Gist

Last updated 2026-04-19 · 1 min read

What Gist works on

Articles, papers, PDFs, YouTube, and pasted text are first-class. Google Docs and X/Twitter are best-effort.

Articles. Papers. PDFs. YouTube. Pasted text. These five are first-class.

Google Docs and X/Twitter are best-effort. Everything else, Gist tells you up front that this is not a candidate page.

First-class

These are the paths we tune for. Expect strong results on the happy path.

  • Regular web articles — blog posts, newsletters, news pieces, essays.
  • Academic papers — arXiv is the reference example. The layered read maps well to the abstract → methods → discussion → limitations structure.
  • PDFs — linked on the web or opened locally in Chrome. Quality tracks the PDF's own readability; a clean text-layer PDF works well, a scanned image less so.
  • YouTube videos — the layered read runs against the transcript. Works well for talks, lectures, interviews, explainers.
  • Pasted text — drop raw text into /try and you get the same layered read without any extension step.

Best-effort

These work when the source is clean. They are not the main path.

  • Google Docs — readable, link-viewable docs work; permission-gated or heavily-formatted docs often do not. See Google Docs not working.
  • X / Twitter status pages — individual threads and status URLs work; timelines and composite views do not.

What Gist does not do today

  • News homepages, timelines, composite index pages. No coherent piece to Gist.
  • Paywalled content we cannot access.
  • Media without a transcript — audio without captions, silent video.

Where /try fits in

/try is the paste-and-upload web surface. It accepts any URL, any local file, or raw pasted text. Use it when the extension is not the right tool — a local PDF not open in a tab, a snippet of text without a URL, or a device where Gist is not installed.

What is next

Related