How Recl auto-sorts everything you save into topics
When you save something, Recl reads it and labels it with the topics it actually covers — from a fixed set like Neuroscience, Startups and Philosophy — then files it in your library under each one. You pick no folder and type no tags. A neuroscience talk lands in Neuroscience; a founder's thread lands in Startups.
Every save-it-for-later tool has the same quiet failure: the folder. You save a brilliant video, pause on the "which folder?" question, pick "Watch later" because it is easiest, and never see it again. The filing was supposed to help you find things. Instead it became the reason you stopped saving.
Recl removes the question. You save the thing. Recl reads it and files it for you.
What happens when you save something
The moment a save finishes processing, Recl has read the whole thing — the transcript of a video, the body of an article, the text of a post. From that it does two jobs at once: it writes the summary, and it decides what the save is about.
"About" is not a free-text guess. Recl sorts each save into one or more topics from a fixed library set, so your whole library speaks the same vocabulary instead of accumulating forty near-duplicate tags like startup, startups, founders and founder stuff.
The topic set covers the areas people actually save in — among them:
- Mind & science — Neuroscience, Psychology, Biology, Physics, Space & Astronomy, Mathematics.
- Work & building — Startups, Business & Marketing, Productivity, Programming, Artificial Intelligence, Design & UX.
- Money & the world — Economics & Finance, Crypto & Web3, Geopolitics, History, Law & Politics.
- Life & culture — Philosophy, Self-Improvement, Fitness & Health, Film & Media, Music, Art & Culture.
A worked example
Say it is a normal week and you save three things without thinking about where they go:
- A 90-minute YouTube talk on how sleep clears the brain. Recl reads the transcript, sees it is about memory consolidation and the glymphatic system, and files it under Neuroscience (with a second tag of Fitness & Health, because it is partly about sleep habits).
- A founder's thread on X about pricing their first SaaS product. Recl reads it and files it under Startups, with Business & Marketing alongside.
- A PDF of a lecture on free will. Recl files it under Philosophy.
You chose nothing. Open your library a month later and the sleep talk is sitting in Neuroscience next to every other brain video you have saved — not buried in a "Watch later" pile with a cooking video and two memes.
Why a save can have more than one topic
Real content is rarely about exactly one thing. A talk on the economics of attention is both Economics & Finance and Psychology. A video on building an AI startup is Artificial Intelligence and Startups. Forcing it into a single folder would mean you only find it from one direction.
So Recl tags the save with every topic that genuinely applies. That one decision is what later lets your saves connect to each other — two items that share a topic are the raw material of your knowledge graph, and of the answers you get when you ask your library a question.
You can still steer it
Automatic does not mean locked. Topics are the scaffolding you get for free; on top of them you can add your own tags, drop saves into folders, and bookmark the ones you want close at hand. The point of the auto-sorting is that your library is never empty-shelved and unsearchable just because you did not feel like filing — not that you are forbidden from organising it your own way.
Why automatic sorting is the part that matters
A bookmark you cannot find is the same as a bookmark you never made. The reason most saved libraries rot is not that people stop saving — it is that the cost of filing, paid at the exact moment you are trying to get back to what you were doing, is high enough that the library never becomes trustworthy. Once you cannot trust it to hold the thing you half-remember, you stop looking, and then you stop saving.
Recl pays that cost for you, on every save, in the background. The result is a library you can actually search by topic on day one — and a foundation the rest of the product is built on.
Frequently asked questions
Do I have to choose a topic when I save?
No. You save the link, video, PDF or post and Recl assigns the topics itself after it reads the content. You can add your own tags or folders afterwards, but nothing is required to get a save filed and findable.
Can one save be in more than one topic?
Yes. Recl tags a save with every topic that genuinely applies, so a video about an AI startup shows up under both Artificial Intelligence and Startups. Shared topics are also what connect your saves to each other.
What if Recl picks the wrong topic?
The automatic topics are scaffolding, not a cage — you can still add tags and folders of your own on top. The goal is that your library is never left unsorted and unsearchable just because you did not want to file it by hand.
Does this work for things other than YouTube?
Yes. Articles, PDFs, podcasts, X and Instagram posts, TikToks and plain notes are all read and sorted the same way, because the topic comes from what the content is about, not from where it came from.
Sources
Asutosh
Co-founder, Recl
Builds Recl, a memory system for long-form video and reading. Writes about why people forget what they watch, and what actually fixes it.
Related reading
- Your knowledge graph: how your saves connect themselves by topicA flat list hides the best part — how your ideas relate. Recl's knowledge graph links your saves by shared topics, and builds itself as you go.
- Building a second brain from video, not just articlesMost knowledge systems assume text you can highlight. What changes when the source is a three-hour podcast, and how to build one that survives it.
- Ask your library: get answers from your own saves, with citationsSearch finds the document. Ask gives you the answer. Put a question to everything you've saved and Recl answers from your library, with sources.
- Why a paid memory system beats free bookmarkingFree bookmarking makes a list of links you never reopen. Remembering is a different, harder job. Here is the honest case for paying for a memory system.