PDF

The PDF App opens any PDF in your Dataspace as a first-class entity. Read, annotate, and let the AI extract structured information from the document.

What you get#

• Native PDF viewer — fast, zoom-friendly, page navigation
• Annotation — highlights, underlines, comments, drawings, all multiplayer
• Full-text indexing — every PDF is searchable across your Dataspace
• OCR for scanned PDFs — text extracted from images
• AI extraction — pull named fields from contracts, invoices, statements
• Citations — AI answers reference specific pages and quotes

Reading and annotating#

Open any PDF entity in a Window. Use the toolbar for highlight, underline, comment, and pen tools. Annotations are stored on the entity itself and visible to anyone with access — true multiplayer reading.

AI moves over PDFs#

A few common AI patterns:

• "Summarize this PDF" — a tight overview with page references.
• "What does the contract say about termination?" — quoted answer, with the page cited.
• "Extract every party, date, and dollar amount from this contract" → a structured Contract entity.
• "Compare these two agreements" — opens both, highlights differences.

For high-volume documents (legal, medical, financial), pair with a Workflow that runs PDF extraction on every new file in a folder.

Bulk operations#

Drop a folder of PDFs into River and the system extracts, indexes, and makes all of them searchable in seconds. For 100+ files, this is the fastest way to build a searchable knowledge base.

Tips#

• For contracts, build an Agent that extracts Contract entities on file.uploaded events.
• Annotations are entities too — search for "highlighted by Maya last week" works.
• The AI cites pages. Always verify the citation; PDF parsing is reliable but not infallible.