Importing PDFs

Papers, reports, and contracts become readable project material rather than files you keep meaning to open.

Importing

Drop a PDF anywhere on the canvas. Text is extracted on upload, so the document becomes searchable and available as context immediately. Scanned documents without a text layer are stored but cannot be read by the tools.

Annotating

Highlight a passage and it becomes its own block, connected to the document and carrying the page reference. This is usually more useful than keeping the whole document as context — the passage is the finding, not the file.

Using a document well

  • Extract the two or three passages that matter and connect those.
  • Keep the full document attached for provenance.
  • Avoid selecting entire long documents as context — accuracy on material buried in the middle of a long input falls away sharply.

A 90-page report contributes more as four highlighted passages than as 90 pages.

Related

Files Supported formats and storage. Sources and provenance Page-level references. Working with context Why less context reads better. Web search Finding material.