PROJECT & FUNDING SCHEMES

Problem Statement:
According to latest statistics, at least 90%of Hong Kong's ~8,000 logistics SMEs remain non-digitalized. These SMEs face sector-specific barriers including cloud connectivity and real-time integration costs of $200K-$500K minimum. This systematically excludes small SMEs from international shipping contracts that require multiple cloud service connections for online shipment booking freight management, warehouse coordination, and real-time status updates.
With 75% of surveyed companies still using paper-based documents, logistics SMEs struggle to manage large volumes of customs and trade documents from various sources, making customs filing and trade financing workflows highly inefficient.
Research Approach:
This applied research project addresses this industry transformation challenge using an innovative approach originally outlined by Transport and Logistics Bureau in a consultancy study awarded to LSCM. Instead of custom integration for each SME individually, we build a unified and scalable onboarding web service that serves all SMEs
for common SaaS use cases.
We applied the latest AI methods to solve inherent data standardization and integration problems through a modular, step-by-step workflow. Our working prototype successfully handles common data and schema issues, cleaning and standardizing data for cloud booking submissions with two SaaS companies （LogFlows and AutoToll）. To ensure accuracy when using large language models, we incorporate cutting-edge AI data issue detection, agentic tool use, and hybrid LLM/SML schema matching and conformal uncertainty estimation techniques.Applying the same automatic pipeline technology to customs declaration automation, we extend our HS Code seed project to cover additional trade products coverage for 2x faster pipeline generation by using declaration elements to automate pipeline verification.
A third forward looking application of the technology is to provide a trusted data service pipeline for cargo insurance and financing. These are the most difficult workflows anticipated by TLB for SME participation in the new logistics B2B ecosystem.