0:00 / 0:00

Obsidian is not your second Brain #claude #llm #aiagent

@pro.glitch
14.6K views569 likes2:41ENJul 13, 2026
525 words2886 characters52 sentencesReadability: Grade 4

Transcript

Everyone building a second brain right now is obsessed with this. The Obsidian Graph. All these beautiful little nodes connected by glowing lines. And I'm going to be honest with you, it's a trap. Obsidian has almost nothing to do with an actual second brain. You could delete it tomorrow and your second brain would work exactly the same. Let me show you what a second brain really is. And why that pretty graph is the least important part of the whole thing. First, what a second brain actually is, it's not an app, it's not Obsidian. It's just your knowledge written down in plain files. Everything about you, your work, your projects, your customers, kept in one place and kept updated. That's it, plain text, folders. The power was never the tool, it's the knowledge inside it. So where does Obsidian actually fit? Obsidian is just one way to look at those files. It reads your notes, finds the links between them, and draws that famous graph, all the dots and lines. It's a visualizer, a really nice one. But here's what almost nobody realizes. Those connections already live inside your files. Obsidian doesn't create your second brain. It just draws a picture of it. Take Obsidian away in every note, every link, every piece of knowledge is still sitting right there. Nothing's lost, and here's why you don't actually need it. That graph looks amazing with 50 notes. But your real second brain isn't going to have 50 notes. It's going to have thousands. And a graph with thousands of notes? That's not insight, that's noise. A giant tangled hairball, you can't read, can't navigate. And honestly, never look at twice. Nobody sits there staring at 3,000 connected dots trying to find an answer. It looks cool in a screenshot. It's useless in real life. Because here's how you actually use a second brain. You don't look at it, you talk to it. You ask an AI a question, and it searches across all of those files in one second and hands you the answer. Ground it in everything you know. That's the entire point. 99% of your time is asking and getting answers. Not admiring a map of dots. The AI is the interface, not the graph. So flip the whole thing around in your head. Your second brain is your files. Your knowledge, the AI is how you actually use it. An obsidian? It's optional. It's a magnifying glass you pull out once in a while. If you happen to like seeing the connections, but you can build, grow, and run an incredibly powerful second brain and never open it a single time. So if you've been stuck trying to make the perfect graph, stop, build the knowledge, point an AI at it. That's the second brain. And if you want to learn how to actually build this, the files, the structure, connecting it to an AI, even building agents on top of it that run your work for you, I teach the whole thing from scratch step by step. The links in my bio come build it the right way.

HD Downloads

Sign in required for HD downloads

Related Videos

Why most first AI projects fail in the first month (and how to build one that lasts) Why most first AI projects fail, and how to build AI automation for a small business that lasts. MIT looked at hundreds of company AI projects last year, and 95 percent made no measurable difference to the bottom line. This isn't only a beginner problem. Meet Maya, who runs a small online candle shop. She watched a few videos, got excited, and spent three weekends building four AI things: an AI agent that answers every customer email, plus automations for orders, reviews and social posts. Week one felt like magic. Week four, she turned it all off. Here is the autopsy, cause by cause, so yours makes it: 1. She started with the coolest idea, not the biggest pain. Social posts were fun to build, but her real time sink was customer email: about 40 a day, mostly "where's my order". The fix: track your week first. The job you do most, the same way every time, is your first AI workflow. 2. She never measured the before. When it felt like it wasn't working, she couldn't prove it either way. The fix: write down one number on day one. For Maya it was 11 hours a week on email. 3. The AI didn't know her business. Nobody gave it her return policy, so it made one up: 30 days, when her real policy is 14. The fix is a briefing: your policies, prices, shipping times and tone in plain files the AI reads before every reply, so it looks things up instead of guessing. 4. No human in the loop. This is the one that actually killed it. At 2 in the morning the AI promised next-day delivery on a custom order that takes a week, and one angry review ended her trust in all of it. The fix: start with drafts and approve every reply for the first two weeks. Once an email type has been right 100 times in a row, let that type go out on its own. 5. She built four things at once. When something broke, she couldn't tell which one. The fix: one build at a time, tested on 50 of your real past emails before it ever touches a live customer. So she rebuilt it: one job (customer email), one number (11 hours a week), a briefing with her real policies, drafts she approves, tested on 50 old emails first. A month later it answers most "where's my order" emails on its own and she's down to 2 hours a week. Then she built the next one. The AI was never the problem. The way she built it was. This applies to any small business that wants to use AI to automate customer emails: online shops, service businesses, agencies, solo founders. #aiautomation #aiagents #aiemployee #learnai #ai

Why most first AI projects fail in the first month (and how to build one that lasts) Why most first AI projects fail, and how to build AI automation for a small business that lasts. MIT looked at hundreds of company AI projects last year, and 95 percent made no measurable difference to the bottom line. This isn't only a beginner problem. Meet Maya, who runs a small online candle shop. She watched a few videos, got excited, and spent three weekends building four AI things: an AI agent that answers every customer email, plus automations for orders, reviews and social posts. Week one felt like magic. Week four, she turned it all off. Here is the autopsy, cause by cause, so yours makes it: 1. She started with the coolest idea, not the biggest pain. Social posts were fun to build, but her real time sink was customer email: about 40 a day, mostly "where's my order". The fix: track your week first. The job you do most, the same way every time, is your first AI workflow. 2. She never measured the before. When it felt like it wasn't working, she couldn't prove it either way. The fix: write down one number on day one. For Maya it was 11 hours a week on email. 3. The AI didn't know her business. Nobody gave it her return policy, so it made one up: 30 days, when her real policy is 14. The fix is a briefing: your policies, prices, shipping times and tone in plain files the AI reads before every reply, so it looks things up instead of guessing. 4. No human in the loop. This is the one that actually killed it. At 2 in the morning the AI promised next-day delivery on a custom order that takes a week, and one angry review ended her trust in all of it. The fix: start with drafts and approve every reply for the first two weeks. Once an email type has been right 100 times in a row, let that type go out on its own. 5. She built four things at once. When something broke, she couldn't tell which one. The fix: one build at a time, tested on 50 of your real past emails before it ever touches a live customer. So she rebuilt it: one job (customer email), one number (11 hours a week), a briefing with her real policies, drafts she approves, tested on 50 old emails first. A month later it answers most "where's my order" emails on its own and she's down to 2 hours a week. Then she built the next one. The AI was never the problem. The way she built it was. This applies to any small business that wants to use AI to automate customer emails: online shops, service businesses, agencies, solo founders. #aiautomation #aiagents #aiemployee #learnai #ai

1.5K2:44
They rejected my application to Hogwarts but I still found a way to be a wizard. 🧹#illusion #magic #harrypotter

They rejected my application to Hogwarts but I still found a way to be a wizard. 🧹#illusion #magic #harrypotter

2.4B0:18
Welcome to the Sisters Christmas Party! 🎅🏻❤️ #fyp #christmas #HolidayBeat

Welcome to the Sisters Christmas Party! 🎅🏻❤️ #fyp #christmas #HolidayBeat

1.7B0:15
Jailbreak - Clue 5

Jailbreak - Clue 5

691.7M0:16