LUMUT, PERAK — 22 July 2026 — Quiver Dive Team Lumut Sdn. Bhd. has begun a pilot covering two areas of its AI transformation programme: intelligent document capture and predictive arrival and capacity information.
On the documentation side, the pilot parses commercial invoices, packing lists, bills of lading and customs declarations into structured fields, with confidence scores attached to each field. Candidate HS codes are produced with supporting reasoning and reviewed by a classification officer before any declaration is lodged. Cross-document consistency checks run before the file is submitted, which is where most border delays originate.
On the forecasting side, the pilot replaces single-point arrival estimates with ranges that update as milestone events are received, and adds an early warning when a leg drifts outside its expected window. Equipment and slot availability outlook is produced per lane to support booking decisions.
The company noted that milestone data is being standardised ahead of wider forecasting use, because forecast quality depends more on event consistency than on model selection. Every automated output in the pilot requires officer confirmation, and results are reported per scenario against an evaluation set rather than as a single averaged score.
The pilot runs with a limited group of regular customers and will be extended only after review of accuracy, handling effort and exception rates.