Andre Klynsmith

Technology and Trends

Classify at the Point of Booking: AI Item Identification Arrives in South African Express

DHL has deployed artificial-intelligence-based item identification inside its customer-facing international express booking system in South Africa. The shipper photographs the item, and the system proposes the customs classification at the moment the shipment data is entered rather than after it reaches a clearing agent.

Technology and Trends

31 August 2026 • 4 min read

Most customs delay is not created at the border. It is created weeks earlier, in a commodity description someone typed in a hurry. Moving classification to the point of booking is an attempt to fix the problem where it starts.

What the Tool Actually Does

The feature sits in the booking flow for international express shipments. Instead of asking the shipper to describe the contents in free text and hoping the description survives a customs risk engine, the system reads an image of the item and proposes a classification. The stated benefit is straightforward: accurate classification at the point of data entry reduces holds and speeds up clearance.

The capability is live in several countries, South Africa among them, with further rollout planned during 2026. It is a narrow tool aimed at a narrow problem, which is part of why it is interesting — it does not require the shipper to learn a tariff schedule, and it does not require the carrier to rebuild its clearance stack.

Why Classification Is Where Delay Is Manufactured

Vague or wrong commodity descriptions are among the most reliable triggers for customs holds and physical inspections. Once a shipment is flagged, the cost is not the duty — it is dwell, storage, a missed onward connection and a customer who has to be told something. None of that is recoverable by working harder at the border.

Africa's customs authorities are moving in the same direction from the other side. The South African Revenue Service is running a Customs Modernisation Programme with artificial-intelligence-enabled risk detection under its Modernisation 3.0 strategy, and in 2025 issued a request for information for an AI-driven digital-twinning system. The Border Management Authority and SARS are also coordinating on high-risk cargo inspections rather than duplicating them.

The Same Logic Applies to Road Freight Paperwork

If both the carrier and the revenue authority are using machine classification, the cost of a sloppy declaration rises rather than falls. A risk engine that has been trained on good data does not become more forgiving of a vague entry — it becomes better at spotting one. The shipper who describes a consignment as “machine parts” is not saving time any more.

For SADC road freight, the transferable lesson is that data quality at origin is now an operational control, not an administrative afterthought. Commercial invoices, packing lists and commodity descriptions prepared properly at loading are what keep a truck moving through Beitbridge, Kasumbalesa or Chirundu. The technology is arriving in express first because express has the volume to justify it, but the discipline it enforces is available to any operator today, free of charge.

About the Author
	Andre Klynsmith Entrepreneur & Technology Enthusiast Andre's expertise spans technology, blockchain, transport, and logistics — with a passion for innovation and driving positive change across industries. info@laaimylorrie.com | laaimylorrie.com