Trade & Logistics AX Cases

Challenge

Maritime transport exposes shippers to weather, routing, GPS-coverage, and voyage-distance uncertainty; before Cargo Tracking, arrival prediction accuracy stood at 40%. That uncertainty forces inventory buffers when late arrivals can disrupt production or fulfillment. Repetitive manual work also persists across warehouse, transport, parcel, forwarding, and overseas operations, while online sellers face slow system links with logistics providers. New packaging rules add another process burden as operators judge over-packaging exceptions, merged parcels, and reusable materials.

AX Intervention

CJ Logistics' TES Logistics Technology Institute builds AI Cargo Tracking to analyze voyage information, route, weather, GPS-signal availability, and vessel distance for arrival prediction. It deploys RPA software robots across operations to absorb repetitive manual tasks. e-FLEXs automates online seller-logistics integration and uses AI and big data to forecast order patterns for fulfillment labor planning. TES also productizes PackCheck for AI-based over-packaging diagnosis and extends computer-vision robotics, including depalletizer arms and humanoid packaging trials.

Outcome

Cargo Tracking raises maritime cargo arrival prediction accuracy from 40% to 85%. CJ Logistics states that lowering safety stock by 30~40% versus before through improved arrival prediction makes it possible to expect lower storage-related logistics costs, manufacturing schedule planning, and prevention of overproduction.

RPA saves 5,600 hours, with about approx. 30K hours additional hours expected. System integration time between online shopping malls and the logistics provider falls to 1/10. AsiaToday reports that CJ Logistics has deployed PackCheck across 26 centers logistics centers nationwide, and NLIC reports that its vision AI depalletizer processes 700 boxes per hour while lifting boxes up to 20 kg.

Challenge

E-fulfillment warehouses combine small orders, broad SKU assortments, and high pick density, which lengthens picker walking time. Legacy WMS, ERP, handheld scanners, and paper pick lists make route and labor allocation harder during demand spikes. The operating constraint is not only automation capital; it is the daily orchestration of people, locations, and urgent orders.

AX Intervention

DHL Supply Chain deploys IDEA as plug-and-play pick-routing software on top of existing warehouse systems. The algorithm clusters orders by location, optimizes walking paths, and allocates tasks against workload and shipment urgency. In parallel, DHL says smart glasses were piloted in 2015 before becoming a standard in its warehouse operations. Vision Picking puts pick instructions, item location, cart placement, and barcode confirmation into the worker's field of view for hands-free execution.

Outcome

In its first commercial deployments, DHL says IDEA reduces warehouse employee travel distance by up to 50% and increases productivity at individual DHL locations by up to 30%. Supply Chain Dive corroborates up to 50% lower travel distance and up to 30% higher productivity, while noting that results vary by location depending on existing WMS capabilities.

Vision Picking international trials show an average productivity improvement of 15 percent and higher accuracy rates. Completed international AR smart-glass trials cut onboarding and training time by 50%.

Challenge

Small exporters often rely on trade fairs, cold calls, and fragmented buyer lists to open new markets. In data-poor regions, they cannot easily tell which overseas buyers already import similar products or are ready to source from Korea. KOTRA frames the problem as a targeting gap: demand exists, but the next credible buyer is hard to rank before sales effort begins.

AX Intervention

KOTRA runs an AI import-probability matching pilot through TriBIG. The model analyzes import/export records across 21 countries and 7.7M records, then predicts overseas buyers with 70%+ import probability for matching with Korean companies. TriBIG broadens the data layer with overseas-office data from 83 countries and 128 locations, totaling approx. 150M records records. KOTRA says its self-collected overseas-company database exceeds 10M+ records and will keep expanding through external data links. The workflow shifts sales teams from open-ended prospecting to scored lists, then concentrates human support on buyers that show demand.

Outcome

Reports say KOTRA's Nairobi office uses the import-probability model to market to buyers above 70%+, reaches 1,100 buyers candidates, focuses export support on 87 companies interested buyers, and closes deals with 5 companies local buyers totaling $1.45M. Dailian reports that the same pilot recommends a buyer with no prior Korean imports to polypropylene exporter K and wins an export contract worth about $900K. Additional TriBIG cases show paint maker A finding a Peruvian buyer and completing an initial export of $30K, while manufacturer M identifies the U.S. as a promising market and signs $83,415.

Challenge

Cold-chain exporters need condition proof before cargo reaches the destination, because temperature or power exceptions can surface too late for recovery. Maersk states that 13% of global fresh produce is lost in the cold supply chain each year. Before RCM data was available, Food Logistics reports that PTI time per reefer ran up to six hours, making blanket physical inspection a utilization constraint.

AX Intervention

A.P. Moller - Maersk turns reefer containers into connected assets through remote container management. Maersk Line said RCM devices combining GPS, a modem, and a SIM card would be on all 270,000 of its refrigerated containers. Food Logistics reports that about 400 vessels were outfitted with VSAT domes to relay reefer sensor data to satellite. The operating redesign uses telemetry before terminal arrival to send reefers for PTI only when needed. By 2019, Maersk says over 3,600 companies had signed up since the 2017 customer launch, with visibility over more than 380,000 combined-fleet reefers.

Outcome

Food Logistics reports that Maersk Line eliminated 225,000 PTIs in a single year by using RCM data to determine container condition on arrival. The same report says PTI time fell from up to six hours to about 12 minutes because data was available before the reefer reached the terminal.

The visibility base also expands from the initial 270,000 Maersk reefers to more than 380,000 combined-fleet reefers. Maersk later says more than 80% of its reefer fleet was upgraded during 2022 to deliver hourly data to Captain Peter, compared with previous daily updates.

Challenge

Importers and exporters need faster exception handling because port disruptions, strikes, conflict, and weather events can turn logistics issues into production and sales risk. Traditional forwarding workflows leave quotes, bookings, tracking, and exception response fragmented across email and phone. Samsung SDS says logistics-risk response planning previously took 1 day, making speed the operating problem alongside freight visibility.

AX Intervention

Samsung SDS builds Cello Square as a digital forwarding platform for online quoting, booking, tracking, settlement, and risk monitoring. Machine learning extracts logistics risks from more than 60K news items global news items collected daily. Generative AI then turns those signals into a risk grade using a model trained on about 20K cases past global logistics-risk cases. DigitalDaily reports that Cello Square performs 70 cases daily AI risk-grading jobs, and BI KOREA reports monitoring coverage across 36 countries and 56 locations. Samsung SDS also adds predictive ETA and ETD, freight-rate forecasts, SCFI market commentary generation, and conversational quote and container-sizing services to shift the platform from transaction portal toward control tower.

Outcome

Samsung SDS says the ML and generative-AI risk workflow cuts response-plan formulation from 1 day to 2 hours. The same official release says ML extracts risks from more than 60K news items daily global news items, while the generative-AI model is trained on about 20K cases historical risk cases. DigitalDaily reports 70 cases daily risk-grading jobs on the platform.

Platform adoption also scales. Samsung SDS says subscribed customers grew 27% year over year to 24,625 companies active companies at end-2025. NewsPim reports Q4 2025 Cello Square revenue of ₩330.7B, up 27% quarter over quarter, but attributes the increase to expanded digital services including warehouse and inland transportation.

Challenge

UPS faces a daily route-sequencing problem where small driving inefficiencies compound across a dense parcel network. The operating problem is not only shortest-path math; delivery rules, safety constraints, address quality, driver habits, and pickup commitments all have to fit the route. UPS's Edelman-verified paper estimates that full deployment could reduce annual driving by 100 million miles and produce $300 million to $400 million in annual driver-cost avoidance and fuel savings, making routing quality a direct P&L lever.

AX Intervention

UPS builds ORION, On-Road Integrated Optimization and Navigation, as a route-optimization system for daily delivery sequencing. ORION processes more than 250 million address data points to generate optimized routes for drivers. As of December 2015, INFORMS reports ORION was used by more than 35,000 of UPS's 55,000 U.S. drivers. UPS later extends the ORION map layer into UPSNav, which the company says runs on proprietary maps containing 250 million locations. In 2021, UPS said dynamic ORION enabled an average two to four miles route reduction per driver, on top of the original ORION's eight miles reduction.

Outcome

INFORMS ORMS Today reports that ORION saved UPS $320 million cumulatively by the end of 2015. Once fully deployed, ORION was expected to deliver $300 million to $400 million in annual savings and cost avoidance. The same source describes the projected annual operating impact as 100 million miles fewer miles driven, 10 million gallons of fuel saved, and 100,000 metric tons lower carbon emissions.

The UPS-authored Interfaces paper states that ORION cost more than $295 million to build and deploy, against expected annual savings of $300–$400 million. ORMS Today also cites a driver who reported daily stops rising from 110 to 115 to 118 in the same working time after adopting ORION routes.