FAQ

What file formats and languages does OrderPilot support?

Complete list of supported input formats, languages, and known edge cases - plus what happens when a format falls outside the supported set.

Published 14 April 2026 · 2 min read formats compatibility

OrderPilot is designed to handle the messy reality of supplier PO flows, not a clean happy-path. Here’s what that means in practice.

Supported file formats

FormatSupportedNotes
PDF (text)YesNative text extraction, fastest
PDF (scanned)YesOCR path, slightly slower
PNG / JPGYesTreated as scanned docs
TIFFYesMulti-page supported
XLSX / CSVYesStructured tabular format
EDIFACT / X12YesVia separate EDI module
DOCXYesConverted to PDF internally
HTML email bodiesYesPlain-text and HTML both parsed
XML (UBL, cXML)YesNative structured parse

Languages

Document extraction works across 20+ languages including all major European languages. Our European customer base uses OrderPilot daily in:

  • English, Dutch, German, French
  • Spanish, Italian, Portuguese
  • Polish, Czech, Hungarian
  • Danish, Swedish, Norwegian, Finnish

We auto-detect the language per document - no config needed.

Handwritten content

Any handwritten fields are flagged for manual review by default. Printed-and-signed POs extract normally; the signature doesn’t confuse the model.

Edge cases worth knowing about

Password-protected PDFs. We don’t break passwords. If a supplier sends protected PDFs, either strip the password upstream or ask them to disable it.

Extremely large PDFs (>500 pages). Split them first. PDFs that large are almost always multiple POs in one file - we flag them in the queue.

Legacy EDIFACT variants. If your supplier sends a non-standard EDIFACT profile (custom segment order, proprietary tags), we may need a 30-minute mapping session. This is rare.

Faxed POs. Fax is a scanned image. Works fine.

What happens when a format isn’t supported

The document lands in the Unsupported formats queue with a clear reason. You can then:

  • Convert it upstream and re-submit
  • Send a sample to support for format evaluation
  • Skip it (we’ll exclude it from accuracy reports)

We support adding new formats for customers with a consistent volume; ask during onboarding if you have something unusual.

Related articles