In logistics, misplaced packages and missing receipts are more than just a nuisance—they're operational bottlenecks.
For Transexpress, I recently built a tool that uses image input to help identify packages even when a receipt or reference number is missing.
It’s a lightweight, browser-based utility that lets staff upload or snap a photo, then get a list of matching items based on visual data. The working prototype is live here:
The idea was to make something useful without introducing complexity. A focused workflow for a specific problem. And since the model improves over time, it’ll only get more accurate as more data comes in.
Projects like this remind me how applied AI doesn’t always need to be flashy to be meaningful. Sometimes, it's about making the right thing simple.