Understand
Read and analyse a PDF
Five tools for documents that resist: a scan you cannot search, a report in a language you do not read, two versions of a contract that have to be set against each other, a file no reader will open any more, phone numbers to pull out without copying them by hand.
All tools
OCR governs everything else
A scan holds no text, only the photograph of a page. Until it has been through OCR it can be neither searched, nor translated, nor compared, nor converted to Word. OCR adds an invisible text layer exactly under the image: the appearance does not change by a pixel, but the document becomes selectable, searchable and readable by a screen reader. It is the first step of almost everything that follows.
Translating a document, not a text
Pasting a PDF into an online translator returns a wall of text: columns run together, tables collapse, captions end up in the middle of a paragraph. Here the page structure is recognised before translation, so columns, tables, figures and formulas stay where they are. It is also the one tool on Atelier Docs where every page calls a language model: the text goes out over the network, and that is stated before you drop anything.
Compare and repair: what the tools can really do
Compare sets the two versions against each other page by page and returns the newer one highlighted, with a summary page for what disappeared. It answers “what was written differently”, not “what moved”: a block that shifted or an image that changed is not flagged. Repair rebuilds the object table of a damaged file: an interrupted transfer, a failing disk, a botched export. What was never written cannot be invented, but the structure often can be mended.