Documents read more accurately
We spent four weeks on the accuracy of the extraction pipeline. You will notice most of it only because some annoying errors have stopped happening.
Classification
- The AI now gets the filename as a weighted hint when it classifies a document. It also gets a list of the abbreviations exporters use: FCT/FAC/FE for invoices, PL for packing lists, BL/AWB for transport documents, FITO for phytosanitary certificates, EUR1/CO for certificates of origin. Previously the AI classified documents without seeing the filename.
- Courier waybills (DHL, FedEx, UPS) are a document type in their own right, so they are no longer filed under the wrong type.
- Admins can recategorise a document from a dropdown next to its type badge. Trackberry re-extracts it as the chosen type and cleans up whatever the old type imported.
Spreadsheets and PDFs
- Hidden columns in Excel packing lists are now excluded. When the same header appears twice, the visible column is used.
- More Spanish headers recognised: Calidad as category, Envase as format/box weight, abbreviated Calib. as calibre. Category values no longer keep their “Category”/“Cat”/“Categoría” prefixes.
- A padded summary sheet no longer takes priority over the detail sheet.
- When a packing list’s header row has no ruled lines, the column names are now rebuilt from the layout of the header itself. Before, all 18 labels could collapse into one cell and leave the AI to guess which number was the calibre.
- Alphanumeric calibres (1X, 2X) now come through extraction unchanged.
Counts, weights and pallets
- Box counts are checked against the declared totals. When they disagree, the check fails visibly instead of the count being silently doubled.
- Missing box weights are derived from net weight ÷ box count. For asparagus packed in “atados” (tied bundles), they come from tie count × weight per tie.
- Pallets sort by number instead of as text, so 10 comes after 9.
References
- Hong Kong phone numbers have the same format as IATA air waybill numbers, and one had been filed as a shipment’s AWB. Candidate numbers now have to pass the IATA check-digit test, and a valid AWB from an attached waybill takes priority over one taken from an email signature.
- AWB numbers that differ only in their separators (“160-14832020” vs “16014832020”) are now matched as the same reference, so they no longer create duplicate shipments.
Alongside this, the Needs Assignment picker is now searchable and grouped into Sea and Air. Each shipment in it is labelled by its container or transport reference rather than an internal SHIP- code.
If a document still reads wrong, Troubleshooting covers what to do about it.
FAQ
Do I need to re-upload documents that were read wrong before this?
Only the ones that matter. An admin can recategorise a document from the dropdown next to its type badge, which re-extracts it and cleans up whatever the previous type imported.
Why does a box count now fail a check instead of being corrected?
A count that disagrees with the declared total means one of the two readings is wrong. Guessing which one would occasionally double a shipment’s box count without anyone noticing. A failed check shows the problem to a person instead.
Does Trackberry only read English documents?
No. Spanish packing list headers are recognised directly, and free-text values in any language are understood — the extraction works from what a document means rather than from a fixed template.