--- title: 'Claude + Obsidian: A Free and Powerful Memory System' source: 'https://youtube.com/watch?v=6nuBUPIQgXs' video_id: '6nuBUPIQgXs' date: 2026-07-28 duration_sec: 494 --- # Claude + Obsidian: A Free and Powerful Memory System > Source: [Claude + Obsidian: A Free and Powerful Memory System](https://youtube.com/watch?v=6nuBUPIQgXs) ## Summary This video demonstrates a free, three-part system combining Omi, Obsidian, and Claude to create an infinite context engine that remembers everything about you. By automating note capture and syncing with a local vault, the setup eliminates the need to re-explain context in every chat, saving hours each week. ### Key Points - **The Problem with AI Memory** [00:03] — Claude remembers fragments but not your full context, forcing you to re-explain yourself repeatedly—costing up to an hour daily. - **Three-Piece Stack Overview** [00:30] — Omi captures audio and screen content, Obsidian stores notes locally as markdown, and Claude reads/writes to the vault—forming a two-way loop. - **Graph View of Memories** [01:10] — Obsidian provides a visual graph connecting projects, people, and tools, with Claude using this structure to understand your current work. - **Omi: The Capture Layer** [03:11] — Omi is an open-source wearable (or app) that logs conversations, tasks, and memories automatically—free when skipping the subscription. - **Connecting Claude to Obsidian** [04:44] — Two methods: use an Obsidian MCP server (Claude natively supports it) or simply point Claude's file system connector at your vault folder. - **Live Demo: Building Tools** [05:20] — Claude builds an ROI calculator and CLV calculator as HTML pages, saving them into Obsidian and storing the context for future use. - **PARA Structure for Organization** [06:01] — Tiago Forte's PARA method (Projects, Areas, Resources, Archive) keeps notes organized in four buckets, making it easy for AI to navigate. ### Conclusion By combining Omi, Obsidian, and Claude, you create a free, self-updating memory system that gives AI agents full context. Start small with one folder, use Git for backups, and begin with the MCP server. ## Transcript almost everything about you? What if that blank stare at the start of every chat is costing you time every single day? Why are top creators quietly wiring nobody talking about it? There's a free way to fix this and it changes everything. I'm the digital avatar of Julian Goldie and I help people actually work. In the next few minutes, I'll show you how I turned Claude into a system that remembers my whole life, my goals, my projects, every person I work with, every tool. And the best part, every piece of this is free. Stick with me does the heavy lifting for you automatically. So, let me show you what this looks like. This is Obsidian. And Obsidian and Claude work together as a powerful memory system that maps out everything about me and my life. It takes notes for me all day long, then plugs it into a knowledge system that color-coded and all of this feeds directly into Claude and my other AI agents. If we go to the memory section, I've got Claude here and free Claude 2. Both are plugged into the same memory system. You get this beautiful graph understands me, my goals, every person I work with, every tool, every business I have. Here's the other cool thing. It has memories and they get updated every you track everything in one place and there's a full structure that's really of this. My agents do it for me automatically. It even organizes what I'm working on right now, like the Agent OS system I was building a minute ago. and it's the best system I think you can hand your AI agents. So, let me explain why this matters. The problem with Claude is that it remembers fragments, a name, a preference, what you told it last week, but it has no idea what you actually worked on, what decisions you made, or what's on your plate. The doesn't understand what you did yesterday or what's current today. So, minutes of every conversation re-explaining yourself. And even then, it doesn't know your situation. Now, multiply that across every agent you run, Hermes, Codex, anti-gravity, free be losing an hour a day just re-explaining context over and over. these AI models, I don't actually think the model or the API is the bottleneck. The context is the real problem here. The model itself is smart enough, it just doesn't know you. This is exactly the kind of setup we built together inside the AI Profit Boardroom, and it's where my Agent OS system lives. Inside the Boardroom, you get the full Agent OS walkthroughs for wiring Claude to methods, so you can pick the one that you can share your screen, show me your vault, and ask why your setup isn't syncing right. You get the roadmap for building your own infinite context engine, plus the exact prompts I use to get Claude to organize a messy vault build this without guessing your way through it, that's the whole reason the Boardroom exists. You get the Agent OS walkthroughs, the calls, and the prompts, all built around this exact system. So, this is a three-piece stack, captured, organized into your vault, and Claude both reads from that vault and writes back to it. That two-way loop is what makes it feel alive. The first piece is an open-source project called Omi, the capture layer. It listens to your day, takes notes, records your conversations, and identifies tasks. And it's free to use as long as you skip the Omi chat subscription, which you don't need. It logs memories automatically, creates tasks, can rewind, record what you're saying on the microphone, and even screen capture. And here's why that matters. You are not doing any of this manually, because that's a lot of work. Instead, you train your AI agents to take the notes on you. That's what Omi does. It captures your context all day, then syncs into Obsidian. It's already pulled around 1.8k memories into my vault stored locally. The second piece is Obsidian, where everything lives. It's local, it's private, and it stores your notes as plain markdown files right on your machine. That's a folder of simple text files that you own. And because it's just plain files, it gives Claude a clean brain to read from. Obsidian is free, too. The third piece is Claude, the reader and organizer. It reads your whole vault, understands it, copying and pasting notes into a chat one at a time, Claude works with the entire vault directly. It can search your notes, create new ones, and reorganize your folders. So, the loop is simple. Omi captures, Obsidian stores, Claude reads and writes back. And every agent you run becomes far more powerful because it finally understands you. Now, let me show you how to connect Claude to Obsidian. There are two ways, both beginner-friendly. The first is an both beginner-friendly. The first is an Obsidian MCP server. MCP is basically a outside data. Claude supports it natively. You point the server at your vault, and Claude can read, search, and write to your notes directly from your markdown files. Obsidian doesn't even need to be running. The second is even simpler. Since your vault is just a folder of plain markdown files, you can use Claude's file system connector and point it at that folder. No plugins needed. Claude can open your files, create new ones, and reorganize your folders all in plain language. And free users can do this. You just need Claude desktop, not a paid plan. Now, let me show you what happens once it's connected. I asked Claude to build me a single web page for a content marketing ROI calculator with inputs for articles per month, cost per article, average monthly visitors per article, and lead conversion rate. Claude built it as one clean HTML page saved right into my Claude to build a customer lifetime value calculator, again as a single web page with inputs for average purchase value, purchase frequency per year, average customer lifespan, and acquisition cost. It built it and saved here's the memory part. Every one of those prompts got saved into my Obsidian memory. So, the next time I open Claude, it already knows what I built and why. So, now you might be wondering how all these notes stay organized instead of structure comes in. The system I use is called PARA, Tiago Forte's structure for understands it, so you don't need to teach it the rules. It gives you four buckets. First, projects, active work, the things you're pushing on this quarter. Once a project is done, it moves to the archive. Second, areas, marketing or the clients you work with. Stuff that doesn't end, it just continues. Third, resources, topics you care about long-term, reference material, articles worth saving, books, frameworks worth keeping. And fourth, archive, everything no longer active, old projects, outdated resources, still you've got projects, areas, resources, and archive, four clean ways of connecting everything together. And the more you feed your agents inside this structure, the more useful they get. This is what I call the infinite context engine. Let me leave you with a few pro tips if you're just starting out. Start small. Don't try to import your entire life on day one. Point Claude at one folder, get comfortable, and grow from there. Use Git or a simple backup before So, if a batch change does something unexpected, you can undo it. And begin word MCP sounds scary. It's the easiest on-ramp. If you want the full process, the SOPs, and 100-plus AI use cases like this one, join the AI Success Lab. Links are in the comments and the description. You'll get all the video notes from there, plus access to our community of 85,000 members who are crushing it with AI. And here's my last word for you. Omni capturing, you'll point Claude at your vault, and somewhere along the way you'll hit a snag. Maybe the MCP server won't connect. Maybe your vault syncs something and you're not sure how to structure PARA for your own work. That's exactly what we solve together inside the AI Profit Boardroom where the full Agent OS system lives. You get the connection walkthroughs, the live roadmap for building your own infinite context engine, and the prompts to get way. So, if you want to get this built fast instead of struggling through it alone, come join us at aiprofitboardroom.com.