Al Munasiq and AI-Driven Classification at Dubai Customs
18 November 2026
By Juma Al Ghaith, Chief AI Officer, Dubai CustomsAccurate goods classification is essential to the confidence on which international trade depends. It helps businesses understand the duties and regulatory requirements that apply to their goods, while enabling Customs to collect revenue correctly, administer trade measures and produce reliable statistics. Dubai Customs’ Al Munasiq Platform applies artificial intelligence to this task, helping traders and officers identify the appropriate Harmonized System (HS) codes. Since its introduction in 2019, the platform has developed into a service that combines classification guidance with tariff information and supervised model improvement.
Al Munasiq gives practical effect to Dubai Customs’ 2026–2030 strategy and its philosophy of “Enabling Trusted Trade”. More accessible classification guidance helps businesses prepare declarations with greater confidence and understand compliance requirements earlier in the trading process. For the Department, consistent classification supports revenue protection and more informed risk assessment. These functions connect the platform directly to the strategic priorities of facilitating trade, strengthening security and compliance, securing revenues and elevating stakeholder experience.
Generative AI for Goods Classification
Al Munasiq uses generative AI to interpret product information and suggest relevant tariff codes. Traders describe goods in different ways, often using commercial language that does not correspond directly to the terminology of the tariff. The platform helps them navigate this complexity by analysing the information supplied and presenting classification suggestions for consideration.
The name “Al Munasiq”, meaning the Harmonizer or Aligner, reflects this purpose. By making classification guidance easier to obtain and use, the platform supports the preparation of goods declarations and assists officers in their assessment of classification issues.
The platform comprises two connected components:
- Al Munasiq Apps allow traders and customs brokers to submit product information through text, voice or images. Alongside suggested codes, users can access information on duty rates, prohibitions, restrictions and potential benefits under trade agreements. Bringing this information into one service helps businesses assess the requirements relevant to their goods before submitting a declaration.
- Al Munasiq Backend provides Dubai Customs with an environment for supervised model training, performance monitoring and classification risk assessment. Feedback supports subsequent training and refinement, with customs expertise guiding the evaluation of outputs and the development of the model.
AI Governance and Professional Accountability
Dubai Customs has incorporated data governance, traceability and human validation into the platform’s design. Classification outputs can be reviewed by officers and retained within an audit trail, supporting accountability in how the service is used. Human oversight remains central to final classification decisions, while regular performance reviews help identify weaknesses and guide improvements. This governance approach allows the Department to develop the service while retaining professional responsibility for decisions with regulatory and financial consequences.
An application programming interface (API) enables internal systems and external platforms to access classification suggestions and tariff information. It also provides a channel for structured feedback to support supervised model improvement. This architecture makes classification assistance available within other digital services and creates opportunities to reduce repeated information entry across the trader’s journey.
Regional Interoperability and International Standards
Al Munasiq’s architecture creates scope for future integration with national single windows, regional customs platforms and Gulf Cooperation Council (GCC) initiatives. Such integration would depend on agreed data standards and the requirements of participating administrations. It offers a practical avenue for cooperation on classification information, supporting more consistent interpretation and a better connected experience for businesses trading across borders.
The approach is consistent with the Revised Kyoto Convention’s emphasis on simpler, more predictable customs procedures. Better classification data can also support the risk-management objectives of the World Customs Organization’s (WCO) SAFE Framework of Standards to Secure and Facilitate Global Trade. As administrations prepare for HS 2028, maintaining current tariff content and validated classification guidance will remain essential to the usefulness of digital classification tools.
Platform Development and Technical Evolution
Development began in 2019 with a machine-learning application available on Android. Subsequent releases expanded access, introduced generative AI and added support for more detailed tariff codes. Each stage addressed a practical requirement, from improving the user interface to accommodating changes in classification detail and making the service available across mobile platforms.
| Version | Date | Key Features and Milestones |
| 1.0 | 2019 | Initial release using machine-learning models, available on Android. |
| 2.0 | January 2025 | Redesigned interface and integration of generative AI. Access through mobile web, Google Play, Apple App Store and Huawei AppGallery. |
| 2.2 | August 2025 | Support for 12-digit tariff codes at GCC and national levels, with mapping from eight-digit codes. |
| 3.0 | October 2025 | Redesigned user interface and user experience. Adaptive generative AI using a cloud-based model to support scalability. |
| 3.1 | June 2026 | Enhanced security, performance, reporting and overall user experience. |
The development team benchmarked Al Munasiq against publicly available government and commercial classification platforms. Hands-on testing, product demonstrations and reviews of vendor documentation informed the assessment. These comparisons helped identify opportunities to improve usability and technical capability, while providing practical lessons for administrations considering similar services.
Version 2.0 introduced generative AI, followed by an adaptive generative AI model and redesigned interface in Version 3.0. Version 3.1, released in June 2026, strengthened security, performance, reporting and usability through secure temporary share links, endpoint protection and rate limiting, Cognos BI report access, improved keyword and commodity search, HS Code copy functions, enhanced spell check, and expanded logging and traceability. Together, these releases improved the platform’s ability to interpret varied product descriptions while supporting secure, scalable operation and continued refinement through supervised learning.
Support for 12-digit tariff codes accommodates the additional classification detail used at GCC and national levels. The internationally harmonized HS structure remains at six digits; regional and national subdivisions provide further detail beyond that level. Mapping between eight- and 12-digit codes helps users navigate this transition and identify the more specific tariff entries applicable to their goods.
Service Uptake and Operational Results
Operational figures covering 2025 to July 2026 show substantial uptake of the AI Goods Classification service and improved completion of classification requests. They provide evidence of how extensively the service is being used and its ability to respond to users’ requests.
By July 2026, the AI Goods Classification system had processed approximately 488,500 requests, compared with approximately 52,800 requests through the traditional Online Goods Classification system. The AI-enabled service therefore handled more than nine times the request volume, equivalent to approximately 825% more requests.
The completion rate for classification requests reached 100%, compared with 67% for the previous system, an improvement of 33 percentage points. This measures the fulfilment of service requests; the accuracy of the classification suggestions is assessed separately through model-performance reviews and professional validation.
Early internal assessments also indicate improvements in classification consistency and accuracy, with fewer discrepancies and a reduced need for manual review. More consistent guidance can help traders prepare declarations and allow officers to devote greater attention to complex cases and classification risks.
Enabling Trusted Trade Through Classification Certainty
Al Munasiq supports Dubai Customs’ role as an economic enabler by addressing a technical issue with direct consequences for business. Clearer classification guidance helps traders anticipate applicable requirements and make better informed decisions. For Customs, more consistent classification strengthens the basis for revenue collection and regulatory controls. The strategic value of the platform therefore rests on the quality of the guidance it provides and the confidence with which that guidance can be used.
The experience also illustrates the importance of developing technology and customs expertise together. Accessible applications, supervised model improvement and accountable decision-making each contribute to a reliable service. Continued assessment of classification accuracy, processing times and the need for manual intervention will help establish the platform’s longer-term operational and economic contribution.
Dubai Customs welcomes collaboration with the World Customs Organization (WCO) and member administrations on AI-assisted classification. The Department is prepared to share its methodology, implementation experience and lessons through relevant working groups and technical meetings. This exchange can support the responsible development of classification services and strengthen the use of shared standards across the international trading system.