Hermes Desktop Runs My Whole Workflow 24/7

0:00Hermes desktop just came out and now I know why everyone is using Hermes. It's because it's so easy to install within 15 minutes. And then this video, I'm going to be putting it through its test, automating a few workflows, showing you how easy it is to get your first AI agent running. Let's get right into it. So, this was just released just a few days ago. News research says the next evolution of Hermes agent is here. This was also demoed on the Nvidia keynote the other day by Jensen Huang. And now it's going mega viral. So, if you're unfamiliar with Hermes and how you used to basically use it was you'd have to actually use it in the terminal. There was other options like using Hermes workspace out there, which is how I had used it previously to automate things for my app called the Magic Hand and

0:46automating content creation for the Magic Hand. But, it was a little bit clunky, a little bit buggy. Now they just straight up came out News Research with Hermes desktop. It is so similar to like a ChatGPT interface or your Codex desktop app as far as running multiple sessions. I like how it looks. It took just 15 minutes to get it installed. After you download the installer from the website, you know, you pick your operating system, macOS, Windows, or whatever. And then you get it installed, and then we can start playing around with it. You can run multiple sessions to get it going. I use OpenAI's my ChatGPT 5.5 subscription because you get so much usage out of that plan to be able to run these agentic workflows 24/7. So, I'm going to plug in a lot of my keys and get it going, and then we

1:32could check out the interface. All right, what I like here is you can just use your News portal, which is one subscription with over 300 frontier models, or you can check your Claude account with Anthropic, which is amazing, too, but you will find you probably will run into usage limits if you're using Opus 4.8 for everything. Like I said, the OpenAI Codex plan is going to be the most favorable because you get so much usage. It's unreal. For $100 a month is what I pay. I mean, I have barely ever hit 90% um with usage limits. It's crazy, and I use it every single day to process information on X, to create content, to do cold outreach. So, like I said, I'm going to use the OpenAI Codex plan, and it's just going to OAuth me with my

2:17login, and I highly recommend just getting the $20 a month plan and playing around with it. So, once you connect you just want to make sure your default model is GPT-5.5. You don't want to be using anything less, like Codex Spark, GPT-5.4, Mini. Don't use those. Just use the Frontier latest model. You just hit start chatting, and here you can already see it's a clean, beautiful interface. I mean, obviously, you want it simple and easy to use for first-time beginners, and you don't want to overcomplicate it. If you've ever gotten Open Claw up and running through the Control UE, it's messy. You know, there's a lot of buttons. Where do I click? Where's my skills? Where's my cron jobs? It's difficult. So, the first thing I want to

3:03do is I just want to see real quick before we keep going. If you're watching this and you want to actually build with some of these tools, not just watch videos about them, then you're going to want to check out our community down below, Shipping School. We [music] have a full Claude code course, a full Open Claw course, and four live bootcamps every single week where we actually help you get set up from scratch. Like, actually set this thing up, not just watch a tutorial and figure it out by yourself. And we also provide one-on-one coaching, so you could book a call with me, we could share screens, and I can help you get Claude code or Open Claw running on your machine. That's it, no fluff. I built this community because watching YouTube's only gets you so far. We launched it just 3 days ago, and we have over 55 members. You need people

3:50around you who are actually building, people who hold you accountable, and coaches who can help you when you get stuck. I'll put the link in the description down below. Get in now before the price goes up. Kind of how it all is laid out. You can sort by sessions, usage. You got your system logs there. That's pretty cool. Awesome. Let's check the settings panel. Change the appearance to Ooh, nice. Pretty cool. Pretty cool. You can set up your voice as well. Text to speech provider. Speech to text. That is awesome. So, you can use the built-in Open AI Whisper voice or 11 Labs as well. That's pretty cool. You can see it right here. That's awesome. You can also change your models here as far as like your vision model, skills hub, approval gate, MCP, memory

4:36and context. So, persistent memory on it uses Hancho memory. So, the memory is already built-in. You don't need to go and find another memory provider like Super memory or even connecting a back-end PostgreSQL database like I have for my Open Clawt. We use Neon. It's very annoying, but this is already built-in, ready to go. That's awesome. Skills and tools. Let's see what it comes with. Apparently comes with over 100 skills pre-loaded and installed. That's awesome. So, you can see you have skills. You have about 90 pre-installed. You have the Kanban board, Comfy UI, Excalidraw. This is great. This is awesome. So, let's start a new session and let's just give it something simple like reviewing the code base for The Magic Hand, which is my AI photo editing app. And let's have it kick off maybe

5:22some other workflows at the same time. And let's just see how fast it is. I would like you to review my repository for The Magic Hand. Make sure everything is optimized and good to go. I'm also going to use GPT 5.5 high and fast. This does consume more credits, but again, you get a ton of usage out of your Open AI subscription. So, we're going to fire that off. I'm going to start another new session. I would like to to check out my YouTube channel Build in Public. See how we can connect via the API, so you can gather more data and see how we can optimize my YouTube channel better. Send that off. Let's do another new session. I would like for you I'd like for you to check out my X account Bo Johnson 89 and see how we can connect to it so we can

6:09gather more data on how to optimize my posts. And I just want to give you some context here just for a moment on why you should be spending your time using Hermes more than any other coding agent out there, especially if you're just looking to automate your first task. Hermes agent on Open Router now has 660 billion token spend versus Open Claw at 249. Why does this matter? Is because people are now starting to look for simplicity in the setup. Simplicity and getting things to work. Open Claw is a little more difficult for a lot of non-technical people. Again, when every update comes out, it seems like something ends up breaking. But with Hermes, it's quite the opposite. They're making improvements, they're listening to the community, and they're not trying to overcomplicate things. So, you can

6:55see here, obviously, Open Claw, you know, used to be the front runner here months ago, but Hermes agent has taken the lead. Over 175,000 stars on GitHub and growing rapidly. So, adoption is taking into effect and people are using it to actually build real things out. Another cool thing is if you just go and type in Hermes use cases, there is over 300 documented use cases that people are using Hermes with. And these can be simple things like an AI employee for my hardest task right here, and you can click into it and get the source link. You can also see that every weekday at 9:00 a.m. summarize my inbox and post to Slack. Something simple like that can save hours a week. My Hermes lives on a VPS, talks over Tailscale. Hermes as my

7:41chief of staff with sub agents per project. We have told it to Google me and ship a landing page to my VPS. We have a meal planner for people who hate logging ingredients. Like there's a ton of build-outs here that you really need to be paying attention to if you're just getting into the AI space. Focus on what people are building and then apply that same type of workflow to your own product. Like I said, we are automating content creation for my app Magic Hand and other than the hour I spent setting it all up with all the back-end API keys through Kai and Seed Dance and Post Bridge, it runs on autopilot now and unless something breaks, I don't need to go into it to ever change it. And it's getting a, you know, these are still getting warmed up but 200 views, 300

8:26views on each platform, TikTok, X, Instagram and YouTube. I mean, that's basically just free marketing for my app that I built out in an hour. All right, jumping back into Hermes, here's the one for just monitoring my X account. I checked Bo Johnson 89 and it said, "I'm building for the future. Follow the vibe court coding journey." Just gives me a quick run down. It said and installed the official X API CLI via Homebrew. So, I need to basically just give it my bearer token which is found in my X developer account. Set it as the default and once that shows the app account is connected and then it can pull all the data, my lists, my bookmarks and all my recent posts and engagement and it will give me a good feedback loop on to iterate with. And for the YouTube

9:12analysis, it's still checking it out and still working in the background which is totally fine. And then for my repo, it is still working but let's see. It is locating the Magic Hand repository. It found it. It's inspecting the structure, dependencies and code quality hotspots. So, this is going to give us a full analysis as well. Very cool, very cool. The X account means the bearer token for me. So, I'm going to need to plug that in to see if it can give us a good analysis. Okay, cool. So, it came back with the YouTube channel information. Basically, I just go into the Google Cloud console, give it my API key, and it can start giving me constructive feedback from click-through rates, when people drop off, as far as like and view to subscriber ratios, to be able to help me dial in my YouTube strategy. So, if

9:58you make YouTube videos, this is something that you're going to want to set up, and it's free. So, basically, like it can only give you so much data without the API access. But yes, the YouTube Analytics API is something you would want to set up in your Google Cloud console to be able to give you some good feedback. Basically, it would pull, you know, video-level performance table, the velocity title pattern analysis, title and thumbnail experiment tracker, retention drop-off flags. This is going to be a very high-quality agent to be able to run in the background. And the other point I want to make up is if you want a recurring workflow, they now have built-in cron jobs that you can just set up and create your first cron job. Where before, in the Hermes workspace that I mentioned earlier, you could do this, but it was incredibly

10:43buggy, and the cron jobs wouldn't always fire off consistently. So, basically, if I want to take this workflow and add it to a scheduled cron job that just maybe kicks off once a week, like a performance evaluation that I have already set up with my Open Claw, but this is something that you could set up in Hermes in just a few minutes. You could have it run every day, daily, weekly, delivered to the desktop. You could deliver on Telegram, Discord, Slack, or email. This is amazing. Even without the API key, really gave me some useful information. Like, it has relevant stats. We have 11,400 subscribers, 567 videos, ship your app in weeks, not months, which is our positioning. It has the quick pay public data read, Open Claw updates, specific tool repo narratives, top public long-form examples were Open Claw

11:30updates when it was popping off the chain. We have current recent content, Hermes agent, Anthropic, Open Claw. Recent examples included awesome and then short patterns, popular shorts that we that it saw. But yeah, getting the API key is basically going to give it all the channel metadata, full video inventory, and much more valuable data to make dis- decisions and help us, you know, basically optimize our YouTube channel better. So, I just went into my Google Cloud Console and just downloaded the JSON file. I just followed the instructions and now I'm going to be able to get daily, weekly updates on the analytics of my YouTube channel and get further insights. So, I'm just going to attach the JSON and allow it to do this. Basically, this analysis that's going to

12:15be more in-depth. Please set up a recurring cron job for me where we can pull this analytic information once a week. Here's all the information you need. And yes, I'd love you to set up those recurring pool that you suggested with all the performance metrics and suggestions. So, that's going to work in the background. So, it did a great public analysis for my X account. That again, you would go through the same process of basically giving it the client secret, your bearer token to be able to actually scrape X, which will be hugely valuable to anyone that does anything with trending information. You need to get on that. It is usage-based now. You don't need to pay $200 a month for $20 a month you can stay top of the

13:00line in your industry by just connecting your X developer account. It's free to sign up. So, this gave me a huge, detailed information. Wow, it's still actually working on the public analysis. So, deliver concrete post profile recommendations. So, if you don't want to go through that, just have it do a public analysis and it's going to scrape all the data. Wow, this is insane. You can also have it take all this data and put it into an artifact, which is really helpful, really nice. I love that. I think it's good to pay attention as it's running these along processes to pay attention to your session usage down here, how long the session's been going, what model you're using. This is important because I don't want you to be spending more than you should, and you need to

13:46keep track of like, you know, your token consumption. Not everyone has the $200 max plan for Claude, not everyone has the $200 OpenAI Pro plan. So, just something to be aware about. You can track all of your token spend right here. So, basically this is the review summary from the Magic Hand, and to be honest, guys, I am completely blown away on how fast, like I've been filming this video for literally 30 minutes, and there's going to be some edits to kind of cut out some of the pauses, but in within 30 minutes, we literally have three agents running in parallel doing all of this work for us. That is insane. Like, I've been using Open Claw for since end of February, and it's still very slow compared to using Hermes, and I use Open Claw with GPT-5.5 as well, and it's very clunky, very slow. Maybe because I'm on an older um version, but

14:33this has so much power behind it. If you could just learn how to use this tool to basically help you with a lot of the redundant things that you do every day. So, basically this just did a full analysis on my whole repository, the Magic Hand. It applied or it suggested low-risk readiness fixes that it were found during the review. I could pull that into code X or Claude code to apply those fixes. It reviewed and tightened up the Magic Hand repo. Good to go with fixes applied and verified. I did not run the full page generation publish pipeline, but I did verify code, dependency security, local render. Oh my, I'm amazing. So, it found five vulnerabilities down to zero. The working tree has now uncommitted improvements, so I can merge that after

15:18I reviewed the diff. One thing to be aware of, the existing post-racial warnings has draft saved to perfect post-racial only. Wow, the recommended next step would be running this command. That is awesome. So, you can see that is done just out of the box. Like just getting this set up, this is just the way of the future right now. No more terminals, no more working in clunky workflows, patching things together. This is directly from News Research to be able to just give you your first agent up and running. Like I mean, it's pretty incredible on what it just did within 30 minutes, giving me analysis on my YouTube channel, my Magic Repository, and my X account. Again, you could take it a deep a step further by automating this as a scheduled cron job, turning

16:04that agent into an actual workflow that runs on autopilot. So, again, you know, you have those pre-installed skills. You can turn things into artifacts. Also going into the settings pane, I mean, you can connect your vision model, which you can change here. You can use GPT 5.5 image, which is amazing. Also, web extract. I wonder if you can use Yeah, you can use GPT 5.5. That's awesome. Also to keep in mind, you can use Anthropic, but you have a chance of being banned, which is something that you probably don't want. Um so, I would probably stay away from that and just use either open weight models or OpenAI. I think OpenAI for what you get on just the $100 a month plan, you get so much

16:50usage. It's pretty incredible. But, I'm pretty impressed by this just with playing around with it for just 30 minutes. You know, there's no reason why you shouldn't be using this in your daily workflow. And don't be biased towards one or the other. I think you should be using all of them. I think you should be using Codex for, you know, code reveal or using Cloud Code for the initial build out, whether using Ultra Code or whatever. And, you know, don't just to a bias towards one or the other. I think all these models used in conjunction with one another is a great workflow. You know, you can have your Claude code agent, you know, set the initial foundation for your code base, and then have Codex come and review it. You know, Codex can fix bugs that Claude code Opus 4.8 can't, and vice versa. So, you're going to need to have multiple

17:35tools in your tool belt to be able to squash bugs, make fixes, ship code faster, and I think Hermes is just another piece to the puzzle to get going. Really impressed. Um you know, like I said, I was using the old workspace. That was just very clunky. It would bug out at times. It was just very annoying. And this is in pro preview right now. You need to get on it. Like I said, we got YouTube information, Magic can repo repo info, and then our X account, which it went into great detail of just like the just this was just from a public scrape, and just gave me some great information that I can apply, um you know, to my future posts and future videos for YouTube. So, no reason why you shouldn't be using it. I'll put a link to it down below. You need to check it out. And if you want to learn to

18:20build actual AI agents just like this, check out our community. We have over 150 builders in there, and we do four live boot camp calls a week, where we go over breakdown tutorials just like Hermes agent, Open Claude, Claude code, Codex, what is AI, AI inference, and AI engineering. Check that out cuz we're adding courses every single week. So, what is your feedback? Have you used Hermes desktop? What is your thoughts? Are you using Open Claude still? Are you using Hermes? I want to know. Give me some feedback. And if you guys haven't subscribed already, please do cuz we release videos just like this every single day. We'll see you in the next one. Have a blessed day.