Saltar al contenido
Volver a todas las herramientas№ 053

Resumir PDF

Resume un PDF o hazle preguntas, 100% en el dispositivo con la IA integrada de Chrome. Puntos clave, TL;DR y un chat privado — sin subidas.

0 B subidos a un servidor

Añade primero un archivo.

This tool reads a PDF and summarises it — and then lets you ask it questions — using the language model built into Chrome, running on your machine. Nothing about the document is uploaded, which is the whole point: the documents people most want summarised are usually the ones they are least able to paste into a chatbot.

You get a choice of summary shape rather than one fixed output: key points for a document you need to act on, a TL;DR for one you need to triage, or a teaser for one you are deciding whether to read at all — each in short, medium or long. After that the chat takes over, answering from the extracted text as its context. Very long documents are truncated for the chat, and it says so rather than quietly answering from half a file.

How it works

  1. Drop in a PDF with selectable text.
  2. Pick the summary type and length, and let it stream in.
  3. Copy the summary or download it as Markdown.
  4. Ask follow-up questions in the chat if you need more.

What people use it for

  • Triaging a long report before deciding which sections deserve real reading.
  • Getting the key points out of a contract you have been sent at short notice.
  • Interrogating documentation for the one answer you actually need.

Frequently asked questions

Is my file uploaded to a server?
No. The file is opened by JavaScript and WebAssembly running inside your own browser tab. It is never uploaded, there is no server-side processing and nothing is stored — closing the tab discards everything.
It says the model is unavailable.
The on-device language model is a recent Chrome desktop feature and your browser is not providing it. There is deliberately no cloud fallback — sending the document to a server is exactly what this tool is built to avoid.
Can I trust the summary?
Treat it as a fast first read, not as a substitute for the document. On-device models are small, and any language model can state something confidently and be wrong. Check anything that will inform a decision against the page it came from.
What happens with a very long PDF?
The text is truncated to fit the model's context, and the tool tells you when it has done so. If the part you care about is at the end of a long document, split it first and summarise that section.
Can I summarise a scanned document?
Only after OCR. Until a text layer exists there is nothing to read — the pages are pictures.

Herramientas relacionadas