Nick Milo Reads My Obsidian Vault  🔗 **Source:** [Nick Milo Reads My Obsidian Vault](https://www.youtube.com/watch?v=bdjYo-x4PUw) [00:00:00] Nick, welcome back to the channel. Happy to be here, Thiago. So, I uh want to get some free coaching on the intersection of Obsidian and AI. I feel like this is the [00:00:18] conversation these days. People want to use AI obviously. They want to own and control and know where their data is. That's Obsidian file over app. and you [00:00:29] have been living and working at that intersection um longer than anyone and I feel like have thought more deeply than anyone I know. And so I thought we would [00:00:40] open up my Obsidian vault and I would ask you some questions. Just like I kind of want to see through your eyes, see what you see, notice what you notice, [00:00:49] and then kind of walk through a series of like questions and experiments and different ways of like uh playing with or interacting with AI with Obsidian. No, that sounds good. I'm game for it. [00:01:01] And let's see what magic we can uncover. All right. So, first thing, uh, as you can see on the screen, we have my Obsidian Vault. You can see down here it [00:01:12] says Obsidian Vault, Fort Labs. So, before we even talk about AI, the first thing I want to ask you is I I've always wondered this. Actually, I've [00:01:22] always wanted to ask you when you see a vault open up, what do you notice? Like almost like a doctor, like what's what is your diagnosis of [00:01:32] >> That's funny. First thing I'm thinking is what theme is it? So, it's like default dark theme. >> So, you already know I'm basic. >> Yes. Hey, and no worries. Um, and then [00:01:41] I'm looking at the folder sidebar to see what you know, is there any um strong folder structure? And so >> that would be right. That would be it [00:01:50] would be here, right? >> Well, like your clippings read wise this section right here. >> Yes. Okay. >> So, the file explorer and I'm just seeing you know what is showing up here. [00:02:00] So you don't have parah in this one because it's not like your go-to vault. And so I'm like, okay, organization just so I know it's giving me context to [00:02:09] understand what I'm looking at. So I look at the theme, I'm looking at the sidebar, it's giving me a that's about as much of a sense I can get. If there [00:02:18] was a note open, I would try to see if there were links. That would tell me a lot about how the person is linking their thinking. [00:02:25] >> Um, and then we go from there. What about if you see my you've probably looked at so many graphs. >> Can you like deduce anything from a [00:02:32] graph view? >> It's funny. They are less helpful than I mean they're great eye candy but they're not terribly helpful. But I can see you have like one big project. [00:02:43] >> Mhm. >> Or something here. What do we what's it called? >> Um for the book life in perspective. >> Okay. Yeah. >> LIP content. All right. So that's [00:02:52] basically what I'd call a map of content and you have all these things linking to that. Then this might be like a tag or something. I don't know why we're not [00:03:01] liked things that you've liked. Is that like a readwise? You know, we could explore that. So read wise because I know that that's here and I know that [00:03:09] readwise imports a lot of data. So that makes sense. And then we have all these um slightly connected light notes and then placeholder notes which is the [00:03:21] coolest thing I think that you can link to something that's not even a created note and it can actually acrue links to it. >> So then by the time you create it, it [00:03:31] could already have like multiple links. So you're never starting from a blank page. That's what those like faded ones are. Oh, I didn't know that. So all the [00:03:38] dark gray are like kind of empty. Look at that. It's already has a backlink to designing for a behavior change. So I mean there are a lot of pe I have a [00:03:47] notes on a lot of people and by the time I actually say you know what I want to create this note. Usually it's when I link to it again and I saw it was [00:03:54] already created I'm like whoa I linked to this two years ago and this and there's like five links already and then that just gives me so much of a richer [00:04:02] context for >> interesting >> for anything. >> Interesting. >> So that's kind of like what I'm looking at graph view. The other thing I'll do is I'll just hover over right here and [00:04:11] then I'll get a sense of the files. So we have 1662 and four folders. So what else? I would just >> What does that tell you? Is that is that [00:04:20] a lot? Is it a little? Like how do I compare to the average Obsidian user? >> You know, I don't know what average is. And it's fine if you have 20 notes. Um [00:04:27] right now and I'm not saying this in any way. There are people that have way more than me. It doesn't matter. But my uh main vault is at uh 17,000 notes and [00:04:36] it's just been a slow growth thing because I've created most of them myself and I'm not really looking to import a lot. >> So I'd go over here, you know, no back [00:04:46] links will show up. Just kind of seeing what's going on in what tabs. So it's a pretty vanilla setup, like very basic, which is great. You're not getting [00:04:56] bogged down in the million distractions that Obsidian allows you to be distracted by. >> Okay. Fascinating. >> Yeah. >> There's so much there's so many little bits of context that you can pick up [00:05:07] just from that. >> You know, that's funny. I need to do more just like show me your like let's get hands on and then cuz I do think at [00:05:16] this point it's like a sixth sense. This one being a little bit of a demo vault or is it? Tell me. >> This is the real vault. [00:05:23] >> This is the real deal. We'll talk about that next. >> Okay. So, to me, this tells me that you're really focused on your work at hand. Yeah. [00:05:30] >> For the book, >> on this specific project there, there were multiple things that you could tell this was a projecentric graph. >> Absolutely, 100%. Because you have your [00:05:38] main project here and then all the readwise supporting uh files that are liked and connected this way. And then I'm not quite sure about kind of the [00:05:47] one-offs, the orphan notes. Me neither, >> the small connections, but yeah, the shape of it is really clear and it makes sense that yeah, if you're using So this [00:05:57] is a true vault. It's for the big book that's coming out. >> Uhhuh. >> This is great. You've spun up an entirely new vault separate as opposed [00:06:05] to trying to like do everything in one place. Perfectly fine. >> And that's where we're at. Um, now let's talk about how we can bring AI into the [00:06:16] picture. So, first thing I notice is I don't want AI to touch my notes, my original canonical notes directly. This graph you see is the result of not as [00:06:28] many hours as you spent, but hundreds and hundreds of hours. The Readwise import was automatic, but everything else, all the linking, the connecting, there's tons of manual [00:06:37] notes, there's tons of outlines with structure was all my personal time. >> Is that a common view, you think? And and do you recommend people point AI at [00:06:48] the canonical notes or create a duplicate or what? >> Uh, always create a backup before you point AI at a folder of notes. And just to take a step back, um, what is [00:07:01] Obsidian? And at its basic form, it's a note app. There are many note apps. Like, let's I mean, I could go like philosophical, but it's just it's a note [00:07:09] app. The difference, I think, is like with Evernote. Evernote holds the notes. Notion holds the notes. >> Obsidian is actually not holding anything. >> It's only a like a skin that you put on [00:07:21] because your computer is holding the notes. So, it's just a file. It's just a folder or maybe it's one folder with four subfolders. Clippings, copilot, readwise, web clipper. That's it. And [00:07:34] then within those folders are text files. >> Mhm. >> So super basic. That's what you know people who use Obsidian really appreciate about it because they don't [00:07:44] have to be locked into Obsidian. I can point anything at the folder of notes. And that's actually how the AI integrations that I'd recommend is, hey AI, just look at this folder. [00:07:56] >> So, okay, just with that little bit of context, if I were to point AI at this safely, there are a few different ways to do it. The first one, always make a [00:08:05] backup >> because if AI can read something, you have to assume even if it even if you don't have the permission for it to write, you have to assume it can write, [00:08:15] which means it can delete, it can change things. Have you seen instances of that where it technically had read only and something was edited? >> I've heard of those instances. I mean, [00:08:23] it's not affected me and it just depends on your tolerance for risk. >> So, I have pointed my AI at all of my notes, but it's because I have a backup [00:08:33] system. So, you have to back up. I do manual. >> I will do automatic backups and then off of the same computer as well. Like, hey, just throw one into Google Drive that's [00:08:44] a zip file. put one on a zip drive, you know, you're ju you're just spreading out your risk that way. So, I haven't been worried about it like maliciously [00:08:54] editing things, but I've heard cases where that's happened. So, if AI can look at a folder, you have to assume it may be able to write, and that's why [00:09:03] those backups are really important. Okay, so that's super helpful. Let's let's actually look at where this data really lives. So, here I have my finder. I have my resources and there you see my [00:09:16] Obsidian vault. >> Cool. >> So, let's assume that I want to create a duplicate. >> Is this really just as simple as rightclick duplicate? >> I would just or I just hit command D. [00:09:28] >> Even faster, >> which I know that you know. >> Yes, >> you're good with hot keys. >> It almost seems like that's too simple. It's hard to believe that it's that [00:09:36] easy. >> Yeah. >> And then I would just do 0709. Am I doing that right? Or 19. Is that something you typically do is date it? >> Yeah. [00:09:46] >> Yeah. >> And I and I'll just do that and then I end up having, you know, a backups folder and that's just where I can kind [00:09:52] of loosely do backups >> of the ondevice variety. >> But again, we want to also have one that's not just on the same device. >> Okay. >> But yeah, it's as simple as that. [00:10:01] >> And if we open this up for anyone who's not familiar, >> so this was a complete perfect duplicated copy of my entire vault, >> right? So these are markdown files. [00:10:12] >> Mhm. >> Which are just plain text, not encrypted, no special proprietary format. It's just plain text. And that's like one of the big selling points of [00:10:22] using an app like Obsidian. And let's really look at this. So if we go back, we just made a duplicate. Like how many duplicates till I fill up my computer? [00:10:32] And if we actually look at the size of this, that's uh what are we 120 megabytes. A little bit larger than I was thinking. So you probably might have [00:10:43] some media. Maybe something with Readwise is pulling in >> maybe >> because with that amount of files, I mean, if they're just the plain text variety, then we're talking [00:10:52] >> like 12 megabytes. >> Yeah, exactly. A tenth of that. >> Probably have photos >> some photos in there. That makes sense. Even. So, that's pretty small. [00:11:02] >> That's tiny. >> Yeah. >> All right. Okay. Well, that's it's nice that I can do it so easily cuz then you can almost create throwaway [00:11:10] duplicates like if you want to try some crazy experiment. >> Yes. >> You just create another one. >> Typically what I'll do. So, it'll be like um [00:11:18] >> pre Nick session like I'll just do little things like that and then that's good. Okay. So, then we don't have to be precious about this one. We don't have [00:11:28] to be scared. We can go crazy. >> Mhm. >> So, let's do that. >> Okay. >> So, this is my next question. What there's so many ways to connect AI to [00:11:38] your Obsidian notes. Now, the three that I know of are just point say Cloudico work at the directory at the folder, but then there's an Obsidian [00:11:46] CLI. Now, then there's like a whole variety of plugins that integrate AI. How do you think what is the >> what is the taxonomy of these different [00:11:54] or the preference that you have for them? just point an AI at the folder. Yeah, I mean, I'll talk through the other options, but at the end of the [00:12:02] day, >> if you use a thirdparty plugin, and there are a couple great ones, there's a third party plugin in Obsidian you can download or community plugin. It's [00:12:11] called Copilot. >> Mhm. >> That's very cool. There's another one that's a little bit different called Smart Connections. >> Sure, use those. But now you're dealing with a few developers and [00:12:21] they're awesome, but they're separate developers. So how are they going to maintain that all this extra all these extra breaking points? >> So it's like what result are you after? [00:12:34] You want AI to be able to look at your notes. You want to be able to work with it. The simplest thing to do is point something like a cloud co-work or GPT [00:12:44] work and just point it at um this, you know, whatever folder of these two that you'd prefer to use. Okay, so let's do that. Let's open up Claude, which is my [00:12:55] LLM of choice. What is yours, by the way? Your main daily driver. >> Yeah, same. Just Claude co-work since they kind of had a currently a first [00:13:02] mover advantage with something that was relatively safe that could do this. >> Yeah. >> Um, they've changed the interface now, by the way. You have to go into home and [00:13:12] then do co-work. Have you noticed that? >> They always change. I'm like, okay, where is it now? So, I'm just going to open up a new uh co-work session. Make [00:13:19] sure it's on co-work, which is much more powerful. Um, what do you think about models in relation to this quantity of data? Cuz I'm giving in 120 megabyte uh [00:13:32] megabytes across 1,600 files. Have you found the either the model or the effort level makes a difference? Uh, typically I just will stick with the opus model using cloud co-work here. um [00:13:47] it because the tokens haven't been as tight as they keep on getting tighter and tighter so we have to make these decisions >> but it depends on the quality of work [00:13:56] too. So I think with where I'm at I just am using it infrequently enough and then I can ask a quality question knowing I'm getting a quality model. But if I'm [00:14:06] doing something like, hey, summarize this article and bring it in and then I have a summary section, then the transcript, that's something I would use a lower model for, like haiku. [00:14:16] >> That's a good idea. >> Yeah. >> Okay. So, we'll stick with that. That's what I typically use. So, I'm most familiar with it. So, I just click here. [00:14:22] >> And that's the magic button really is just project or folder, >> right? There's so much power behind this button. >> And that's all you have to do though. I [00:14:29] mean, that's what's so cool about it is you just click the button and you're like and that's the control mechanism. So, when we talk about like [00:14:35] privacy and owning your notes, because you don't have to point it even at your main Obsidian vault, you could say only the notes that I've manually put into [00:14:44] the subfolder. And then you point Obsidian or you point your AI at the subfolder. >> So, for those that are a little bit more private with what they're sharing, [00:14:52] they're like, "No, only the subfolder with these 10 notes." >> And it's like you have that granular control at any moment with how much you want to share or not. So, this is super [00:15:02] interesting because this implies that you need folders. >> Yes. >> Is there any way to do what you just described with anything other than folders? >> Folders are back, baby. Folders are [00:15:11] back. Okay. So, I'm going to click there, choose a different folder, go to uh resources. Oh, no. Resources. And there's our very nicely labeled vault. >> Mhm. I'm kind of going through this in [00:15:27] slow motion for the benefit of our viewers, but I actually myself get stopped here sometimes like what to even ask? What is even the question? Do you have any sort [00:15:38] of like diagnostic or exploratory questions that you've used or that you turn to? If I may, even before we get there, the first thing I want to tell it because I [00:15:51] don't trust and I want to be free from any of the actual AI tools. >> I'm going to say look at three notes >> and those are like the orientation [00:16:00] notes. >> So like it doesn't have to be three notes. It could be a single note, but it needs to hit three buckets. >> And so but when I whenever I start a new [00:16:09] chat co-work doesn't matter. It's hey look at the my me file. It tells it like here's who I am. look at my um vault map file. Here's how the folder structures [00:16:18] are. Here's how you create new notes. Like if I want it to be a project note versus an idea note or something, it knows the templates. So it can do all [00:16:27] that for me. I just say, "Hey, create a new project on this." >> And then the third one is the skill map. So like I want to So look at the skill [00:16:37] map. The skill map is um you know we can use skills with AI and these are like defined ones that then get to the next question. It's like what can we do? [00:16:47] >> And so what happens is there's it's like you say it's the just in time situation >> and it's like okay what's a problem that I'm now really feeling urgent about or [00:16:59] that's coming up and say let's say a little bit more about this >> and we'll put the graphic on screen. This is what you call your [00:17:06] AIOS >> which has three parts. Can you kind of briefly explain those three parts? >> Yes. AIOS is just like call it whatever you want but it's the idea that there [00:17:16] are three layers to how we can interact with AI. So at the core of it is a second brain a knowledge system an idea verse your notes. That's at the core. [00:17:28] Everyone has that. Before AI was around who those who had notes we had this AI is around. So now what do we do? Well, there's the outer layer. So we'll go [00:17:38] skip the middle layer. There's the outer layer that is claude. That's open AI chat GPT. You know what else do we have? Like any of these models that are out [00:17:48] there, Gemini, Frank, >> you're you have to launch a product called Frank now. And you plugged it so much. >> I know. That's funny. It's like my go-to [00:18:01] madeup AI model. Frank and uh so the outer layer is the least important. It's the most ephemeral. It's always changing. One one month it's cloud co-work and then something else will [00:18:12] come along and we'll go oo I want to play with that but I'm stuck in co-work. Well, let's not get stuck anymore. And that's one of the value propositions of [00:18:20] having your notes on your computer. Like the notion apps, the other apps, they are great and they're built in. So it's a little bit easier to use AI features [00:18:28] that way. But the advantage of something where you can use Obsidian to look at the notes on your computer is that you can use any AI at any time. They [00:18:38] increase their prices. Well, that's okay. You are not actually locked in. Um, so you're sort of I haven't thought of it this way before, but I feel like [00:18:47] you're inverting the whole model. The way most people like really almost everyone thinks is okay, which AI should I use? And that's at the center, even if [00:18:57] they don't realize it. Mhm. >> And then they think, "Oh, let me add some integrations or connectors or, you know, a working directory." But but just that way of thinking and speaking, [00:19:07] your data is sort of at the periphery. It's like, oh, it's plugging in and it's enriching the answers. But no, you're inverting that. You're saying your knowledge and information and data is [00:19:16] the core. And the AI is almost like these little robots that just like come in and just like plug in and interface at different points, but they're not [00:19:24] essential to the system. And doesn't that seem like how it should be? >> Yeah. >> And the reason it's not is because the tech companies, rightly so, that doesn't [00:19:34] give them the same revenue growth knowing that you could leave their platform at any time. So just incentive based, their incentive is to lock you in if they can do that. So the idea behind [00:19:46] this, and we'll get back to like the three layers, but it's digital sovereignty. like how can we actually own the land that we think on as opposed [00:19:56] to digital feudalism when we're doing all of our work on somebody else's land or in somebody else's software. >> So crucial. >> Yeah. >> So in the middle your knowledge the [00:20:09] outside layer is the tech layer the AI. So we need some way to connect the two to translate and it's not complex at all. That's what I mean just three [00:20:19] files. >> Those three that you mentioned >> yes that is the translation layer. It's just a playbook. It's an operating manual. It's hey AI it doesn't matter [00:20:25] who you are. Here's how you're going to work with me the individual. >> It's that I mean so we can talk big about it but at the end of the day just [00:20:34] you have your knowledge here. Choose any tool you want and then read these files. Now we can work together. So the closest concept that I think we have to [00:20:43] that translation layer is the master prompt which I do have here in my settings. there's this pretty extensive master prompt um that has personal context about me. But you know what I [00:20:55] noticed in your explanation is what's not in there is what you said about like how do projects start? What are the structures that we turn to? What are the best practices for [00:21:07] how we work with information? I don't have anything. I just leave that up to the AI. >> So have you found like how much of that is necessary? Aren't the LLM smart [00:21:18] enough to already know how to structure things or is it a question of taste or what? That's a that's a great question. So, and we can go a lot of different [00:21:31] directions with that. The one that I think I would go to is the more that you've now you've done a one-off project. It was a huge one [00:21:42] >> writing a book. The more that you've done your own notes and made links and connections, the more these links and connections are telling any AI tool what are the important [00:21:56] notes. Mhm. >> So that one that had 50 plus links to it, it knows that is has high gravity >> and it can interpret that intelligently and say [00:22:06] >> it has the most links. So I'm going to make sure I read this note because the bigger the vaults get, >> it's not able to read every single word. [00:22:16] >> So it can make assumptions. Assumptions are what lead to the errors or hallucinations or overconfidence and that sort of thing. And we might not always know that it's not reading every [00:22:26] bit of content. >> So that's where when we do personal organization, it we're telling the AI models what is meaningful and it can read that by the [00:22:36] amount of uh links that a specific note might have. >> Interesting. >> So it prioritizes the important notes. >> Yeah, I like that. It's so similar to [00:22:44] how Google approached the internet with page rank. >> Yes. It's like instead of having a thousands, millions of humans over there cataloging, let's infer meaning from the [00:22:54] graph, from the connections. And now our personal note collections are large enough to need that same approach. >> Yes. And not every like use case is going to need all that like complexity [00:23:05] like it can be a lot simpler and you don't have to organize. You can have zero folders too and just it depends on what the prompt is like what problem [00:23:14] that is trying to be solved. Hm. Okay. Interesting. So, we have some of the translation layer. It knows about me. It doesn't really know about my preferences for using Obsidian, but I [00:23:25] also don't have many preferences. I just kind of like took notes in it. >> I didn't really have much in the way of like structure. So, I want to get to a an [00:23:36] experiment where we which I've done some experiments, but I'm curious how you think about it. >> Mhm. to see if the AI could add that structure for me. [00:23:45] >> But before that, are there any like I was saying earlier like diagnostic exploratory like just to get a sense of to give you a sense of what [00:23:54] even is in here like in what state is it in? What what is the kind of question that you would ask for that? Oh, I would [00:24:01] say something with the term dossier, like scan all of these >> and write a dossier about me >> and because I want you to make the memb [00:24:13] like the MI file. >> So basically that's going to be the assignment. Um, and that's the best way I think is always just to start with that if you feel comfortable having an [00:24:21] AI look at all the particular notes. Now this one being project specific, I'm wondering if we want to change that at all. >> Could it be adusted just on the project? [00:24:30] Yeah, >> read through this entire Obsidian Vault, gather the context you need, and write a project dossier that summarizes everything that someone might want to know about the project that I was using [00:24:48] this vault to execute, which was my book, Life in Perspective. Nice. This might take a while, right? Yeah. I' I'd say yeah, a couple minutes. >> What do you do while the LLM is working? [00:25:07] This is an important question these days. >> Yeah. Well, I think when this is happening, I'm checking my email. I'm um checking my calendar. I'm not doing deep [00:25:16] work. Then I'm like, what else can give me a shot of dopamine? So then I go on YouTube and I'm like, oo, this is cool. Watch another video. And I'm like, where [00:25:23] was I doing? >> Do you ever run multiple sessions in parallel? I won't consider myself a power user that way. Sometimes I want to keep the [00:25:33] context very specific in this one though and then I'll spin up another one and ask a side question. I'd say not more than at three. >> Usually just if just a secondary one on [00:25:43] the side. >> What about you? >> Yeah, I find it so hard to run them in parallel. I honestly if I'm doing something substantial like this is [00:25:52] substantial I often find that I need almost the whole time that it's working to just like think my own thoughts. Do you stay here then? >> Sometimes sometimes I just like watch [00:26:03] that little orange thing go and I just kind of like ask myself questions like is this the right approach? Is this worth like meta questions? Is this worth [00:26:13] is this a good use of my time? Well, you you're good at the meta questions. I love those. I do like this coming up though as like these like [00:26:21] little reminders like, "Oh, yeah, that's right." You know, this AR checklist that might resonate in a certain way or like you know the Charlie Hoon notes [00:26:30] like, "Oh yeah, I should >> make sure like reference that material again personally." You know what? It's true. Even this isn't just eye candy. This gives you little reminders, [00:26:42] little bits of context, little clues. Yeah, >> it's true. >> They're like little shots of like I don't think dopamine in this case. I think like bursts of aliveness. [00:26:52] >> Um and whatever that is for you because it's your material and if it takes longer, it's actually it's probably doing a better job because there can be [00:27:00] some cursory scans, >> you know, um you know about like the micro wikis and Carpathy and >> and so there's a way that can be really effective. But if the vault size [00:27:12] gets beyond let's say around 500 notes and it depends on how long the notes are then again it's a situation where um what's being indexed is not with 100% [00:27:22] fidelity meaning again it's scanning notes >> and you might get poor results. So there there's always a certain >> you need an ability to um kind of [00:27:34] distill >> those layers up. So then AI can just take a fraction of the context or what I like to do is if let's imagine you had [00:27:44] multiple projects in this vault then you say >> oh I and we haven't shown this but I'll just see if I can pop back over to it. [00:27:54] So we would go into Is it okay to click on? Yeah. >> Okay. I'd go here and there are a couple hotkeys, but we can just hit command or [00:28:01] control P and I'm going to type copy which gives me the commands that use copy. So, this is something I do all the time because I want to give very [00:28:10] >> like look this is taking a while and what I want to say this time is use this as the core note >> to do your research around. [00:28:19] >> Oh, >> so I'll do copy >> using the Obsidian URL. No, I'll do system route is the most guaranteed. It's ugly, but it goes all the way to like your whole computer [00:28:30] system, but even in this case, vault folder would work. >> So, um I like to So, these are all hotkeyed. It's like command shift C or something for me. But then like if we [00:28:40] did the system route, we can see what that looks like. And now we see that final folder. And then and there's this is in no subfolder. So when we do the [00:28:51] other version which is just copy from vault folder we get something that's a lot cleaner. So sometimes if I want absolute certainty I don't want to waste tokens. I don't want [00:29:04] to waste my own time even wondering if it's going to be in the exact right place. I'll just use the um system root path and it's like start with this note [00:29:13] help build a brief around this >> and then it's just like super accurate. It's only looking at the link notes and maybe any other interesting ones. This [00:29:21] one has had to work widely and it's looking at readwise files. Some of them are related but maybe not all of them are. And so this way would just [00:29:30] give you a more targeted result >> depending on what you want. And with this the note that you refer it to up front is typically a map of content [00:29:37] file. >> Yeah, typically. >> Okay. >> Or it could it doesn't have to have a lot of links. So it might just be on research. So maybe there's some [00:29:47] angle of research, but it's like an article and then I've linked maybe one note to it, but I'm saying look at this. Here's some questions I have on this [00:29:56] line of reasoning. What do you think about X, Y, and Z? But I but I'll just have the actual um file path there so we can just start a conversation. [00:30:05] >> Smart. >> Now there are ways you can like have the third party plugins and then you can ask AI questions from inside of Obsidian. [00:30:15] >> I just I don't want these multiple ways to do things that exists. And then you know we >> Obsidian CLI I don't even want to [00:30:23] touch that. I know it's extremely powerful, but until I have time and a really good reason to do like the coding that's involved to figure it [00:30:32] out, I don't need it. Do you ever do you ever point it at a collection of notes and say make a map of content just on its own? Can it can it do that? [00:30:41] Is that does that result in a high quality map of content? >> If if the prompt's good, um it's only as good as the content of notes that you [00:30:50] have. like you can't have a map of content if you don't have any content. >> So we yeah there's this a skill a couple skills that put together we call the [00:31:02] sherpa system and essentially I say hey train on my these are my favorite 15 to 20 maps of content I've ever made. Some of them are more like an index others [00:31:13] are like a thinking space and I like you know arranged everything. Others are like all pros with all these links built in. So it can be a lot of different [00:31:20] varieties. I want you to look at all my favorite ones and then reverse engineer what makes a good map of content and then I worked with the model and I was [00:31:28] like now we have a style guide for maps of content. >> Interesting. >> Yeah. >> And that's part of the Sherpa system. >> Yes. So that's one of the skills like [00:31:37] think skill here and then the running system and then so there's that skill and then the other one is like the questionnaire. So, so what I do is I'll [00:31:46] just go to the message here and I'll say >> run the sherpa >> and then it taps into the system and then it starts the questionnaire skill. [00:31:54] It's like what topic are we learning about? >> And then what's cool is the result of that isn't just here's Wikipedia on the topic. No, it's actually connecting. [00:32:03] It's like you talked about something similar here and we're connecting all these related notes. I'm like oh cool. Now I can analyze from that perspective and make it meaningful. It's only [00:32:15] when AI does a map of content for you, it's only potential meaning >> and that can be disconnective if you don't connect it afterwards. You know what I mean? Like it's just this wave of [00:32:28] new cognitive debt. >> It's it's not helpful until you actually have the energy to make it your own. Is that kind of the difference between a map of content as you uh describe [00:32:42] it and just a table of contents? What's what's the difference between just a simple index and a map of content in your mind? >> I think a map of content is fluid and [00:32:53] and an index feels a bit more static. M and what I mean by the fluid nature of it all and what gets me really excited is that we can move text around really [00:33:04] fast now and we can link to all these other things just by one thing and we can basically work closer to the speed of thought in a single note a single [00:33:14] dedicated space that can hold all of these disperate thoughts and it's like wow finally I can just be in this space and not worry about like what am I [00:33:25] forgetting to think about. I've actually collected everything into a single note and I start to move things around quickly. So I'm the ctographer >> of my ideas [00:33:36] >> of my knowledge. And so it's not so much I want to create an artifact, but it's more let me think about this topic. How are things related? Because working [00:33:48] in this mapping way, it's not organization for organization's sake. organization in this way is ideiation >> and that's like not all organi not all organization is created equal but I feel [00:34:02] like through this sort of like mapping method that the organization is a form of ideiation >> that I love cuz I kind of skew towards organizing like more that architect [00:34:12] >> but then in this mapping method it's gardener and I'm just feeling like >> like I'm the avatar >> of like pulling on fire and air and [00:34:22] and water and earth together. >> That's how I feel like in this. So, it's like all parts of my mind activate. >> So cool. >> If someone from the outside looks at it, [00:34:30] they might go, "Oh, it kind of looks like an index, >> but they can't see the meaning that you made through the making of the map." [00:34:39] >> It's like it's like you're mapping your own relationship to the topic, which is not a it's not a fixed it's not an objective thing. It's [00:34:50] inherently subjective. It's the priorities you hold, the interpretations you impart. There's like a pieces of you in the mouth almost. >> 100%. It's like it's a meaning [00:35:03] making process. Um it can be more utilitarian, but the one the stuff that usually feels that makes me come alive is when I'm like, "Wo, this is a really [00:35:14] meaningful thing." It might just be about trying to learn the guitar, but if I'm mapping it myself, that's a lot different than, you know, reading a Wikipedia article. [00:35:25] It's it's um it's personal. >> Totally. Okay. So, this took maybe like 7 to 10 minutes, something like that. Um let's open this up. So we have the title [00:35:39] what it what this is which is accurate uh a snapshot of the project. >> That is a really good imprint. >> Yes. One of the best. Um this is all correct. [00:35:54] Correct. It's drawing on my master prompt for broader context. Oh >> what the book is about. Wow. This is like a fantastic summary structure, publishing and business track, key [00:36:09] people, research, foundation, source material, status, key files. Yeah, these are the two. The the content, which is sort of a massive outline of every single idea that I [00:36:26] wanted to include in the book, and then the outline, which is those ideas put into a book form. So what is what would you do with this? [00:36:34] What is this? I would not read it from here other than to check to see if it was working because >> I create a system and what it would say [00:36:42] at the end of this is great. I've put this in your AI folder in your note system. >> Yeah. So we can talk about the boundaries of [00:36:51] where notes go, but as part of the master prompt or the vault map sort of the orient the translation layer um I would have a folder and it would be some [00:37:00] version of yeah I would just call it AIOS like that >> and then within that um depending you might have a couple other folders but I just have a place called history. So [00:37:13] anything that's created it I just know goes in AIOS history. >> So it'd be like hey I've just put the file in there and it would you know it [00:37:21] would give it the time stamp first so we could always just sort by date and then it would give it the title. And so there's just this running you know [00:37:28] history of notes it's created and this is crucial. There's a very clear line of separation between my notes, my thoughts, and then where AI's notes and thoughts go. [00:37:41] >> And >> and that's the AIOS folder. Only AI stuff goes in there and nowhere else. >> That's the only place it can write to that I've given it instructions to write [00:37:51] to. >> With the one exception being if I say, "Hey, create a new idea note um using, you know, the template and all that stuff that you know, [00:37:59] >> then it knows where to put the note." >> Okay. But um yeah, when there's anything that's juicy and it's creating a new note because we're researching [00:38:07] something or in this case we're doing a dossier, it knows just AIOS history. Never have to think about it after that. >> Okay, perfect. I like seeing how you do [00:38:17] that. So let's say I like your point too about it's true. If you don't save this proactively, it's gone. It's it's you'll never see it again >> because Claude all the AIs love to just [00:38:29] create artifacts upon artifacts, but like it's in a chat interface. So the chat goes up, you know, think about a WhatsApp group or [00:38:38] a group chat. Like do you know how I guess you know how to find the files or photos that may have been shared, but it's not easy. So [00:38:47] how do you typically do that? Do you just copy here? >> No, no, no. We could I'm like, "No, don't do it." Uh, can if we exit? Okay, [00:38:55] we'll go back here. So, what I do is I would say um, great. Looks good. Save this to the AIOS/History folder. Prepend it with the timestamp, year, month, date, and then the title. [00:39:14] And this becomes a skill that's done automatically. You know, it knows to do this, but let's see if it if we can oneshot it and if it knows what to [00:39:22] do. And if it doesn't know what to do because it doesn't have the full file path, >> we can then take it from the system route. >> The folder exists. [00:39:31] >> All right, we're in the right spot. >> Cool. >> And and what is the advantage of doing it that way versus just copy paste? >> Like this is the automated way to do it. [00:39:42] I only am giving it this instruction once. But what you want to do is it becomes a skill. It's part of one of it's part of the master prompt, let's [00:39:49] say. and it knows, okay, it will ask, hey, should I save this to your vault? And you just go, yep, and go, okay, save to AIOS history. [00:39:58] >> So, what is the line that you have there? Is it is it just a bullet point that says every time you create an artifact, ask me where to put it and [00:40:06] wait for my confirmation? Something like that. >> You know, I want it to assume that it's just going to go to AIOS history, >> everything. But there are sometimes it [00:40:14] might create an artifact before I kind of asked for it to and then it's like I created this if you want I can save it and then that's when it's sort of asking [00:40:24] but other times it just knows immediately this is where it's going to go if I've asked it to create something. >> Okay. So >> so there's no friction. [00:40:31] >> Okay, this makes sense. So this is this history folder is the default save location of all artifacts created by Claude. So if you ever wonder where it is, it's not in anthropic [00:40:45] server somewhere. It is in your own owned and controlled but segmented part of your vault 100%. And then any of these other folders, doesn't matter what they are, but that's where all my other [00:40:56] information, all my other knowledge lives. So there's a clear separation. If I ever feel there's too much noise in the history folder, uh especially like after years of these things that really [00:41:07] have zero value, it's that easy to clean it. You just open it up and you say, I'm going to clean all these the po these previous years. I'll put them in the [00:41:16] deep storage in the archive and don't have to worry about them. Once we have this file here, what's here I'm really what I'm really looking for and what [00:41:25] gets me the most excited is to see what it linked to. Mhm. >> So once the notes open, I want to see in the sidebar if it did any back links for [00:41:33] us. It didn't. So in the prompt that I have, it will connect things. But and that's what I would say is like as you create this dossier, you know, make sure [00:41:42] it's connecting to what you think are the most valuable notes. >> I could have it do that, right? Why not? >> Could I say? >> Yeah. >> Now take this new project dossier note [00:41:55] that you've created. find every single term inside of it with a corresponding existing note and link it to that note. Anything else? Um, we could I like going with [00:42:11] that. If we did anything else, it'd be and create a little map of content section if you're familiar with that term um for these links. >> I think it will be [00:42:20] >> okay. Also add a new section to the project dossier where you create a map of content, an MOC. Use all of the advice and recommendations issued by [00:42:34] Nick Milo for that. >> Who that? That'd be insulting. This is a great task. It likes this task. So, while this is working, what would you do next? Like, [00:42:52] I'm trying to It would be interesting to get to a point where we're doing something that I couldn't or wouldn't do myself. Like, cuz I can read [00:43:03] through that project dossier and I don't actually that didn't exist before, but I kind of know it. Like, there's no surprises there. What is something that's surprising or that is [00:43:12] disproportionately valuable? >> I'll think about it through two lenses. Maybe I'll stumble into something. one would be through this project lens and then the other might just be more [00:43:22] broadly. So if I'm thinking through this uh project lens u making assumptions or something I mean you're at the process where the book's written >> it's they're pre-sales already so it's [00:43:36] really far along. >> Mhm. >> Um well that kind of suggests something now we're starting on the marketing. >> Okay. So I wonder which is a whole it's [00:43:49] it's an endeavor at least as challenging as writing the book itself. So how could I and we do have a marketing plan. We have a marketing strategy and stuff. [00:43:56] >> This would be good though. Yeah. >> But I almost wonder what's interesting here is this is the full content of the book and even beyond what [00:44:04] made it into the book. So, I wonder if we could do something like give it the marketing plan and say what existing content in my notes could be could fuel [00:44:16] these efforts or something. >> You already have a marketing plan. >> Uh-huh. >> I like that a lot. Um, and just depending on where that marketing plan [00:44:27] goes and where it doesn't go, one thing I'd be really curious about is based on everything from our existing audience building a second brain and everything in between, the [00:44:38] testimonials, the people who really loved it, what's going to resonate the most with them from this? And that's probably part of there in some way. Um, but the sort of there are a few [00:44:50] different things that AI can do uniquely well in a complex project like this. But one is going to be to analyze a huge amount of data. And this is like the [00:45:00] imagination. It's like, oh well, without this tool, we wouldn't be able to do that. And we were just trying to like feel it out and just gut instinct it. [00:45:08] But this way we can actually have a more datadriven sense of who are all these individuals in an aggregate and like are there avatars that we can put [00:45:17] on them? >> Second mention of avatar. >> But um like who are these individuals? Are there any surprises? Are there any surprises in the data? Are there certain ways to reach out? Um, [00:45:33] and you might have you will have more ideas based on what's already in your marketing plan. >> But, um, I think the marketing side is where AI can really come alive is [00:45:44] since I have so much source material. I just don't know how that source material can be sort of routed or translated to marketing [00:45:53] efforts. But that seems like a great task for AI because it's not wholesale invention. It's really just mapping. This goes to this. Mhm. >> Should we try it? [00:46:03] >> I think so. >> So, verified. Okay. Let me copy into the vault. 84 link targets to resolve. So, if we go back, I'm curious if there's if the dossier has some links in [00:46:18] it. >> Oh, it actually says 180 wiki links. A new section map of content. Okay. Couldn't write directly to the vault folder. So, I built the file in my workspace and [00:46:31] copied it into via the shell. >> Interesting. >> Not sure what that >> That's worryingly creative. >> Uh, okay. Oh, so they just appear all these links. [00:46:46] >> Oh, and because they're underlined, that means that they're probably external links to like an Evernote file is my guess. >> So, these seem kind of like literal. [00:46:55] There's not a lot of like creative interpretation. It's like if I mention a book, it links to the book. If I mention quotes from a person, it links to those [00:47:01] quotes. >> There may be a new section, though. Let's see. Oh, yeah. There it is. >> So, let's analyze this. How do you feel about this map of content? I think we'll [00:47:15] both have feelings because it's your material, which I don't am not familiar with the new book. Actually think it'll be very helpful for future marketing. you've just given the [00:47:25] AI on the next project, >> not the book writing, but the book igniting >> um a map of content for it to like start to compare like where to grab this. Now, [00:47:36] you you'd definitely want to have the manuscript clean in there if it's not already. >> It's like this is the key source. This is at the very top of the map. [00:47:44] >> Reference this >> and have other specific directions. cut up this, you know, like h have um a list of topics that can be, you know, whatever [00:47:55] medium you think is best. Like a newsletter issue where I focus on um Iikitu, >> if I'm saying that correctly. >> I don't know. What do you think? [00:48:05] >> I can see how these sections will be very useful for um our marketing. But what strikes me in your comments too is like there's no objective standard of [00:48:14] what makes a map of content. It's not it's not pure science. It's what it means to you. It's how it makes you feel. It's how it what it reveals [00:48:23] to you personally. Right. Absolutely. And and so the goal is that this shouldn't just be valuable to you out of the gates. It might be to AI in a future [00:48:36] project. As I'm making that organizational decision, it's an ideational outcome >> because I'm like, yes, now I'm giving context to all of these >> concepts. M >> so that I can understand them better [00:48:48] which is like how I might communicate these things >> and it's just it's all personal and it might not be needed. I find it really valuable. This is like how I make sense [00:49:00] of all you know like we get over 20 things. I want to know what's its relationship. What is there an order to it? Yeah. It's like it's like you take [00:49:09] notes. Taking notes is always kind of bottom up. You're just capturing what resonates and what you like. Um but then that implies some later stage where you [00:49:18] come in and you group creates some structure like make some sense out of it. Mhm. >> Um I kind of wonder if we can So it's [00:49:27] this is kind of interesting to notice is um if we go to the project folder uh right there. Um so here is the launch of marketing plan [00:49:46] um which was compiled very with a lot of time and effort and manually. So I'm going to download this as markdown uh current tab and first actually I feel like I want to [00:50:03] I want to save it in obsidian. How do you usually what is your go-to? There's so many different ways to do it, but what is your favorite way if you [00:50:10] just have a free floating markdown file to add it to Obsidian? >> I like to have the sidebar open so I can see. And I'll just drag and drop now to [00:50:19] the file explorer section. Yeah, right there. And now we see that it's I'll just do the drag and drop that way. >> Okay. >> Yeah. So, in a funny way, we now have um [00:50:31] my launch and marketing plan and its launch and marketing plan. So, I almost want to like compare them and contrast them. >> We don't have its plan yet. Or have we? [00:50:41] Did we make it? >> Well, not a plan, but this is almost kind of like a at least like a brainstorm of ideas. >> Does it cover the marketing? Like you'll [00:50:49] know better than me with >> Yeah. Like all the marketing and launch stuff is in the vault as well. >> So, it has publishing and business, [00:50:58] marketing bonuses, companion product, live program. That is the upsell. Hm. So, what it's missing is just like I guess some it has a lot, but some of my [00:51:09] very specific opinions on the launch, but I almost want to know like it be it'd be just interesting to know how it would launch this based solely on my [00:51:19] source material cuz it might come up the models are smart enough now. It might have like a whole other approach to a because I'm doing kind of like [00:51:27] the most standard like the classic launch playbook. I'm almost curious if it would come up with a different approach. If you if that's what's most curious, go [00:51:35] for it. >> And downstream after you did that one, then you could say, "Now look at knowing what you know now, look at my launch plan over here and say what works and [00:51:49] what is not good at all." Like be argumentative, have strong opinions >> and have alternatives suggested. >> Okay? you know, but that would be like the downstream after it gets a [00:52:02] sense of what you're suggesting. >> So, there we go. >> Let's copy I just learned that from you. Let's copy the system route and say look at this note. [00:52:16] This is my current marketing and launch plan. Compare what I am planning here with the raw material related to marketing and promotion that you have found in this vault. Is there anything [00:52:30] I'm missing? Any gaps or blind spots? Anything that the source material suggests that I haven't thought of or that I'm undervaluing or not taking into account? It's it's going back and doing [00:52:42] a wide search of the whole vault. It looks like actually this is a question I have is like I find that everything that AI does is summarizing anytime it looks [00:52:52] like looks at a lot of information and writes anything or creates any artifact it's a summary like you mentioned it's always a subset uh and it's often a low [00:53:00] like 10% 5% 1% I often ask it what percentage of the notes did you actually look at 0.5%. But then once I have this maybe that's a good distillation but I often [00:53:13] find a summary of a summary is not good. The more times you summarize all the specificity and the signal gets stripped out of it. >> Yes. >> And so I really don't trust summaries of [00:53:23] summaries. The problem with that is it limits the usefulness of artifacts like this because if I don't if I never refer to this project dossier for any further [00:53:33] step then what was like what was the point of creating it? It doesn't it doesn't quite feel like a compounding asset. In some ways, it's like a [00:53:41] depreciating asset because of that. >> Yeah. It It's almost a distillation that's lost its soul. >> Yeah. And I think if the distillation can have cited sources, you [00:53:55] know, like references to killer lines, you know, something that really grounds it in the specifics. Um I think those instructions can kind of help it [00:54:03] stay grounded as oo so it doesn't get lost in like the platitudes as this is happening. >> So you can give it some guidance to reduce that tendency. Cite your sources. [00:54:15] >> Keep details. Use my verbatim >> use. Yeah. Absolute verbatim quotes. >> And there's probably a few others a bunch of others that we could come up [00:54:23] with. Are those guidelines in the templates and the prompts that you you've designed? >> Mhm. No, absolutely. The big one that's like the only giant call out box is do [00:54:33] not rewrite me >> because I want to know I mean if I'm asking for feedback or whatever it can give feedback but I don't want it to [00:54:42] rewrite my own words when I'm asking like what have I said in the past and well you said this it's like I know for a fact I haven't said that now what [00:54:49] happened there it's like well I summarized like >> I don't want the summary so we have to be really specific um so it knows how it can be helpful. [00:54:58] So, the useful work here isn't cataloging overlaps. It's the handful of things your own source material pushes hard on that the plan underweights or emits. Interesting. >> Blind spots. Great. [00:55:10] >> Interesting. >> So, it's saying you're naturally drawn to the contemplative side. >> Yes. >> Knowing it's a book that needs to reach a broad audience. It the productivity [00:55:22] side is really important as well. >> Yes. >> Is that the tension that it's referring to? So, I think it's it seems like it's picking up on I had a list of like [00:55:33] the emails in the email series that I want to write. One email would be dedicated to why the annual review is the missing piece of your second brain. [00:55:40] Another one is the bridge from BASB to the annual review. >> Here's a question. Is it overindexing on one piece of information? >> It is. It is, which it does a lot, [00:55:53] >> which is great. You bring your own discernment. This is not truth. This is not reality. And that's a that's this is actually a great point because when this [00:56:01] happens, we have to say, well, this is far from true. But is there anything useful? This is overweing on the fact that the list of email topics [00:56:12] that I wanted to write on was very brief and incomplete. >> Not most or many of the emails will be about these things. They'll be about the [00:56:21] annual review itself, but those are kind of obvious, so I didn't list them. So this is such a funny thing. It's like I think that the tendency is [00:56:31] AI always assumes that whatever information it has access to is all the information that exists. >> You know what I mean? It's like it's like every time you put in a [00:56:41] prompt, you give it some working files, it's like a little universe, a closed universe that you've created. And it's kind of understandable like it can only know what it knows. it can it only knows [00:56:52] about things in that little pocket universe. It doesn't know about all the other universes. Mhm. >> So I think it's important to just think what you said is like what is [00:57:02] it overindexing on and how can we build in rules in the future like hey AI is great and a lot of people that are going to talk about it [00:57:12] are going to say the amazing things it can do but it does take tinkering and tweaking and so in this case how can we prevent overweaiting [00:57:22] overindexing of a blind spot? No, but I think you're right cuz what's so interesting is despite its misconceptions in a way. This is a good [00:57:33] point that if I try to launch this book as what I did before but for your year and it gets specific, you have three costs. You attract productivity readers who [00:57:46] will find the book softer and stranger than promised. I think that's a real risk. That's going to happen actually >> when you've written two books on one [00:57:54] topic and then you kind of like grad like shift to something that's related but different. That's going to happen. >> Uh you blur the exact thing that makes [00:58:02] it unconable. I don't know what it means by that exactly. And interesting it found this idea of the wisdom worker my most ownable idea. I don't think I agree with that. [00:58:15] Sitting unused in the outline. Okay. Yeah. So, audiobook has no tactics. We just haven't gotten to that yet. January is your Super Bowl in its thinnest. That's because that's the very last [00:58:25] stage of the launch and we just haven't gotten to it. Um, it's just it's pointing to real holes, but holes that we're aware of. Middling results, not [00:58:32] the most exciting for this particular effort. >> Yeah. I mean, part of what this is uh pointing to also is that it needs more context. Um and the fact that we [00:58:44] had to go to Google Drive to you know download something like >> I tend to really I put a lot of effort into centralizing things. I always want [00:58:52] one source of truth but despite that my information is definitely scattered across >> probably eight to 10 different platforms. Do you ever this must be the same to some extent. What do you do when [00:59:06] you realize the answer is not good because it needs context that's over in this other place? >> We haven't covered it, but using a tool like Claude, we can connect it to, you [00:59:17] know, to Google Drive to ClickUp and so we can connect to other tools and then we can get better feedback. But we have to point to, hey, check out this [00:59:27] folder. This has I think this is where the marketing plan is. um you know include this conversation over here in the project management software that we use. So it's again it's just like that [00:59:39] becomes a meta skill with AI. It's like it to if it doesn't have the context it's going to make poor decisions. So you have to become really [00:59:49] great at pulling it all into one similar place. >> Yeah, it is indeed a skill and not a cloud skill. >> No. >> Well, that too, but [00:59:58] >> yeah. Um, Nick, this has been awesome. Wow, I learned so many different things. If people like what they heard today and they want to learn from you and your [01:00:06] team, uh, or buy one of your products, how do they do that? I think the real value of linking your AI is everyone is trying to learn about AI, [01:00:20] but really we need to teach AI to learn about us. And when we can point it at our notes and give it the translation layer that the manual like here's how to [01:00:30] interpret and work with me, it becomes really powerful. And what's more, it becomes futurep proof and it doesn't matter what the AI is because we own the core. So important. Highly [01:00:43] recommend you check it out. There's no one on the planet that knows more about this stuff than Nick. And uh I hope we get a chance to collaborate again. [01:00:51] >> Me too. Thanks, Diego. Well, that was nerdy. Every video is like that. Watch how I made a million dollars in 7 minutes with Claude Fable. It's like you're just not [01:01:07] believable on the surface. I think that's a big part of both of our sort of brands and reputations is like not trying to manufacture certainty where it doesn't exist., Nick Milo Reads My Obsidian Vault, Tiago Forte, 01m270c5jvjqn4ydfcrkv1jd10.