Transit documents

Transit document automation is a common issue among international freight forwarders. While T1/T2 formats are largely standardized, the documents can be lengthy and have minor variations between countries, complicating data extraction. Due to the volume of documents, manual or semi-automated processing is time-consuming and error-prone.

Common challenges

• Due to listing individual items, the documents can easily span hundreds of pages
• Minor variations in the format between countries
• Long reference sequences (e.g. MRN) need to be extracted exactly

Shortcoming of previous-generation solutions

• Need to split longer documents
• OCR issues surface in the presence of tables and long sequences (“1 v. l”, “O v. 0” problems)

Solution with Clerk

Clerk’s AI can natively handle longer documents without the need for explicit splitting. Once a transit document is recognized in the inbox or submitted to Clerk directly, it is automatically split into individual positions and the line items are extracted. Clerk then returns a list of individual positions with the respective amounts, which can be used in the downstream automated process. Its state-of-the-art AI leaves behind familiar OCR and template-based extraction issues.