Can I use Jarvis to summarize PDFs?

Yes. Open any PDF in Preview, Acrobat, Chrome, or your browser of choice, press Cmd+/ on Mac or Ctrl+/ on Windows, and ask "summarize this PDF" or "pull the action items from this contract." Jarvis (getjarvis.eu) screenshots the visible page and, for connected files (Google Drive, OneDrive), can pull the full text via the connector API. Routing goes between a faster model for quick triage and the strongest available one for nuanced summarization (with OpenAI realtime for voice), with very long PDFs handled by a frontier model. PDF content sent to the LLM is not retained for training by Anthropic, OpenAI, or Google enterprise APIs. The free tier covers daily PDF summarization; paid plans at $16 and $20/month raise the quotas. Works on Mac, Windows, and Linux from inside any PDF viewer.

Jarvis can summarize a PDF two ways. (1) Screen-aware: the PDF is open on your screen in any app (Preview, Adobe Acrobat, Chrome's built-in viewer, PDF Expert, Skim). Press Cmd+/ and ask "summarize this." Jarvis screenshots the visible page and sends it to a multimodal model — a frontier model for the strongest visual reasoning. The screenshot captures one page at a time; for multi-page summaries, you can paste the file URL if the file is in Google Drive or OneDrive. (2) File connector: if the PDF lives in Google Drive, OneDrive, or another connected storage, paste the file URL or share link and Jarvis pulls the full text via the connector API. Multi-page PDFs flow through a frontier model without losing accuracy.

(1) Executive summary: "give me a 5-bullet summary of this PDF." (2) Action items: "pull every action item, decision, and date from this PDF." (3) Compare two documents: "how does this contract differ from the standard MSA in our Notion folder?" (Combines PDF + Notion connector.) (4) Cite by page: "what does this PDF say about indemnification, and on what page?" Jarvis returns the answer with page references. (5) Translate while summarizing: "summarize this German PDF in English." Jarvis handles dozens of languages via the multilingual models. All five work from Cmd+/ on Mac or Ctrl+/ on Windows whichever PDF viewer you prefer. Best practice: use the file-connector path for any PDF over 5 pages so the model gets the full text rather than one screenshot.

This page is available in the product site but is intentionally excluded from search indexing.

Docs & files