. This saves valuable time, enabling rapid adaptation of operations and ongoing compliance.
III. Data security and digital sovereignty
Data from customs processes is, by its very nature, highly strategic: it concerns import and export flows, the value of goods, their classification, the identity of suppliers and customers, and so on. They are also subject to strict RGPD (General Data Protection Regulation) obligations, which means knowing precisely where this information is stored, how it is processed and who has access to it.
In our R&D projects, we insist on opting for private AI instances, i.e. environments on sovereign infrastructures, rather than simply exploiting public APIs managed by third-party providers. This approach guarantees total control over the processing of sensitive data, and helps preserve digital sovereignty. It also avoids potential data transfers to countries whose legislation is less protective than that of the European Union.
In addition, the question of the quality of training data is particularly acute: for AI to produce reliable results, it must be fed from official, validated sources. A poorly trained algorithm, or one trained on approximate data, risks generating erroneous suggestions and jeopardizing corporate compliance.
IV. MyTower, Innovation Leader in International Trade
4.1. A comprehensive, adaptable offering
MyTower's TMS and GTM solutions are characterized by their modularity and scalability, giving companies global visibility of their commercial flows and the ability to manage their operations centrally. Through our R&D work, we enrich our solutions with advanced functionalities: to enable companies to navigate the changing and complex regulatory environment.
The ambition is to offer a range of tools enabling international trade players to adapt rapidly to legislative changes and limit errors, while respecting constraints linked to data protection and digital sovereignty. Each company can configure the platform according to its specific needs, whether for import, export or transit flows.
4.2. A partnership for sustainable co-innovation
Innovation is rarely achieved in isolation. At MyTower, we are convinced that co-innovation is the key to developing solutions that are truly adapted to the realities of the field. We are already working with partners, and therefore invite shippers, forwarding agents and all those involved in international trade to collaborate with us on these R&D projects.
By pooling feedback, comparing concrete needs with the latest technological advances, and ensuring the quality of training data, we can create a secure, high-performance digital ecosystem. The ultimate goal is to transform the challenges of international trade (multiplication of regulations, customs complexity, ESG traceability) into opportunities for growth and differentiation.
Conclusion
The integration of artificial intelligence into customs and export control processes is not just a technical evolution, but a strategic lever for mastering the complexities of international trade. At MyTower, we take a research and development approach to these issues, convinced that artificial intelligence, when properly managed, can bring significant added value to our customers.
All our initiatives are based on strong principles: compliance with RGPD, protection of strategic data, private AI instances and use of reliable sources for algorithm training.
We invite visionary companies to join us in this digital transformation, by participating with other customers our R&D projects or sharing specific needs and constraints. Together, we can design and refine solutions capable of meeting the challenges of increasingly complex international trade. Our ambition is to make MyTower a trusted partner for modern, secure customs management, where AI serves as a catalyst to optimize processes while preserving digital sovereignty and information quality.
MyTower - Innovation for international competitiveness.
