The BEST Way to Capture Into Obsidian (just got better) — Transcript
Full transcript
- 0:00I want to be able to take any webpage in
- 0:01milliseconds without AI and turn it into
- 0:04perfect distraction-free [music]
- 0:05plain text ready to drop into Obsidian
- 0:08or anywhere else I need it. That's
- 0:10Obsidian Web Clipper and in this video
- 0:12I'll show you my exact workflow, what
- 0:14I'm capturing, how it lands in Obsidian,
- 0:17[music] and how to set up every detail
- 0:18so that you can actually make use of it
- 0:20later. I'm Nick Milo and I'm here to
- 0:22help you build an idea verse, a home for
- 0:24your best thinking. So let's look at a
- 0:26page I want to save. Watch what happens
- 0:28when I click the web extension up here.
- 0:30Look at that. Instantly it's translated
- 0:32the page into clean markdown and even
- 0:34populated metadata, things like the
- 0:36title, who it's by, the source, the
- 0:38published date, and the description. And
- 0:40if I click the book icon here, I get a
- 0:42clean preview that I can read right
- 0:44away. No ads, no navigation bars, cookie
- 0:46banners, just the actual article. And if
- 0:49I just hit add to Obsidian here,
- 0:52done. It's immediately in my Obsidian
- 0:54vault. Here's what it looks like. Look
- 0:56at how clean this is. The title is
- 0:58there, who it's by, the published date,
- 1:00the source, the full article in
- 1:02beautiful readable markdown. [music]
- 1:03This is exactly how I want text in my
- 1:06idea verse without ads and all sorts of
- 1:08distractions. And again, before I even
- 1:10clip it, sometimes I just want to read
- 1:11it. So I'll enter reader mode, just use
- 1:13the built-in hotkey, which is on Mac
- 1:15option shift R, and I can just toggle
- 1:17right in and out. Look at that, back and
- 1:19forth, back and [music] forth. That is
- 1:21pretty cool. Whenever I find something I
- 1:23want to read and I'm tired of ads just
- 1:25like this one that popped up, guess what
- 1:27I can do? Immediately hit that's pretty
- 1:28cool, right? The reader alone replaces a
- 1:31lot of what people use Instapaper or
- 1:33other read-it-later apps for. But the
- 1:35difference here is that this is free,
- 1:36it's open source, and nothing leaves
- 1:38your machine. So how do you set up the
- 1:40web clipper? It's a browser extension,
- 1:42so it works in all the major browsers.
- 1:43I'm using Chrome here, you can also use
- 1:45something like the Arc browser, you can
- 1:47also use Firefox. I find that Firefox is
- 1:49always a few steps behind what's
- 1:51possible in Chrome. All you need [music]
- 1:52to do is search for the Obsidian Web
- 1:55Clipper in your browser's extension
- 1:57store. And this is all happening locally
- 1:59on your machine, no cloud processing. So
- 2:01when I clip something, it goes into
- 2:03Atlas, then sources, then clippings.
- 2:06That's its home. It never moves, I never
- 2:09have to sort it or file it or think
- 2:11about where it should go. It just lands
- 2:12there. And I have a clippings collection
- 2:14note that links every single clip sorted
- 2:17from newest to oldest. So whenever I
- 2:19want to see what I've been saving, I
- 2:21just go here and I can see everything in
- 2:23the best order, reverse chronological
- 2:24order. That means that the most recent
- 2:27thing I captured is always at the top.
- 2:29And if you just want to copy this to
- 2:30your clipboard or save it to a file
- 2:32instead of Obsidian, that's right there
- 2:34next to the add to Obsidian button. Just
- 2:36click on this little arrow. In fact, and
- 2:38this is really cool, you can actually
- 2:39use this functionality completely
- 2:41independently of this extension of the
- 2:44web clipper. The creator of this
- 2:45extension, Kapano, actually built a
- 2:47standalone website for it called
- 2:49Defuddle. So all you need to do is add
- 2:52defuddle.md
- 2:53>> [music]
- 2:53>> before any URL. So look at this, I put
- 2:56my cursor at the front and I have a
- 2:58hotkey for it, defuddle.md. Now let me
- 3:00hit enter.
- 3:02Whoa.
- 3:04Clean text with metadata at the top.
- 3:07This is truly a gift to humanity. I know
- 3:10that is kind of hyperbole, but when you
- 3:12think about how much we're bombarded by
- 3:15ads and all this content and everything,
- 3:17this is such a gift to be able to have
- 3:20the text, the information that we
- 3:22actually want and nothing else. And if
- 3:24you're working with AI, this is crucial
- 3:27because AI reads clean plain text
- 3:29dramatically better than a screenshot or
- 3:32a messy copy and paste from a webpage.
- 3:34This is the cleanest possible way to
- 3:36hand a webpage to a language model. And
- 3:38again, it's totally open source and can
- 3:40be built [music] into any other
- 3:41workflow. Now let me show you a few more
- 3:43things you can capture because this is
- 3:45where it gets fun. Twitter X posts, blog
- 3:47posts, recipe pages which are
- 3:49notoriously aggressively cluttered,
- 3:51Insta cleaned, tech docs, help pages,
- 3:54and then YouTube. So on any webpage,
- 3:57same idea, if you click into reader mode
- 3:59or use that hotkey, option shift R,
- 4:03immediately I can watch the Cosmos
- 4:05episode here and as it's happening, I
- 4:08can auto scroll through. And what's so
- 4:10cool, I can scrub
- 4:13through the text here.
- 4:16Just by clicking, dragging, and holding
- 4:18on the timeline on the left. I love
- 4:20that. [music] It reminds me of my
- 4:21editing days working on Better Call Saul
- 4:23and Avid Media Composer. If we look at
- 4:25what's happening here, as Carl Sagan is
- 4:27talking, we're seeing line by line where
- 4:31he's at in the script. So we can both
- 4:33see him and watch him, but we can also
- 4:35read the text at the same time. Multiple
- 4:37ways to learn, to interact, to take in
- 4:39the information that might matter to us.
- 4:41Additionally, [music]
- 4:42we can click on something. So how about
- 4:45the surface of the Earth is the shore of
- 4:47the cosmic ocean? So I select that,
- 4:51click highlight, and boom, it's
- 4:53highlighted. Now what's so cool about
- 4:54this is if we go back up to the web
- 4:56clipper, I can go to the highlights and
- 4:58just save what I've highlighted. And if
- 5:00you go into the settings, you can switch
- 5:01what is being highlighted. So right now
- 5:03I have replace page content, but if you
- 5:06want to keep the whole [music] text as
- 5:08well, you can say highlight the page
- 5:10content. Now if we go back to Cosmos,
- 5:12let's try that one again, and if we go
- 5:14to highlights, now we'll have the whole
- 5:18transcript here. And if we scroll down
- 5:20to that section where I highlighted, and
- 5:22we can see it's right here and this is
- 5:24telling me that it's highlighted. That
- 5:26means that when I click add to Obsidian,
- 5:29which is blowing my mind that I have the
- 5:31whole Cosmos episode, almost 6,000
- 5:33words,
- 5:34>> [music]
- 5:34>> in my Obsidian vault. We can go to 5
- 5:38minutes in and see exactly what I
- 5:40highlighted. Pretty sweet, right? But as
- 5:42I'm looking at the Cosmos version, this
- 5:44isn't the final one that I want to use.
- 5:47What I actually want to do is go to
- 5:50summary plus [music] highlights. So now
- 5:52I'm incorporating AI and it's thinking,
- 5:55it's writing a summary, and it's going
- 5:56to add this to my idea verse. So now we
- 5:59see this version has a summary, it has
- 6:02headlines that I love, and then it has
- 6:04things, the things that are mentioned in
- 6:06the episode, people, places, numbers,
- 6:09things, and concepts. This is amazing.
- 6:11[music]
- 6:12What I want to do right now is finally
- 6:14do the thing that I've actually I've
- 6:15wanted to do this for a couple decades
- 6:17now, is to take all 10 episodes of
- 6:19Cosmos, throw them into my idea verse,
- 6:21and then just look at the things that
- 6:24are mentioned in each episode,
- 6:26high-level view, and then dive in as I
- 6:28need to. And if Archimedes is mentioned
- 6:31here, you better believe that if I hit
- 6:33to find the rest of the article, here's
- 6:36Archimedes at the 43-minute mark.
- 6:38How cool is this? This is blowing my
- 6:40mind right now. So let's try another
- 6:42one. This one is Michael Sugrue who
- 6:45recently passed away. He was an amazing
- 6:47philosopher. Here he's talking about
- 6:49Schopenhauer and we can do the same
- 6:51thing, option shift R, and this is going
- 6:53to take us into reader. But notice, for
- 6:56this particular YouTube video, we now
- 6:58see that there are timestamps. There are
- 7:01headers essentially. So if I'm looking
- 7:03at this, I can say Schopenhauer and then
- 7:05I see the eight noble truths, that's
- 7:07Buddhism. How do those two things
- 7:09overlap? Well, I can now just click
- 7:11straight to it. I can read through here.
- 7:13I can see exactly what this overlap
- 7:15might be. Very cool. If you were a
- 7:18Tolkien fan and you're just so into Lord
- 7:20of the Rings, you could go into this
- 7:22channel, In Deep Geek, which is amazing.
- 7:24Basically, [music]
- 7:25transcribe all of the Lord of the Rings
- 7:28video. The Lord of the Rings from
- 7:29Sauron's perspective. Let's open this
- 7:31bad boy up. As I'd go into reader view,
- 7:34and again, we can see the entire
- 7:35transcript is here.
- 7:37And what I'll do is go ahead and just do
- 7:41summary and highlights. I don't have any
- 7:42highlights, it doesn't matter. It'll
- 7:44still do the summary and the full
- 7:46article. There it is, [music] it's done.
- 7:48Extremely fast. Summary, headlines, and
- 7:52things, what's actually being mentioned.
- 7:54This is really cool. Sometimes I just
- 7:55like to kind of it's like a form of
- 7:57being in a dark cave and just
- 7:59echolocation to get a a visual map, like
- 8:02the sense of where the walls are. That's
- 8:04what this thing section does so well for
- 8:07me. Now this is an optional AI layer
- 8:09built right into Web Clipper called
- 8:11Interpreter. To set it up, go to the Web
- 8:13Clipper settings. So if you click on the
- 8:15Obsidian icon, it will be the gear icon
- 8:17here. Go to Interpreter, make sure to
- 8:20enable Interpreter, that this is toggled
- 8:22on, and then add your provider of
- 8:24choice. After that, you can select your
- 8:26model below. Now Obsidian Web Clipper is
- 8:28completely model agnostic. It doesn't
- 8:30care what you use. You can use Claude,
- 8:32GPT, Gemini, Deep Seek, or local models
- 8:35through Ollama, and a lot more. If a
- 8:37better model comes out tomorrow, you
- 8:40just switch. So the Web Clipper template
- 8:41that I've honed over many iterations
- 8:44generates three distinct sections above
- 8:46the actual article. The first is the
- 8:48plain text summary, just a clean,
- 8:50concise, in-summary paragraph. The
- 8:52second are the headlines. These are the
- 8:54top key points I need to be aware of,
- 8:56formatted as a short bulleted list. The
- 8:58third is an unorthodox, but
- 9:00surprisingly, it's what I like to read
- 9:02first. It's called things and it gives
- 9:04me a hot list of key people, [music]
- 9:06terms, and ideas that are mentioned in
- 9:08the clipping. And then of course, the
- 9:10full article content itself is just
- 9:11below that. Now I tried just doing the
- 9:14generic summarize this prompt for a
- 9:16while and the problem was the results
- 9:18were way too long or too vague or too
- 9:21inconsistent. You'd read the summary and
- 9:23think, why am I reading this instead of
- 9:24just reading the article? These three
- 9:26distinct layers let me scan the summary
- 9:29in 5 seconds, read the headlines in 15
- 9:31seconds, or go deeper with all the
- 9:32things that are actually mentioned in
- 9:34the article. Now if you want to get the
- 9:35full guide to setting up the prompts in
- 9:37the Interpreter, I've set up a special
- 9:39page here and you can be processing
- 9:41articles in minutes. Now what's great
- 9:42about all this is that when your idea
- 9:44verse is built on clean, well-structured
- 9:47plain text files, you own them for a
- 9:49lifetime. Plus, you're building a
- 9:51foundation that any future AI model can
- 9:53easily read. Your files remain yours and
- 9:56they're available offline. They're ready
- 9:58to connect to any intelligence, human or
- 10:01artificial, whenever you're ready. And
- 10:03this is absolutely essential if you want
- 10:05to build your own AI assistant. If you
- 10:08have the files, you have the core of
- 10:10your AIOS. Then you just build a
- 10:13playbook that teaches AI how to interact
- 10:15with your notes. Then the AI model
- 10:17doesn't matter anymore because you've
- 10:19future-proofed all of your notes, and
- 10:21that is a wonderful feeling. If you'd
- 10:23like my exact interpreter setup, the
- 10:25specific prompts, and how I've
- 10:27configured this three-layer summary,
- 10:28scan here or check out the link in the
- 10:30description below. And for a full
- 10:31breakdown of my AIOS, essentially how
- 10:34I'm using Obsidian and AI together as a
- 10:36complete system today, check out this
- 10:39video next.
About this transcript
This page contains the full transcript of The BEST Way to Capture Into Obsidian (just got better) by Linking Your Thinking with Nick Milo, generated from the public captions YouTube serves with the video. The transcript has 2,104 words across 304 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.