HMRC Adds AI Commodity Classification to Online Trade Tariff

HM Revenue & Customs (HMRC) has introduced AI-enabled commodity classification into its Online Trade Tariff service, in work involving digital transformation consultancy Transform. The tool allows businesses to describe goods in everyday language and receive guidance on identifying the appropriate customs codes. Those codes help determine how goods are classified and establish applicable duties and requirements. The announcement describes a deployed government service—not a new solicitation or a disclosed contract award—and provides no contract value, contract number, or procurement contact details.

The service addresses a usability challenge for smaller companies and occasional traders. HMRC’s existing tariff service works well for experienced users, but can overwhelm newcomers with extensive information. At Think AI for Government 2026, Philippa Evans, HMRC’s head of tariff and quota policy implementation, described users arriving at the service and being “hit with a kitchen sink’s worth of data in the face.” The Online Trade Tariff has been available since 2012 and has around 550,000 regular users, handles approximately 35 million searches through its interface annually, and receives around 350 million API calls. These figures illustrate the scale and operational context in which the new capability has been introduced.

For GovCon suppliers, the deployment is market intelligence about applied AI in a high-volume public-facing service, not evidence of an open bidding opportunity. The announcement identifies Transform as supporting the implementation but does not specify its contract, delivery role, or commercial arrangement beyond that involvement. It also provides no information on a procurement timeline or future solicitation. Contractors should avoid treating the news as a direct sales lead or inferring award details that have not been disclosed; its value is as an example of a government problem-led AI implementation and a potential indicator of capabilities agencies may seek in similar services.

The reported project began with the service problem rather than a predetermined AI solution. Jose Puyana, Transform’s director of trade and growth, said, “We didn’t start with the question, how does AI fit here? That was not the question. The question was a problem.” After several years of work with HMRC to improve the tariff service, the effort included an eight-week AI lab, followed by 10 weeks of discovery, before development and production. The source also highlights the data and knowledge challenge: customs classification information was spread across approximately 15 separate sources, and some expertise resided with experienced staff rather than being formally documented. HMRC and Transform involved traders throughout the process to reflect the difficulties businesses actually encounter. The source says AI could potentially extend to customs valuation and rules of origin, but does not say those capabilities have been implemented.

• HMRC has deployed AI-enabled commodity classification in its Online Trade Tariff; the reported service lets users describe goods in everyday language and get code-identification guidance.
• The tariff codes inform customs classification and help traders establish relevant duties and requirements, making accurate, usable guidance important to business operations.
• The Online Trade Tariff dates to 2012 and has around 550,000 regular users, approximately 35 million annual interface searches, and around 350 million API calls.
• The service is intended to make complex customs information more accessible to smaller businesses and occasional traders, for whom classification mistakes can contribute to delayed shipments, unexpected costs, and potentially lost customers.
• Transform is identified as HMRC’s implementation partner, but the source does not detail its contract, scope, delivery role, or the value of any agreement.
• The work followed several years of service improvement activity and comprised an eight-week AI lab and 10 weeks of discovery before development and production.
• HMRC’s classification knowledge was distributed across approximately 15 separate sources, with some expertise held by staff and not formally documented—a relevant data-integration and knowledge-management challenge for suppliers.
• The source identifies customs valuation and rules of origin as possible future areas for AI support; it does not report that either capability is currently deployed.
• Treat this as an implemented-service signal for AI, user-centered design, and fragmented-data integration—not as an open solicitation: no contract value, number, procurement contact, or future procurement timeline is provided.