OpenClaw Crash Course For Beginners

Hey guys. So, you've probably heard of Open Claw by now, which is essentially your own personal AI assistant that can integrate into just about any aspect of your life. And there's been a ton of hype around it. I'm not one to to ride hype waves. That's why I'm never really the first to release a video on anything. I like to actually use the product in a real world scenario and and kind of get a feel for it before making any videos. So, I've been using OpenClaw for a while now, both locally and on a VPS. I think there's reasons and advantages for both, and I'll go over those. But what I want to do is instead of just making some hype video, I want to show you a complete walkthrough of um not only getting set up with OpenClaw, but also create your own workflow similar kind of similar to what I have with my uh content research and creation. So basically we'll have openclaw reach out get us the latest webdev and AI news and tools and all that and create a digest for us and we'll have it do that with a cron job so it happens automatically and that's just one very simple use case. I'll talk about some some of my other use cases. uh and take a look at the timestamps because after setup I'm going to show you some cool stuff that I've been doing personally and give you kind of an idea of of my own AI workflow and uh and how I have agents talking to each other and stuff like that. Um but before we do any of that, I want to start off by talking about exactly what Open Claw is and how it works. So, I want to start off just talking about what OpenClaw is. And this is mostly for those of you that may be a little overwhelmed with the amount of AI tools that are being released, but you keep hearing about OpenClaw and you want to look more into it and know what it is. So, it's essentially what I said, it's an AI assistant and it's a self-hosted gateway. Let me just grab my laser pointer here. So, it's a self-hosted gateway that uses chat apps such as Telegram, WhatsApp, Discord, Slack. There's a whole bunch of them and you can use these to basically route messages to an AI model provider. So, basically, you install this open claw gateway and that could be locally whether that's on your the machine you're using, your laptop, whatever, or a server that you have at your home. Uh, or you can install it on a VPS in the cloud. And there's pros and cons to both of those, and I'm going to go over what those are in a few minutes. So, basically, you have this gateway. Uh, it routes your messages, it runs the agent loop, and then has access to models. Now, as far as the models, you can use GPT, you can use the anthropic models, you can use um DeepSeek 4, which was just released, GPT 5.5, you can use um Olama. So, if you have local models, you can also use those. Gemini, just about anything really. And you have to keep in mind that the the better the model that you use, the smarter your Open Claw. I've tried using it with some really shitty local models and it wasn't a good experience. It just took forever to reply and it would just keep asking me the same questions over and over. So, you want to make sure that you you use a good model and and of course you can test it out with different ones. Now, one one thing I do want to mention is with Anthropic, you can no longer use your subscription, which really sucks because that's what I was using at first was Opus 4.7, but now uh Anthropic doesn't want you using the subscription with outside tools. So, they basically banned that. Um the good news is you can still use open your OpenAI subscription. So, whether you have the the $20, you know, chat GPT subscription or $100, $200, you can use that. And then of course you could still use the API key with anthropic but that's going to cost you quite a bit uh depending on how much you use it and what you do. Now in addition to communicating with these models it also has access to all kinds of tools. It has total access to the file system or the files that you allow it. Uh it can read and write files. You can it can write code. And as far as writing code, I've been using it for a lot of local tools. like I had it create this um this mission control. I can't show you guys a lot of this stuff because a lot of it's personal like I have sponsors and stuff like that but it created this whole mission control system which is tightly integrated with my network and I can have u tasks where I can assign Travis is actually the name of my local OpenClaw. I can through cla code or co-work assign Travis a task and he'll pick it up claim it and do it. So I can have these agents, you know, doing different things in the background, which is really cool. And I'll towards the end of the video, I'll show you a little more of that workflow. So in addition to that, it can uh it can run web search. There's different providers like for instance the Brave search API. You can use that and and other ones. It can connect with MCPs, run scripts. It can you work with your Google calendar or whatever kind of calendar you have. And that's one of the the great things about it is the level of integration. You can integrate it with any Google service, Dropbox, Notion, pretty much any popular uh platform has integrations usually through what's called a skill, which I'll get into in a little bit. Now, with that said, since OpenClaw can do just about anything a human can, you have to be very wary of security, right? you have to um you have to pay attention to what you let allow it to use, what it can access, things like that. And I'll talk a little bit more about security uh considerations in a little bit. So, let's quickly talk about where to install OpenClaw because you can install it locally on your own machine or your own server or on a VPS in the cloud and there's cases for both and that's what I want to talk about. Now, I'm always 100% upfront and and transparent with you guys. If you've watched me for a while, you know that. And Hostinger is sponsoring this particular video and they offer managed OpenClaw VPS's, but that doesn't sway my advice or anything that I would say whatsoever. And they know that they wouldn't want it any other way. And I've been working with them for years. So really the case for VPS is it's just ready out of the box. You don't have to deal with installation. You don't have to deal with um you know your system going down and all that. It runs 24/7. doesn't matter about your Wi-Fi, doesn't matter about your your power, no power bill costs, etc. But the case for local, you have obviously more control. It's on your machine. You have full control of that machine. You have a little more privacy. Uh if you want it to access your network and basically work as like your network admin like I do with my OpenClaw Travis, then you probably want to go local. If you're running local AI models, maybe you have a really beefy machine with a GPU, then you might want to just install OpenClaw, you know, locally as well. Um, so it's really, it really just comes down to what you're using it for. And I'd say bottom line, local is your hardware, your rules, but a VPS is always on, reachable from anywhere, and you're going to get that reliability and support. Now, just to show you to give you an idea, uh, Hostinger has a couple packages. the managed openclaw. Basically, you don't have to do anything. You go through a couple UI steps in the hostinger panel and you just have access to your dashboard. With this version that the V regular VPS, you SSH into it and you install it just like you would on your local machine, right? And you have much more control. So, it's really up to convenience. And just to show you, if I have the manage package, I just click on OpenClaw, click it here, and it just brings me right to the dashboard. I didn't have to go through any weird installations or anything like that and I can just chat. Now, usually you're going to use like Telegram or Discord or something to communicate with your OpenClaw, but the dashboard does give you a chat section and I can say like what what can you do, right? And it's going to basically do the same thing as if I were talking to it through open I'm sorry, through Telegram or or any of those tools. So it says a lot uh broadly in four buckets. Answer questions, explain things, access your files and code, debug, refactor. So I can I can interact with it from here just like I could anywhere else. Now with a regular VPS, you would come over to VPS manage and then you could just SSH into it from your terminal. You can even use the the integrated terminal from here. And from here, I could go ahead and run the curl command to install and set up OpenClaw, which is what we're going to do now. All right, guys. So, I just want to show you how to get set up, how to get up and running with OpenClaw. I'm going to install it locally on this Mac. My local OpenClaw, Travis, is actually running on my server that's in my basement, my server rack, but I'm going to do a fresh install from here. Now, if you're using a a regular VPS that you have SSH access to, you would install it the same way. So, basically, you're going to go to openclaw.ai. You're going to grab this command right here, this curl command. And we're going to just open up a terminal. And I'm going to minimize that and that. And then just paste in that curl command. And it's going to do a few things. It's going to check to see if Node.js is installed. If it isn't, it will install it. It's going to check to see if git is installed and then install OpenClaw. All right. And then the first screen you're going to see in the onboarding process is this security disclaimer because obviously open claw is very powerful. You have to be careful of what you give it access to. So it's just basically giving you some recommended baselines telling you you can run this openclaw security audit command. So you just basically need to accept that um those terms. And then setup mode manual just gives you a couple extra options like port numbers and stuff like that. I'm going to choose quick start. And then this is obviously really important. This is your model selection. So if money wasn't an issue, I would want to use Anthropic. I think Opus 4.7 is is the best model to use for OpenClaw. But like I said earlier, you can't use your claw subscription anymore. You have to use an API key, which gets very expensive very quickly if you use it h how I do, which is just all day for literally for everything. I just used it at the gym for logging my my workout. So, um, for me, the next best option is to use my codeex or open AAI subscription. As you can see here, you can do open AAI, which is the the API key, and then codeex, which is your subscription. It could be the $20, the 100, or the 200. I have the $100, and it seems to last me, you know, as long as I need it. And then there's also other models like uh Deepseek 4 was just released. You can use that. You can use Gemini. You can use um there are some free models from like Open Router. You can use local models through like Olama, LM Studio, Hugging Face, but I've used some of those some of the free models and they just it didn't work out for me. It was just it just took way too long to answer and it just was it was an idiot. So, the the best for me is the is uh GPT 5.5 right now. There's also Miniax, which it actually says recommended. So, I've heard decent things about it. And I know there's a $10 subscription, which I think you can use with this. So, just know if you want your OpenClaw to be efficient, you want it to work well, you're probably going to have to spend a little bit of money on models. Um, I would, again, I would suggest the OpenAI codeex subscription, the $20 subscription. It's definitely worth it. So, I'm going to choose that. And then we don't have an API key, so we need to log in through the browser. So, you just need to authenticate. Authenticate through Google and just authorize. Click continue. And it says authentication successful. So, I'm going to jump back over here. Now, you'll probably see a list of GPT models. Just make sure that the one you pick, if you if you have a subscription, make sure it's OpenAI-CEX whatever. It could be 5.4, 5.3, any of that stuff, but make sure it says codeex. If it doesn't, then it's going to fail. It's going to expect an API key. So, I'm going to choose 5.5 codecs. And then you want to choose your chat uh what you want to use for chat. So, for me, I'm using Discord, but I would suggest starting with Telegram, which is what I want to do in this video. I like Discord because you can have separate channels where you can actually spawn sub agents and you can have different contacts for different channels. But Telegram is very simple. It's just like a onetoone chat. So I'm going to use that. Use local plugin and then we're going to select enter telegram bot token. Now to get your telegram bot obviously you need telegram. So if you don't have it in uh install it on your on your mobile device and then you can download the desktop app if you want and you can go through that which is what I'm going to do. All right. So you can see I have my Travis local uh uh open claw. What you want to do, the first thing is atbotfather. And it's going to be this guy right here, this top one. Make sure it matches this exactly, botfather. Click on that. Click start. And then you're going to add a new bot. So right here, new bot. And then you're going to give it a name. So I'll call mine um I'll call it uh Dave. Okay. So that's my name. Now, a username you have to pick that ends with underscorebot. So, I'll say Brad's Brad's Davebot. Okay, now it's and that has to be unique. No one else can have it. And then you have your token, which you can copy now. And you can paste here. Just make sure you don't give that to anybody. Don't let anybody see it. Uh, I don't care if you guys see this because I'm going to delete it after, but I'll probably still blur it out. And then for a search provider, a lot of these need API keys. I use Brave, but I do have an API key. I think the search XNG doesn't need one. So, we're going to choose that. Just hit enter again. And then for skills, I'm not going to configure skills now. Um, skills are used basically to package things together and you can have integrations like Gmail and Dropbox and stuff, but I'm going to say no to that for now. And then skip hooks for now. And then it should start the gateway. So it basically just runs a service. It runs as a damon on your machine in the background. And there's commands to start and stop. So uh right here I have the open claw dev sheet which I created. So I'll have this in the description. And this has a lot of helpful commands including how to start and stop and restart the gateway, how to get your status, how to get logs, so open claw logs, how to get to the dashboard, which we'll do in a second. the onboarding process. What we're doing right now is called is called onboarding. And if you get kicked out for some reason and you want to start over, you can just do OpenClaw onboard. Or if you want to go to specific parts of it, like let's say you want to just go to the model and configure that, you can do open claw configure dash section model. And that will um you know, that'll bring you just to the model part. And then let's see, you have more configurations down here. if you want to configure a different model like sonnet. And then let's see, we'll get into this in a little bit, the workspace files and all that. So, as far as this right here, you can hatch in the terminal. So, you can actually use OpenClaw through the terminal or through the web UI, which is the dashboard, which I actually want to open up manually. So, I'm just going to say do this later. And then this should just kind of kick us out of the onboarding process. Okay, so it kicked us out. Now we can just say opencloud dashboard and that will open the the web UI and this is the same web UI that I showed you in the hosting our manage plan. Right? So you just go through the hostinger UI and it pops you right into this. But now we have this installed locally after going through that process. And you can chat from here. I wouldn't recommend it. It's it's not it's pretty slow and just it's not a good experience. you definitely want to use Telegram or Discord or something. But you have other options here. You can see your sessions. You can see any cron jobs that you want to get set up. Um your agents, your models, and so on. But just to show you this works, I'll just say hello. Okay. So, it says, "Hey, I just came online. Who am I? Who are you?" So, that's always going to be the first message you get because it doesn't know who it is or what it what it's supposed to do. It doesn't know who I am. So, it needs some information. Now, I could tell it through here, but like I said, I don't want to use this. I want to use um uh Telegram. So, I'm going to open that up. And then I'm going to search for my bot, which I created, which I called at Brads Dave_bot. And you just want to make sure it matches exactly what you used. And then I'm going to click start. Now the first screen we're going to see here is to pair our telegram with Dave which with your bot which is this command here openclaw pairing approve telegram and then your ID. So we're going to copy that and then we're going to jump to the terminal and we're going to paste that in here. And then you should see approved. Once you see that, let's just clear that up. Once you see that, you can come back over here and uh I just want to clear this. I just want to clear this chat, this history. And then I'm going to just say hello. And you can see it says Dave typing. Now, this stuff here where it says working, it just it shows the tool calls and and like what it's doing, what it's thinking. You can shut that off. Um, I'll get to that in a little bit, but it says, "Hey, I just came online. Who am I and who are you? I know your Telegram name shows as Brad Traversy, so I can start there, but I need your help shaping the rest. What should you call me? What kind of creature am I?" So, if you want to make it like a freaking I don't know, a rabbit or something, you can do that. Um, now I could just type something in here, but I want to show you the the core files. So, I have those here. And basically there's you have a workspace which is just a folder and you have these files there. User MD is who you are. What do you want it to call you? Um identity is who it is, what its name is. And then soulm is basically its personality, its vibe. Do you want it to be just, you know, straightforward? You want it to be funny and witty, whatever you want its personality to be. Then you have memory MD, which is the persistent memory across sessions. agents MD which is like it's instructions and rules and purpose and then tools is just available tool notes. Heartbeat MD is is uh periodic background tasks so you can have you know u uh scheduled events and things like that. So these files reside in your workspace which by default is this home home directory.openclaw/workspace. If we look in there, you can see those files. There's also a state folder and it'll usually create a memory folder as well. Now, what I want to do is actually open that up in VS Code. So, we can see the the actual file content. So, if we look at user, you can see it doesn't have anything about me here. Doesn't have my name. Uh identity doesn't have its own information. So, what I'm going to do is tell it. And I actually have some prompts. I'm probably not going to use all these, but I have some prompts here that I want to just grab and use. Okay. So, I'm going to replace this name with with its name, which is Dave. And then this right here, you and I'll have these prompts in the description if you want to use them. It'll say I'm Brad. All right. So, what it says is you are Dave, my my AI research assistant. I'm Brad. You are a calm, sharp, practical research assistant. Keep the tone concise, useful, and not hypy. Your job is to help me track web development and AI AI news, summarize what matters, and turn it into content ideas. So, that's what I want the main purpose of this specific open claw to be, but you can have it be absolutely anything. You can have it be your gym buddy. You can have it be just your all-in-one life manager, which is what Travis is, my OpenClaw, my local OpenClaw. But, uh, you can just define what it is here, and you can always change it up as well. So, yeah, what this is going to do is is actually write to those files that I just showed you. So, it says, "Got it, Brad. I'm Dave. Comot practical, etc." And now, if I go back to VS Code and I show you these files, you can see an identity. He's got Dave. His creature is an AI research assistant. Comsh shot practical. So he'll use these qualities when communicating. In user MD, we have my name. So what to call me, Brad, where I am, my time zone. Brad wants Dave to be a calm, sharp practical uh research assistant. And then soul is its basically its personality. So be genuinely helpful, not performatively helpful, have opinions. And you can edit these files directly if you want, but you can just also just talk to it. I have some other some other uh commands I wanted to do here. I want to just create a simple little workflow and then I'll just kind of talk about more just better things you can do with it and a little bit about my own setup. So let's jump back in here and I'm going to paste this in. So it says let's set up a news gathering workflow. Create a feeds.md file for the sources to monitor, a topics MD for for relevance filters, a digest folder for or digest folder for daily markdown digest and an ideas MD file to store distilled content opportunities. Make the workflow focused on daily tech research and content idea generation for developer focused creators. So, basically, I'm telling it that I want it to reach out and and have specific resources where it can get me the latest AI news or web development news. So, obviously, this is geared towards a content creator, whether you're a YouTuber, uh a blogger, a newswriter, whatever, podcaster, it's going to get you the latest stuff. And you can just tell it to run this every day, or you can set it on a scheduled task on a cron job. I'll show you how to do that. and you'll just have it, you know, every morning. And that's something very simple. You can take it much further than that. You can have Dave or or Travis or whoever your OpenClaw is create an extensive application, a dashboard um or even a a a public-f facing blog uh blog website with articles that it generates based on that content. You can have it send emails. You can have it do anything. So, it it looks like it created these files and it's going to be in the workspace. So if I go to VS Code, you can see now there's a feeds MD. So this is all the the where it's going to get its data from. MDN blog, web.dev, uh Chrome developers blog, Netlefi. So all very reputable sources, Typescript blog, Noode.js. So if anything new is released on these these feeds, then it's going to know about it and it's going to put it in this digest folder. Um let's see, we also have topics. So this is the topics it's going to look for. So uh JavaScript, TypeScript, Node.js, security issues for developers. Lower priority would be like pure enterprise news content formats to consider 5 to 10 minute news breakdown, beginner friendly explainer. So obviously you can edit this if you want uh in the way that you want, but let's look at these prompts. I think we can just ask for a digest now. I think we have we give we've given it enough information. So, I'm going to say give me today's digest. So, it should reach out to those sources, get me whatever it thinks that I need and uh and give me create a dated digest file in that folder. So, he's saying I'll scan the main source categories and save a digest for today. And again, these these tool calls that you're seeing, those can be hidden. We can ask Dave to to not show those. There is a config setting. I just forget what it is. All right. So, today's digest is done and saved in this location here. Top items, you have OpenAI Symphony, OpenAI uh responses API websockets, GitHub co copilot cloud agent. So now if I look in my digest folder, I should have that. Here it is with by the date we get our top items. So open AI releases symphony. So now I could take any of these and and make content with it, right? And I could even ask for ideas. I don't think we did that yet. There is an ideas MD, but I didn't really add that workflow yet. I do have it in in these prompts, but I don't want this video to be too long, so I'm not going to go through all of these. But let's I I do want to show you how to set up a a scheduled task or a cron job. So this right here, I'm going to grab this prompt and I'm going to say set up a daily scheduled task that runs every morning at 8 a.m. It should check the sources and feeds, apply the relevance topics, um remove duplicates, summarize the top stories, explain why they matter, and so on. So this is going to go ahead and set up a cron job. that will run every morning at 8. And obviously, you can set it to run whenever. You can set something to run every 10 seconds if you want. Now, I do want to say something that's important. Don't cram a bunch of tasks into one of these crons. Um, because it'll time out and you'll have issues. And you can extend the timeout, but I would really suggest not cramming a lot into uh into one task or one cron job. For instance, with these prompts, I have another one that says, "Set up a second daily schedule task to get ideas from those digests." I'm not going to run it, but it is here. So, I would split that up. I wouldn't I wouldn't create one task or one cron job to fetch them the data and then also create the ideas. I'd split it up into two. That's why this one says set the second one up at 8:10 a.m. This one's at 8:00 a.m. All right. So it looks like it's set up a scheduled job runs every day at 8. It has a job ID. The output will be in the digest folder. And if you go to your dashboard and you go to cron jobs, you should see it here, right here, daily tech research digest. So it gives you the prompt. It shows you when it's going to run next, which is in 14 hours. So really cool stuff and really easy. You just talk to it like it's a person that you want to do stuff and it can it can edit any files. It can run any command on my machine, but just be careful of what you have it do and what you give it access to. Now, you can also create skills. And skills are basically just packaged tasks. So, if I want, I could say create a skill that will um that will create a digest and generate five video ideas with titles in the ideas file. So you can literally make anything a skill and then you can just have your your OpenClaw run that skill. Now you can also while that's going I'll just talk about this. You can also install let's go to clawhub clawub.ai. You can also install install thirdparty skills. Now you have to be very careful with ski. Don't just install whatever just because it looks cool. If you're going to use any of these, I would suggest going with the top ones like the trending. Like this one here has a 413, what is that? 43,000 downloads, 3 3.4,000 stars. So like these are generally going to be safe. Um, you don't want to just go and and look for some obscure skill with no downloads, no because they can do some serious damage to you. So always, or I should say never trust a skill. go in and read the skillmd, right? Because most of them just have a skillmd file uh down here. Let's see. It'll always show you the structure. So, this has a bunch of markdown files and sometimes like like if we look at gogg for Google services and you need the Google CLI for this, right? So, it's not just a markdown file, but this will basically integrate your Gmail, your your Google Docs, your Google Drive. It'll make it so that your open claw can can work with that. Can send emails on your behalf. And again, you have to be very careful with that stuff. I can't stress that enough how important security is with with OpenClaw. Um, let's see. I'm going to go back here. It looks like it's still working on creating that skill. I just want to show you that what it creates. Okay, so it says done. Created it. Now if I look in the folder here, there's now a skills folder with a tech digest ideas and a skill MD. Got some front matter, a name and description. Gets a purpose. So create a concise daily research digest. Uh distill the strongest opportunities into five developer focused video ideas. All right. So you can tell it to do whatever you want. And I can just say run the you know tech digest skill and it will it'll know what I mean. All right. Now to talk a little bit more about security. So basically like I said there's going to be security with skills. But I think one of the biggest security things you have to worry about is access. You know don't don't give openclaw access to anything it doesn't need. Give it the minimum necessary access for what you want to do. If you're using OpenClaw on a VPS, you want to make sure that you lock it down. You want to make sure that your web UIs aren't accessible to anybody else. You can use SSH tunneling. There's different ways to do that. And a lot of and and like a managed VPS account is going to is going to do that for you. Anyway, be careful with secrets. So like API keys, obviously credentials and and obviously things like social security numbers. Don't don't ever put those in your open claw. Don't don't put them in the chat. Uh let's see. Be careful with prompt injection and use trusted devices. So, I mean, you can there's there's just so much that I can't really go over it all here because this video is getting pretty long, but I just want to show you kind of my my my little workflow that I've been working with over the past couple weeks. So, I have my OpenClaw Travis, right? And I use this through Discord. I guess I could show you actually I don't really want to open the Discord server. I know it's there's nothing there that is like bad to show, but I don't even want to show like the name of my private server because you can do access my whole life from it. So, basically, I can ask Travis anything because his workspace is actually my Obsidian vault, which you're looking at here. This diagram is in my Obsidian vault. It's just it's using the Excal plugin for these for the graphics or whatever for the diagram. But all these files here, business calendar, courses, um guides, my home network snippets, just uh YouTube, basically everything, personal. I have my gym stuff in here, my finances. My openclaw has access to all this. And I even have uh I have a core folder here which is for all the agents I use. So basically if I use claude code or claude co-work with those are the other two agents I use. They look at this first which gives them the context they need. Current state will tell them exactly what I'm working on. They can see each other's sessions. So these are the open claw sessions. These are the clawed sessions. And they always have context. And that's really important is to is to have your agents know the context of of what it is you're working on. Now, I have everything documented. You can see my network. I have all kinds of stuff here. It knows every every device that's on every port of of my switch in the basement. It knows even my electric breakers, my uh my um breaker numbers. So, if I were to ask Travis, I'm going to say, "What breaker is is my microwave on?" And I could just I could do this from my phone as well, obviously. So, it says your microwave is on breaker 41. It gives me the source, which is household, systems, electric. So, that's over here, household. So, it's all in there. I'm not going to go through a lot of these folders because again there's private stuff here. Um, but I can say like cuz Travis has access to my entire network. He can SSH into any machine. I have like seven or eight machines. So I'll say like when was the last time that Trav-dev, which is the machine behind me, the Abuntu machine was rebooted. So, it's going to check the machine directly rather than relying on memory. So, it was last rebooted April 25th, 4:08 p.m. And then I can ask about projects. So, I'll say like what was the last commit on Vidpipe, which is one of my side projects. And notice there's no tool calling. That's been shut off here. I just I forget exactly how to shut it off. So, the last commit was April 26. So, refine article quality. And I could have it make a commit. I could tell it to to update a feature or or fix a bug. Like right now, I'm just showing you how I can get information from it, but I can tell it to do anything that I can do. Earlier, I went to the gym and I logged my gym. So, I'll say what what did I do at the gym today? Now, what's nice about using Obsidian for your basically your brain, I I call it this my my second brain or my shared brain between all my agents. Um, it syncs with any machine you have it on. I have this Obsidian. I have Obsidian on this one, the machine behind me, that the Windows one over there. They all sync together through Obsidian Sync, which is four bucks a month. and every agent on every machine has access to it. So, today was legs. You logged plate leg press, uh, leg extension, um, from the f first log session, leg session, and you want to drop the plate, you want to drop the leg press weight slightly next time for better reps because I had I had struggling with 450. Um, but yeah, so I've been using it to literally manage my life. And another cool thing is I've been having them talk to each other. So like Claude Co-work, I can say give Travis a test task. Now I usually do this from my dev machine. It should work from my Mac though because it's all in my Obsidian brain. And then I should see that in my mission control, right? So if I go to mission control here, we should see a task show up. So let's see if it works. All right. So cloud cloud sandbox mode no LAN route to Trav AI falling back to a direct write. So basically I created an API route for task. It's saying it can't reach it for some reason. So it has a fallback of writing directly to Obsidian which means it'll still do it, right? Or it should still do it. Yep, there it is. Test task from co-work. So that's for Travis. As you can see now, Travis picks it up. And if we look in Telegram right here, starting test from co-work. So to me, that's freaking awesome. Like I couldn't imagine that kind of doing that stuff two years ago or even a year ago. And I'm having these these agents, Travis, Claude Code, and Claude Co-work. I treat them as two different things. His co-orker is more of my planner, like my manager, and Claude Code is obviously my software developer, and Travis just does it all. And and he's also my network admin, right? He has access to all the machines and and does health checks. Um, I would show you my network tab, but it shows my my uh local IP addresses and stuff. I don't really want to put that on camera, but um I I don't know. I think this stuff is really cool and I didn't want to make this video too long. So, I'm not going to show you like how I set all this up right now, but I will make a video going deeper into it, deeper into my workflow. And I'm still learning. Obviously, this stuff is is pretty new. Um, and I'm not really used to it. I'm having all these agents do things in the background while I can focus on, you know, one thing at a at a time. So, I just think it's really cool. And you should see right here. So, the test starting the test and then once it's done, Travis notifies me here. And then in Discord, I have a test um channel. So, every test that's run, no matter who it's by, it shows in that channel. All right, guys. So, hopefully you you got something from this. I just my the base idea of this video was just to show you what OpenClaw is and show you how to install it and get set up, but I did get a little carried away here cuz this stuff just I don't know. I feel like it it's like reignited a passion cuz I think it's it's just really cool and and there's so much that I have left to learn and uh you know I've been doing this stuff every day. So uh that's it. Hope you enjoyed it and I will see you next time.