All logistics case studies

DHL Aruba: BOL and invoice automation

Invoice & Bill of Lading automation

A look at the document-processing platform built by Esnaj for DHL.
DHL logo Aruba

For DHL Aruba, Esnaj replaced manual Bill of Lading (BOL) and invoice entry in an older internal system with a custom document automation platform using image preparation, OCR and AI. It processes around 1,000 documents a day in approximately 4 minutes per document, including human review, versus 15–20 minutes previously. Extraction accuracy exceeds 96% across extracted fields. Staff resolve flagged fields, keep reviewed data in the platform and export it when needed.

documents processed per day
~1,000
per document, including human review
~4 min
accuracy across extracted fields
96%+
year in production
~1

Every document meant another round of manual entry

Staff read each BOL and invoice, then copied its fields into an older internal system. Processing took approximately 15–20 minutes per document. Reading, typing and checking the same information kept people tied to data entry instead of other work.

Why a fixed template was not enough

A fixed document template was not enough. Supplier formats varied, equivalent fields had different names, and some scans arrived tilted or incorrectly oriented. Handling that variation with deterministic rules alone was difficult. The platform combines image preparation, OCR and AI to accommodate more of those cases.

  • Correct document angle and orientation before extraction.
  • Interpret equivalent fields with different supplier labels.
  • Flag problematic fields so a person can correct them or decide what to do.

How we process BOLs and invoices with OCR and AI

One platform, from upload to reviewed data. Automation handles extraction; people retain the final decision.

  1. Upload BOLs and invoices

    Staff upload documents directly to the platform that replaced the older system.

  2. Prepare and extract

    Image preparation corrects orientation. OCR and AI extract and interpret fields across different supplier formats.

  3. Review and resolve

    Staff verify the extracted information. The system highlights problematic fields for correction or a user decision.

  4. Store or export

    Reviewed data normally stays in the platform. Staff can export it when needed.

From 15–20 minutes to around 4

Approximately 73–80% less time per document

From 15–20 minutes to around 4
What changed Before With the platform
Time per document 15–20 min ~4 min
Staff task Read and type each field Upload, verify and correct
Working system Older internal system Replacement document platform
Document handling Manual field-by-field entry Image preparation, OCR and AI

Calculated from the approximate processing times: 11–16 minutes saved per document. The current ~4 minutes include human review; this is not an automation-only runtime.

Less typing. People still make the decisions.

The objective was not to remove staff from the workflow. It was to remove repetitive entry. People verify the data and decide how to handle flagged fields, while the reduction in manual work frees time for other tasks.

Questions about the DHL implementation

What documents does the DHL platform process?

The platform processes Bills of Lading (BOLs) and invoices. It combines image preparation, OCR and AI to handle different supplier layouts and field names, including documents that need angle or orientation correction.

How much time does processing take, including review?

Processing takes approximately 4 minutes per document, including human review, compared with approximately 15–20 minutes previously. That represents an estimated 73–80% reduction based on those approximate times.

What does 96%+ extraction accuracy mean?

Esnaj reports extraction accuracy above 96% across extracted fields. This is a field-level figure, not a claim that more than 96% of documents need no correction. The system flags problematic fields for staff to review, correct or decide on.

Did Esnaj integrate with or replace the older system?

Esnaj replaced the older internal system. Staff upload and review documents in the new platform, where data is normally stored and can also be exported. The solution processes around 1,000 documents per day and has been in production for approximately one year.

Still entering freight documents by hand?

Tell us about your documents, supplier formats and review process. We can discuss whether a custom workflow fits your operation.

Discuss your document workflow