About half the PDFs moving through a freight back office — invoices, rate confirmations, bills of lading — are already digital text, not scans.
Firecrawl open-sourced pdf-inspector, a tool that checks each PDF in milliseconds and flags whether it's real text or a scan — before any OCR runs. The text-based files, which Firecrawl puts near 54% of documents, go straight to extraction; only the scanned pages take the slow OCR path. What comes out is clean, structured text with tables and reading order preserved, and it needs no GPU for the text-based files.
For document-heavy operations, the point isn't the tool — it's the habit. OCR shouldn't be the default. Sorting text PDFs from scans first, and only running OCR on what actually needs it, cuts the time and cost of turning freight paperwork into usable data across a day's invoices and BOLs.
Freight document processing: skip OCR on the roughly half of PDFs (invoices, BOLs, rate cons) that are already digital text.
https://www.firecrawl.dev/blog/fire-pdf-launch