What research tool keeps the lectures, podcasts and threads a reference manager doesn't hold?

Papers have a DOI, a publisher page and a place in your reference manager. Much of what shapes a project has none of that: a conference keynote posted on YouTube, a lecture series on a method you are learning, a podcast interview with an author whose book you are reading, a long thread on Bluesky or X where people in the field pick apart a new preprint, a Reddit discussion among people who have run the method.

That material is often where you first meet an argument, a critique or a practical caveat. It ends up in browser tabs, a Watch later playlist and the saved lists of four different apps. Six months on you remember the point and not who made it or where, which is the worst state for anything you might need to check or cite.

Reference managers can hold a record for this kind of source. Zotero, for example, has item types for a video recording, a podcast and a forum post, and you can attach notes and files to the record. Unless you write a note while you watch, what was said and why it mattered to your project stay in the source.

• PhD students and early-career researchers who follow their field through recorded talks and seminars on YouTube.
• Students working through a lecture series who need the material to stick well past the week they watched it.
• Researchers who keep up with the discussion of new work on X, Threads, Bluesky and LinkedIn.
• Anyone whose field moves through podcasts and interviews.

Paste the link and it usually comes back as a lesson within a minute. A YouTube talk with captions brings a transcript taken from them, and a post on X, Threads, Reddit, Bluesky or LinkedIn brings its text. Every lesson has a one-line core idea, the key points, the one sentence worth keeping marked as Remember this, a Try this with one thing to do when the post suggests one, and the link back to the original with the creator's handle.

For research use, the link and the handle matter most. The lesson tells you what was argued and by whom; the source is what you check and what you cite. Treat the lesson as a pointer and a memory aid, and go back to the original before you quote anything.

Each lesson is filed into a topic on its own, named after what you keep saving, so you don't have to build a folder tree. Someone following lectures on statistics and a podcast on sleep science ends up with topics that follow those subjects. Change moves a lesson to another topic, or to a new one you name yourself, and when several small topics pile up the app suggests grouping them with Group topics, and you decide.

Each topic has a big picture, a short summary across all of its lessons that is rewritten as the topic grows. It is a quick way to see what your lessons on a subject say so far, across talks, episodes and threads. It summarizes only the sources you saved, so it maps your own listening; a literature review still needs the literature.

Related lessons sit under a lesson when some are close in meaning, and under a search. That is how a claim from a podcast guest can turn up next to the lecture that covered the same idea from another angle.

Today's quiz asks a few questions a day about lessons you saved, each written from the lesson. You answer in your head, tap Show answer and grade it: Forgot, Kind of or Knew it. Forgot brings the lesson back tomorrow and starts it over, Kind of brings it back in 4 days, and Knew it in 14 days, then 35, 88 and 219 days, then a year.

Before an exam or a seminar, Quiz this topic runs through one subject on its own. The topic's knowledge score counts every lesson that has come up for a quiz, graded or simply due, and shows the share of those you remembered. Quiz days you miss count too, so a topic whose due lessons went unanswered shows 0%, and a topic with nothing come up yet shows no score at all.

My takeaway holds up to 500 characters of your own words on any lesson. For research it is a good place for the link to your own work: which chapter a claim bears on, or which paper to check it against. Search covers takeaways along with titles, key points, the core idea, Try this, Remember this, topics, creator handles and the exact words of transcripts.

• Lectures and talks with captions, the speaker's own or YouTube's automatic ones, come back with their transcript. YouTube lists very long videos, poor sound and overlapping speakers among the reasons automatic captions can be missing, which describes plenty of seminar recordings.
• A long recording is read from beginning to end. The lesson is written from the whole transcript, so a 90-minute seminar comes back in four or five sections, one for every part of about 20 minutes. When a point you need is not among them, add it to the key points yourself with Edit lesson, the pencil on a lesson.
• A YouTube video without captions gives no lesson. The speech itself is not transcribed, so the app has only the title, and on video platforms a lesson is never made from a title alone. It asks for the video file instead, and the upload takes files up to 50 MB, which fits a short clip better than a full lecture.
• Podcasts work when the episode is on YouTube with captions, and the lesson is written from the whole episode; the example above is the kind of lesson such an episode becomes. A Spotify or Apple Podcasts episode is transcribed from start to end when its show has a public podcast feed, as most shows do; a Spotify exclusive gives only its title and description, so a short description gives no lesson.
• On X, Threads, Reddit, Bluesky and LinkedIn, the text of the post you link becomes the lesson. For a thread, that is the post you paste, not the replies under it, so link the post that carries the argument. When the post's text is under 300 characters, a video on X, Threads or LinkedIn is transcribed, and one on Reddit is read from its own captions; when that fails, the post's text makes the lesson. A longer post is the lesson on its own. LinkedIn articles are saved as well, and a long one is read whole.
• A post that has been deleted or made private can't be read, so paste it when you find it.

• It holds no papers and keeps no file library: a photo of a printed page, a voice memo or a video file you add becomes a lesson or a card.
• It produces no citations or BibTeX, and its export is lessons as Markdown, CSV or JSON, with no sync to Notion, Obsidian or a reference manager.
• It does not save journal pages, preprint servers, blogs or any other website; LinkedIn articles are the only articles it takes.
• Transcripts have no timestamps. Search finds the exact words of a transcript, so to quote a talk, find the claim that way, then open the source and find the moment there.
• There is no shared library or team space, so a lab, a supervisor or a reading group can't work in it together.

A split that works: papers live in your reference manager, and scrollandlearn holds the talks, episodes and threads around them. When one of those becomes something you will cite, add it to the reference manager from the link on the lesson.