Omi + Obsidian = LLM Wiki Supermemory!

So today we're looking at Omni, which is a powerful way to basically store all of your memories, conversations. It can see your screen, hear your conversations, and tell you what to do next as an AI, right? Now it's open source and local, or you can download the Mac app like this. And if you've ever been familiar with, for example, like WhisperFlow or that sort of thing, well, you can use this directly inside Omni, right?

And you can, for example, use this to store your memories and use it as a second brain. Now, this is super powerful because what this allows you to do is get loads and loads of context on what you do day to day. And then you can use that, you can export it to Obsidian as a second brain. You can also use those memories inside OpenCLR and it will sync across all your favorite AIs, right?

The other cool thing about this as well is like you can see here right now, it records the audio that I plug into the system too. And I can set what goals I want to focus on and it will tell me what I need to do based on my memories, right? So it has loads of context of me, who I am, what I'm working on, what I'm working towards, et cetera, which is pretty cool. Now, so far this is free, but I'm assuming at some point it will ask you to pay or upgrade once you've stored a certain amount of memories.

And you can also import all of your memories right here, which is pretty cool. Additionally, you've got this rewind section so you can see what you've done recently and see your screen recording. So again, it's recording my screen, which is pretty crazy. And we can change our settings, what we want to exclude, so you can exclude passwords and that sort of stuff.

You can also pull this up as a menu bar at any time. So for example, if I press command and option, I can actually pull up this chat, which has memories on me, right? And I can ask it questions. It can actually see my screen, right?

So for example, if I say, OK, what should I change in my settings to optimize on me? It can see my screen right here, right? So for example, it says we should turn on voice questions. So I'm going to turn that on now and we can pull up that chat anytime and it can see our screen and it has access to us.

Now we can delete our data at any time as well, which is good. We can also change our privacy settings as well. So we can actually choose not to sync to the cloud, which is good. And the other cool thing about this as well is that we can sync it across our AI agents too, which is really useful.

And that's just the settings, right? We also have these apps that we can connect. So you can see, for example, here, I've synced this to Obsidian, which means that inside my Obsidian vault, right? I can kind of use this as an LLM wiki, right?

Where I have all of my memories connected together. And then if I go inside this section, you can see how everything connects together and we can get the AI to really organize our knowledge, right? Using Obsidian as well. So with a combination of Omni and Obsidian, we're giving loads of context to our AI agents so they can understand this in the future.

Now we can also, for example, connect this to OpenClaw, right? We could connect this to Notion, to ChatGPT, to Claude, etc. If we actually install the OpenClaw app, we can open it up now and we can connect all of our context to OpenClaw, which is pretty amazing. We could also, for example, export the memories and the conversations and the context that we have inside Omni into Obsidian and then link the Obsidian MCP to OpenClaw, which is pretty amazing.

We can even link this to Twitter, to, for example, our emails, our calendar, etc., our local files. Really easy to set up. And then if we go to the conversations tab here, we can see what we've been doing recently. Now this is open source as well, so you can connect it to open source.

So you can see here Omni is available on open source. You can connect it to Cursor, to Gemini. It works on desktop too. And it's basically a second brain, right?

So Omni can capture your screen, conversations, transcribe in real time, generate summaries and action items. And it basically remembers everything. The other cool thing here is you can connect it to your desktop, your phone, your wearables, right? So it can remember everything on your phone as well and sync across multiple devices with your AI.

And it has an MCP server, right, which means that you can link it to any AI agent. So, for example, Cloud Code, that could be, for example, OpenClaw, Hermes, any AI agent that you have. But basically what you're doing here is creating the ultimate level of context. And you're doing it in a passive way where you don't have to organize it, right?

Your AI is organizing it. This is organizing it. You can also choose what you don't want to connect. So you could say, OK, I don't want it to screen record for me or I just want it to only select my microphone and not anything else.

And additionally, you can get OmniDevice, which connects to your devices here, right? So, for example, people could connect it to their headphones even, right? The Omni glasses and everything else, which is really crazy. So let's talk about why this is so powerful, right?

It's basically your second brain and it's insanely powerful, right? Your brain is for having ideas, not storing them, right? So Omni stores them. Now, the problem we have is when we're using AI across the board, it doesn't have that much context on us, right?

It can remember some things, but it doesn't have that much memory. This sort of solves that problem. So even, for example, right now, you can see it's sending up a new memory where it's taking all the research that we're using on Omni and then plugging that in, right? We can even make our memories public if we really wanted to.

I mean, obviously, you need to be careful with privacy here, so don't connect anything that you're not comfortable with. You can also add memories manually here and you can see the full brain map, right? So you can see this brain map here. This is really cool.

It's a chart. So if we open this up, we can also see we have tasks as well. So we can set tasks and goals for us. And this is basically like a living LLM wiki, right?

Karpathy style, right? So Andrij Karpathy, one of the most respected AI researchers on the planet, you know, he was the head of AI at Tesla and a founding member of OpenAI. His approach to knowledge isn't just like, read things, remember them, right? He builds like dense, structured, interconnected notes that compound over time.

It's like a Wikipedia, but it's written entirely by you, right? Every idea links to another idea. Every concept has context. Every piece of knowledge builds on the last.

And so the result is a living document that gets smarter every day, right? Now, doing this manually would take a lot of time. Most people would give up. But Omni basically automates the whole thing.

Now, here's the other thing. Here's the other interesting thing as well, right? And most people miss this. Like, every conversation you have is a knowledge event.

Every meeting is a data point. Every idea you speak out loud is a node in your knowledge graph. But right now, all of that disappears into thin air, right? Like, imagine trying to take everything you know and then plugging it into Obsidian, right?

It would just take too long. It would be so much work. Omni can capture all of it without you trying. And with the right setup, it doesn't just save it, it structures it, links it and makes it queryable by AI agents, right?

Your conversations become your wiki and your screen and your recordings and your conversations directly become all your wiki. So you're not just building a second brain, you're building and training a personal LLM on the most valuable data set that exists, right? Which is your own life and expertise. Now, it works in several different layers.

So you've got, for example, Capture, right? Omni captures everything you say and hears all day across your devices, which you've got inside the conversation section. Then we have the layer and structure. So Omni automatically converts those conversations into summaries, decisions, action items and memories, which we've got inside this section right here.

Then we have those structured outputs getting pushed into a living knowledge base, right? I like to use Obsidian for this. So, for example, if we go to my apps and then we scroll down to Obsidian, I can sync everything that we've done and talked about into this memory, right? And now you can see here that it's exporting all my memories, all my details, etc., directly into this Omni memories document, right?

Now, if we go back to our knowledge graph, we can organize our notes, right? We can get them set up better. We've got all of our memories over here and it's already written 163 memories into Obsidian directly. And then we can plug Obsidian's MCP server into OpenCLR or into Hermes or into Cloud Code or whatever we want.

We can also, for example, copy the memory pack and plug it into Cloud and chat GPT in Gemini and export this stuff into Notion too. And then you have the agent context. And this is where it gets really powerful. So your AI agents can query that knowledge base via Omni's MCP server or API to have your context when they act.

And the problem is like most people have AI agents, but they build agents with zero personal context, right? The agents are just so generic and mediocre. Whereas with this, your agents will know everything you know. So in terms of getting the most out of this, well, you can capture your meetings, for example.

You can run the Mac app during every online meeting, let it run in the background during brainstorms, client calls, team stand-ups, casual thinking out loud, etc. And you're not trying to record everything perfectly here. You're just building out a raw data set of your expertise, decisions and thinking. And then the more you capture it, the richer the wiki becomes.

So you want to think of every conversation as a training example for your personal AI. And then you can convert that chaos into knowledge, right? So the raw audio itself is useless, but Omni converts it into structured outputs automatically. So summaries, decisions, action items, memories, right?

So for example, we can add a new task over here. We can open up our conversations, we've got the memories, we've got the rewind section, right, where it actually sees our screen. And then it gives us the goals, the tasks and everything else, right? When you first use it, it's not going to be that useful, but the more you use it, the better it's going to get.

And it kind of compounds on itself. And then from there, you're going to build up the wiki, the living knowledge base, right? So you can integrate this into Obsidian, all my knowledge, right? Everything I know, everything that I'm working on, all my projects.

And it becomes like your external cortex, essentially, right? It's a pipeline that fills in automatically. And then from there, you can get the MCP server and plug it into your AI agents. Now, MCP stands for Model Context Protocol.

In plain English, it's a way for your AI agents to directly query Omni's knowledge and knowledge base, right? So for example, if you go to the apps over here, we can open up the OpenClaw one and then we can integrate OmniClaw with OpenClaw. And that means that you can ask Omni anything and OpenClaw implements it automatically. And all you need is OpenClaw installed, OpenClaw bridge and the tunnel to connect them, which you can do right here, right?

It's a really quick setup to do. So that's basically it. It looks really promising, super interesting. I'm going to keep testing it more and seeing how it performs, but I'm super excited to test this out more.

And also, I think one of the biggest downfalls of using AI agents right now is they just don't have enough context, which this solves. So thanks so much for watching. If you want to get more training on stuff like this, feel free to get it inside the AI Profit Board and you can connect with me inside there. You can ask questions inside the community.

You can jump on weekly coaching calls where we go deep on this sort of stuff. And you can get more trainings on OpenClaw, Hermes, AI agents, creating LLM wikis, etc. Using CarPathy skills and Cloud Code, etc. And this is all inside the AI Profit Board.

So feel free to get it. Link in the comments description or just go to the AIprofitboard.com. You can also connect with people in your local city who are doing stuff like this as well. So feel free to get that and hope to see you inside there.

Cheers for watching.