How FRB works

PDF order processing with review before export

FRB supports purchase order data extraction from PDFs, collecting line items, quantities and a reference into a draft. PDF order processing also involves checking the result against the original: a readable document is not, by itself, a correct or approved order.

Text PDFs and scanned documents

PDFs containing text are read directly. Scanned pages with legible printed text can be read using local optical character recognition (OCR). The order parser uses rules and code; documents are not sent to OpenAI or another external AI model for interpretation.

Unclear print, handwriting, ambiguous columns and unfamiliar layouts may need manual handling. Scanned product codes and quantities need particularly careful checking. FRB does not promise to interpret every PDF design correctly.

Size, units and review

The maximum file size is 10 MB. PDF reading is limited to 100 pages, including up to 20 scanned pages within the processing time limit. Damaged files and exceeded limits produce a controlled error or warning; the document is not automatically treated as a correct order.

Explicit weight and length units can be handled when the quantity and product are unambiguous. FRB must not guess how many kilograms are in a sack or how many metres are on a roll. Unclear information needs clarification.

Export reviewed orders as CSV, Excel or XML

Approved orders can be exported as CSV, Excel or FRB XML. Excel provides a grouped order overview, while FRB XML uses FRB’s own documented format. Standard CSV includes the order ID, reference, customer email, SKU, product, quantity and approval time; prices, discounts and totals are not included.

FRB supports configurable export profiles, but the destination system’s import format must be tested before use. A downloaded CSV does not mean the order has been registered in the ERP. The customer checks the import in their own system.