Create an extractive summary with key points, topics and verifiable page references. Scanned pages use local OCR; your document never leaves the browser and no external AI service receives its contents.
100% private: everything runs on your computer, your files never leave your device.
Summarize PDF reads the text of a document, ranks its sentences by how representative they are of the whole, and returns a Markdown report with an executive summary, a list of key points, the main topics and a few figures about the document. Every sentence in the summary is quoted from the original and carries its page number, so you can open the PDF and check the context in seconds. The analysis runs in your browser.
It is an extractive summary: the tool selects what the document already says instead of writing new sentences, which is exactly what you want for a contract, a technical report, a set of minutes or an academic paper where wording matters and where a paraphrase could introduce errors. Because nothing is generated, nothing is invented, and because nothing is uploaded, the document can be as confidential as you like.
Sentences are scored on how specific they are, how well they represent the rest of the document, how central they are to the vocabulary of the text, and whether they appear near the beginning or share words with the title. The best candidates are then chosen one by one, penalising sentences that repeat what an earlier pick already says, so the summary covers different parts of the document rather than one paragraph.
Text extraction, optional recognition of scanned pages and the ranking itself all run inside your browser tab. The PDF is never uploaded, no passage is sent to any service and no third-party script is loaded. If a scanned page needs recognition, the engine and its Spanish and English data are downloaded once from OléPDF's own domain and run locally. The Markdown report is written in memory and saved to your disk.
That is why the summary can be used on due diligence files, medical reports or unpublished manuscripts without a confidentiality problem.
Yes. Summarising one PDF at a time is free and unlimited, with no account. Summarising several PDFs in one batch requires the 7-Day Pass or Pro.
No. The text is extracted and ranked in your browser. Nothing about the document leaves your computer, and the report itself says so in its footer.
No. It is extractive: every line is a sentence taken verbatim from the PDF, with the page it comes from. The Focus field only changes which sentences are preferred.
A Markdown file with sections for the executive summary, key points, main topics and document figures such as page count, word count and estimated reading time. It opens in any text editor and pastes cleanly into email or notes.
Yes, up to 50 scanned pages per document. Those pages are recognised locally in Spanish and English before the selection runs, which takes a few seconds per page.