PDF tools suite: convert, merge, split and images

Anonymization is Anonimatum’s core, but teams stumble earlier: the case file arrives mixed (phone photos, Word, native PDFs, scans, 80-page annexes). If you anonymize too early or in the wrong format, you redo work and multiply the risk of leaving PII in an annex you «were no longer looking at».

This guide proposes an operational order for the PDF tools hub — convert, merge, split, images to PDF, OCR — and finishing with purposeful anonymization.

The most expensive sequencing mistake

Anonymizing each loose file and *then* merging seems logical… until the merged PDF reintroduces metadata, wrong page order or an untreated annex. The reverse also fails: merging everything and anonymizing without splitting means an error on page 2 forces reprocessing 200 pages.

Hub map (which tool for what)

Playbook: typical administrative case file

• Input: 12 HEIC photos + 1 Word report + 1 scanned 40-page PDF + 1 native resolution PDF.
• Step 1: images → PDF (group by logical annex).
• Step 2: OCR / anonymize the scan; native Word anonymization if it must stay editable.
• Step 3: split the long PDF if only 5 pages are publishable.
• Step 4: merge the final public version in index order.
• Step 5: global verification (identifier search + visual sampling).

Merge and split with a compliance mindset

Split applies different policies (publishable vs reserved) and reduces the blast radius of a mistake. Merge serves delivery and archive, but should not be the first step if PII is untreated. In both cases, document page order: in transparency and courts, the index matters.

Images to PDF: the mobile link

Staff capture minutes, boards or IDs with phones. Those images carry EXIF metadata (sometimes location) and uneven quality. Move them to PDF in a controlled way, check orientation/dpi, *then* decide OCR + anonymization. Do not send the original album «because it is already in the chat».

GDPR criteria when using converters

• Each conversion is processing: minimize who can launch it.
• Prefer EU processing without sending content to external LLMs.
• Output format alone does not «clean» PII: convert ≠ anonymize.
• Keep traceability: which tool, which version, which operator.

How this fits anonymization

The hub does not replace the anonymizer: it feeds it. Practical rule: *normalize → structure (merge/split) → anonymize → verify → deliver*. Anonimatum concentrates those pieces in one product so you do not jump across generic conversion sites with opaque retention policies.