Archives, document management systems and some e-filing portals accept only PDF/A: a PDF with every font embedded
and nothing that depends on the viewer's computer. Saving as PDF/A by hand works for one file. For a folder of
contracts, tables and slides you want it done by an API. Every output below is from a real run on 7 November 2026.

Disclosure: I built the tool used here, Office to PDF Converter on Apify.
It runs LibreOffice in a container. The source code is open:
AndriyVK/-apify-document-tools.

One call, several files 

pip install apify-client 
from apify_client import ApifyClient client = ApifyClient("<YOUR_APIFY_TOKEN>") run = client.actor("ai_admin/office-to-pdf").call(run_input={ "sources": [ {"url": "https://api.apify.com/v2/key-value-stores/FloomuI2DrQCJazY8/records/doc-uk.docx"}, {"url": "https://api.apify.com/v2/key-value-stores/FloomuI2DrQCJazY8/records/sheet-uk.xlsx"}, ], "pdfA": True, # PDF/A-2b instead of a plain PDF  "mergeIntoOne": True, # also one merged.pdf in input order (free) }) for item in client.dataset(run["defaultDatasetId"]).iterate_items(): print(item.get("fileName", "merged"), item["pages"], "pages", item["pdfUrl"])

The two sample files are public, so the code runs as is. Formats: DOCX, DOC, ODT, RTF, TXT, HTML, XLSX, XLS, ODS,
CSV, PPTX, PPT, ODP and images.

The real output 

doc-uk.docx 1 page 44,040 bytes pdfA: true sheet-uk.xlsx 2 pages 32,574 bytes pdfA: true merged.pdf 3 pages 34,927 bytes documents: 2

What I checked in the PDF

A file that says «PDF/A» is not always one, so I opened the Word result:

  • The PDF/A label is there: the XMP metadata says pdfaid:part 2, pdfaid:conformance B - PDF/A-2b.
  • All fonts are embedded (pdffonts: emb yes for every font). This is what PDF/A is about: the file looks the same on any computer, in ten years too.
  • The Cyrillic text is real text, not a picture: pdftotext gives back «Звіт про постачання», «Склад у Львові отримав 1240 палет…» - so the archive's search finds it.
  • Fonts were replaced. The document asks for Calibri and Cambria, which are Microsoft fonts. The converter uses Carlito and Caladea instead - free fonts with the same letter widths, so lines and pages break in the same places. The look is close but not identical.

What I did not do: a full validation with veraPDF, the reference PDF/A checker. If a portal
rejects files that fail veraPDF, run it on a sample before you convert the whole folder.

Price

$10 per 1,000 documents, whatever the number of pages; merging is free; a file that fails to convert is not
charged. The run above cost two cents.

Limits

  • Password-protected files are not converted.
  • Fonts that are not installed are replaced by metric-compatible ones; an unusual font may change the look slightly.
  • Macros are not run.
  • PDF/A is about how a file is stored. It does not sign it: if a court or an agency needs a qualified electronic signature, sign the PDF/A afterwards with your own signing tool.

From an AI agent

The converter is also an MCP tool: https://mcp.apify.com?tools=ai_admin/office-to-pdf, so Claude, Cursor or any MCP
client can turn the files it has prepared into PDF/A.

Originally published at andriyvk.github.io.