Hermes Agent Kanban Swarms: Build a Free 24/7 AI SEO Team (Orchestrator + Agent OS Setup)The video demonstrates Hermes Agent Kanban’s new automation upgrade that lets an orchestrator agent take a single prompt in the Kanban triage column, break it into subtasks, assign them to specialized agent profiles, and run work in parallel to automate SEO and other business workflows. The presenter updates Hermes via the terminal, opens the Hermes dashboard Kanban, creates tasks like writing a blog on AI SEO automation and building a local HTML site, then shows how an Agent OS syncs Claude, Hermes, and Kanban into one mission-control interface with memory context. They explain assigning tasks to a writer agent (“Julian”), viewing run history and handoff summaries, and how tasks move from triage to ready, running, and done. Examples include generating a 1,350-word blog and producing a keyword research plan split into multiple subtasks, plus growth results on websites reaching up to 1,134 clicks/day. They recap the Agent OS concepts (mission control, profiles, Kanban orchestration, memory, and self-healing), mention CLI Kanban commands, and direct viewers to the AI Profit Boardroom and a free SEO strategy session at goldie.agency.00:00 AI SEO Swarm Intro00:58 Kanban Upgrade Overview01:50 Update Hermes and Launch02:46 Create First Kanban Task03:23 Agent OS Mission Control06:13 Debugging and Workflows08:19 Assign Agents and Run09:29 Review Outputs and Memory12:05 What Is an Agent OS14:10 SEO Results and Deployment15:36 CLI Commands and How It Works16:29 Push Split Route Deliver19:10 Profiles Parallel Execution21:53 Beliefs Setup and Reliability23:49 Recap Community and Q&A28:00 Final Wrap Up
Full transcript
Hermes Agent Kanban Swarms just changed AI SEO forever and in this video I'm going to show you how to run an entire team of AI agents all working at the same time, all doing your SEO for you completely free. One agent plans your content, one agent builds it, one agent publishes it, all at the same time, all on autopilot. That means you can create more content, you can rank for more keywords, you can get more traffic to your website way faster than doing it yourself. And here's the thing, you don't have to touch any of it.
You give it one task and the agents figure out the rest for you. I'm also going to show you exactly how I set this up, how I used it to grow one of my websites over a thousand clicks a day, and the one thing you absolutely have to do inside your Kanban setup to make this work properly, which I'll show you later in this video. If you stick with me until the end, you're going to walk away with your own AI SEO team running 24-7 and you can actually apply this to anything inside your business that you want to automate. Let's get into it.
We have a brand new update from Hermes Agent Kanban, which makes it super powerful for AI SEO. I'm going to be walking through how you can basically automate anything, including AI SEO, using the new Hermes Agent Kanban. So you can see this announcement from the co-founder of Hermes Agent, he says that Hermes Agent Kanban just got a big automation upgrade. So you can basically drop one prompt into the triage and the orchestrator agent can take it from there.
And basically what it does is it takes all the subtasks necessary and automatically assigns agent profiles of it that specializes. What that essentially means in English is that with the dashboard inside Hermes, you can basically drop in a task and it will create loads of subtasks using the orchestration agent to build and automate whatever you want. So we're going to be testing out today and seeing how it performs. The way that we can use this is if we go over to terminal here and then from here we're going to make sure that we have updated to the latest version.
So we can fetch the latest update from Hermes as you can see and then it can pull in the latest skills. So it can pull this in as you can see here. This just means that we update to all the latest skills inside Hermes right. So to do that you do Hermes update inside your terminal and that will update and download the latest updates for using Hermes Agent.
Now in terms of how this works whilst that's updating here, basically this is a multi-agent profile collaboration. So multiple agents can work together when we're doing this. So let's try this out now. If we go to Hermes here we can run this inside terminal as you can see and then we can open this up in a new tab and we can say okay Hermes dashboard.
That opens up the dashboard for Hermes Agent and then what we can do from here now is we can scroll down to Kanban and with Kanban we can create a new task inside the triage section and that should start orchestrating it. So let's test this out. We're just going to try this out now. So I'm going to put in a new idea here.
So let's say for example I don't know something super simple. We'll keep it simple to begin with. So I'm going to say write a blog about AI SEO automation with OpenCLR and then we're going to try that. Plug that in, click create and we can see that working right here.
We can also add that as part of another task or setup here as you can see and we can start working from there. Now what I've actually done over here as you can see and this is even better is you can actually build this into your agent OS right and so this is an operating system basically a mission control for managing all of your agents. You can see for example I have like Claude here, I have Hermes Agent and we have a Kanban section and so what we can do is use the new orchestration features with Kanban and start plugging in the tasks and asking our agents to create them right. Pretty powerful stuff.
So for example here you can see the task we just created inside a dashboard shows up inside a Kanban right. So when we go to the Kanban section now this is synced with everything else. Now I want to be 100% honest with you because when I actually look at this setup I can see that the actually this is working pretty nicely so now it is I thought it wasn't working for a sec there but it is working nicely. So you can see here that it has started to work on that task right so it says needs an assignee and it start working on the task here so it's moved it to the ready section right and so basically this will automatically triage the task using the orchestrator agent and then start delegating that out right which is pretty cool and you can see inside our agent OS over here this is fully synced up to our Kanban board so this is just like an operating system for managing every part of our SEO agents.
So for example we've got a section here where we can create SEO content we can see the content we've created over here and we've got the Kanban section too and so we can use this to start delegating and creating tasks using our AI agents. So let's say for example I've got a new task here I'm like okay I want to create a new website built locally in HTML about AI automation and just as a little test let's say for example it can just be a basic HTML website that looks nice right plus sends traffic to the AI profit ordering community. Let's say for example you were trying to create a website ranks for a keyword so we'll add that in now we'll click add here and that's synced into our triage section so now the agents are going to start working on that task. Now what this means is that we can control it inside our Kanban board here we've got this nice operating system we can wait and then switch over to SEO and start doing some SEO in the background we can go back to our Kanban tasks here and you can see how it's beginning to be worked on now.
Now what I want to do inside Claude, Claude is helping me build this agent OSL and you can build an agent operating system with whatever agent you want to use. So I've got Claude over here and the way that I set this agent OSL is I literally gave it the documentation and then it built that in right it said here we go yes I gave it the documentation in terms of what the Kanban automation involves what it does etc how it works etc and then you can see inside the chat here it's built out and I've got two tasks ready to go. There's two things I want to check here so number one I'm going to ask Claude here is the Kanban working e.g. is it being delegated to Hermes agent what's being worked on it seems stuck in ready how do I see the progress so let's check on that.
So just note here the Hermes agent the Kanban board is pretty cool it's a great idea but sometimes can be a little bit buggy right just being 100% honest with you so if you find that problem you can always go into Claude or into Hermes agent and just say listen I've settled this task why is it not working official says hello bro good to see you thanks for joining here and so you can see it's fixing that over there. Now let's talk about what else you can do with this stuff right so you've got two different ways you can talk to your tools or three different ways now because you got the agent OSL. So the agents can drive the board through a dedicated Kanban tool set if you're using and speaking to Hermes agent directly or you can use the board directly which is what we did with the agent OS right and so it can work for terminal with Hermes agent or it can work directly through Kanban too. Now this covers workloads that normal sort of delegate tasks wouldn't so if you're using Hermes agent normally there's certain things you can do with Kanban that you wouldn't do with for example something like the normal sort of way of using Hermes agent so for example you can research triage right you have parallel researchers one analyst one writer with a human in the loop that's an example of an agent team you could also for example have scheduled operations you could have recurring daily briefs that build a journal over weeks right you could have for example persistent named assistants like one for inbox triage one for reviews that accumulates memory over time you could have engineering pipelines you could have multiple agents working in parallel together or you could have for example one specialist managing like multiple different accounts or websites and that sort of thing so you might be saying why would you use the Kanban instead of just Hermes agent it's quite different so you can basically build a team with this process and also you have this human in the loop system with Kanban that you don't have with Hermes agent normally which is pretty cool now we can also just build this in now into the agent operating system so we can also assign this to different agents so you can see all of my agents here if we want to give this to an agent we can just select it from the assigned section here so let's try this out now we're going to assign this to Julian which is my Hermes agent for writing content and now you can see it's running right so the task is actually running which is great so basically what this does as you can see here is it moved from ready over to running because we assign that task to our AI agent which is this one over here right and it's already working now which is pretty cool as well if we want to get the other tasks running so for example this one we can just switch this over and we can assign this to say Julian as well right and so what that would do is it will move from ready to running once it syncs over right which is pretty cool so you can see it now has moved over to running here right now one thing that you can do as well as you can see the status here so you can see the events used etc and you can also see the run history of the tasks as well which is great if you ever need to archive the tasks you can just click on archive over here you can block it you can mark it's completed etc and you can also see what it's done right so it says latest handoff summary produced a complete original 1350 word blog post called AI SEO automation with open core covering all requested stuff file saved to workspace here now if we put a comment on that and we say where is the content it should reply to us in a minute as well and you can see all the tasks have now been moved automatically to done so basically the agents wow that was so fast the agents go through the task create it as you can see we give it the details and then it works and we can see the log, we can see what's done etc.
One thing that would be super useful on this is if you can actually see the outputs, which is what I might do in a second, and you can see how much easier this is to use than if you were to use this inside terminal without OriginOS. If we went into a new terminal here and we put this in, that is if you're just using the terminal to manage your Kanban, whereas with this process you can see it's nicely organized, it's a nice UI, it's easier to organize, we can switch over to this one etc. It's super easy and simple to just switch between everything, right? We can see all the tasks have been completed inside the chat right here, which is pretty awesome.
And then you can see the actual blog has been created by the Hermes agent right there. The other crazy thing about this is I didn't ask it to create content around OpenClaw, but because it has a memory layer built into the AgentOS, it knows, okay, people are interested in OpenClaw, that'd be a great blog to create for people watching, let's create that, right? And it creates this whole blog, as you can see right here, using the Kanban board, and it did it directly because it understands and has context about us, which is pretty amazing. Now what we can also add inside the AgentOS, which is what I'm doing over here, is we are creating a workspace file viewer.
So any sort of files or blogs or websites that I create inside the Kanban board will be viewable and we can just watch those created, which is pretty amazing. Let's see what we've got inside the questions. Best laptops for running local models? I think the one I've seen a lot of people praising recently is like a DGX Spark.
Honestly, local models are not that great. You can also use like a Mac Studio, but again, local models, not that great. I don't like them that much, right? Universes, I can't find any courses for Hermes.
So if you go to the AR Profit Boardroom, I don't know if you remember or not, but if you type in Hermes course here, you can find like a full course on Hermes inside this section. So you can just use the search bar here, type in Hermes course, and then you can get access directly there. So you can see the full course here. If you want like a full step-by-step course, but if you want like new advanced daily updates, then you would go to this section over here.
Alright, that should be set up right here as well for the Kanban board. So let's try that out. We just refresh our AgentOS. Now the other cool thing about this is like you can really build an AI assistant team using this process, right?
So with this operating system, what you've got here is like a powerful way to build an infrastructure that runs everything for you, right? And you just become the CEO of a machine that's running 24-7, right? So you set the direction and the machine implements everything, and that's what this operating system looks like, right? So you might be wondering, okay, what is an agent operating system?
So if you're using, for example, the terminal here, it's like a chat window, right? It's the same way as using like chat cheapity or something like that. And you might have Claude in one window, Hermes in another, etc, right? And it's much harder to organize and orchestrate all your agents when you're doing that, right?
Nothing really links together. You don't have everything saved. You can't really see the chat history. You definitely can't see the Kanban board.
So when you're using this process instead, what you've got here is a nice setup with all of your agents in one place. You've got Claude, you've got OpenCore, you've got Hermes, everything working together nicely. And it's just much easier to manage in one place, right? I can even, for example, go to my studio, I can go to the Kanban board, I can go to the SEO section as well, and everything's saved in my history as well, which is really useful as well.
Let's see what we've got in the comments here. Amit says, very nice, appreciate it, happy to help. And so you might be wondering, okay, what is an agent operating system? Your computer is an operating system, right?
It manages memory, processes, files, and programs. It runs everything in the background without you telling it to. An agent operating system does the same sort of thing, but for your business work, right? So it has the mission control dashboard.
It has all of your agent profiles. It has a Kanban board, which in the task delegation and orchestration engine that connects everything. And then you have a self layer, which is the memory system that makes your agent smarter over time. So for example, when we were creating SEO content, the reason that it knew to create content around OpenCore is because it has this map of me and everything that I work on and everything that I do and all the people in my life and what I do day to day, et cetera.
Nice context, really awesome machine with all my memories. And so it can pull that data, look at what I've worked on recently, and then use that to create AI SEO content that's really tailored to me, right? Which is super useful. And so let me show you an example of that.
If we look at these four websites, Hermes, Claude, et cetera, they can all deploy to Netlify for me. And so when I'm creating SEO content, we can deploy directly to our websites. It's pretty simple and easy to do. And you can see the trajectory of these websites as well, right?
Like this one has gone from one click a day, all the way up to 95 clicks per day with AI SEO, right? It's ranking for some nice keywords. This one too, similar sort of trajectory. It was at zero clicks a day.
Now it's 71 clicks per day, right? And this one too, this was basically 15 clicks a day and it's grown all the way up to a peak of 1,134 clicks per day, right? And so these websites are using this process I'm showing you to create better content, to organize it better, to make it as good as it possibly can be, and also to make it as useful as it can be as well. So when we're using the Kanban board to create content, it's really easy to assign new tasks to our agents.
Our agents can publish our websites. We can give them new websites to create. We can create new content, new ideas of clusters of content to create. It's simple and easy to navigate.
The other cool thing with this AgentOS system is that we can actually use it to generate SEO content directly with this section, right? And so what this section actually does is we just plug in our keyword. We pick the case study that we want to use, and then it goes off and deploys it, which is pretty nice as well. Now, what can you do with the Kanban update as well?
So you can manage boards from the CLI if you want to. You can manage them from the dashboard as well. You can do these quick start commands inside the CLI, as you can see, and you can say, okay, Hermes Kanban create, and then give it the topic and it'll go off and create it. And you can assign it to, for example, your researcher or whatever you want.
If you want updates, you can launch Hermes Kanban watch inside your terminal and watch the Kanban list. And you can also see the boards, right? So you can do Hermes Kanban list or Hermes Kanban stats and get the latest information on it there directly too. So you can see, for example, here, we even have a studio inside the agent operating system that we can create images, videos, text to speech, et cetera.
We can manage our Kanban board. We can go over here, plug in keywords, and see our history of deployed content as well. It's just really easy to organize in one place. Now you might be asking as well, okay, how does this work?
So essentially you open the mission control, you go to the Kanban triage column, as you can see right here, you type one sentence, one idea, one goal, one deliverable, and that is your only job in this entire operating system, right? You are the owner given the direction and the operating system handles the implementation from there, right? So let's say, for example, here, I'm like, okay, I'm going to assign this task to Julian, Hermes agent. We're going to say, okay, create a keyword research plan for my AI automation blog, and we'll plug that in here related to OpenCore and Hermes agent topics.
We add that in, that's going to add it inside the triage section, and then what that's automatically going to do is triage your task and move it over to ready and running once it's ready to go, which is pretty cool. And phase two is really like the split. So the orchestrator agent looks at your goal, reads the sentence he gave it, looks at every agent profile you've built and their descriptions, and thinks, okay, what needs to happen to complete this task? It breaks your idea into a full project plan.
It creates little tasks for every piece of work required. It decides what runs in parallel and what runs in sequence as well. So it assigns every task to the right specialist profile as well. It links the dependencies as well, so nothing starts before it should, and your one sentence becomes a fully structured project, as you can see, right?
So you see how we said create a keyword research plan for my AI automation blog here. You see how it's assigned these tasks to other AI agents, and we've now got a running section, right? And so it split the task of keyword research into three tasks, which is map the core topics around AI automation, OpenCore and Hermes agent, generate keywords and long tail variations, and then define evaluation criteria and data sources. And so these are running in the background to complete this task over here.
So this will still be in the to-do list until it's completed, and then it'll be moved to done once it's completed. So it's a really powerful system for just having a team of AI agents, and if you've got multiple profiles with multiple specialties inside your Kanban board, this is how you can operate them. And so we've talked about step one, which is the push, right? So you give it a task inside the Kanban.
Phase two is a split, so the orchestrator agent decomposes a goal, and then splits out tasks, right? And you can see how these tasks are now marked as done, which is awesome, and there's just one more task that's running. From there, you've got the routes, right? So tasks go to the right specialist, so each task lands on the profile that's built for it.
So your researcher profile gets a research task, your writer profile gets a writing task, your SEO analyst profile gets the SEO tasks, your ops profile gets the operational tasks, right? And so the routing happens based on the profile descriptions you write during the setup, and if you're not sure how to set up profiles for your AI agents, we have a full step-by-step guide on that right here, right? In terms of how to create different profiles for your AI agents. So basically all you do is type in Hermes profile create inside your terminal, and then you name the agent, right?
And then to set up, you run this command, but we've got a full setup guide and video tutorial on that right there. So that is completing the tasks as you can see, and then your agent implements in parallel, right? So you saw before we had three running tasks, two have been completed, this one is running right now whilst I'm talking to you, and so it's all running in the background, pretty simple and easy, right? And so multiple agents work simultaneously, the researcher digs into angle A, the second researcher digs into angle B, right?
The SEO analyst runs an audit, the writer is queued and waiting for whatever test you give it, and so all of this happens at the same time on the same day. board, tracked in real time. So when the research is finished, the outputs get passed to the writer automatically, the writer would pick up the structured handoff and implement. And so the board tracks every heartbeat, every update, every completion, and you watch the whole thing in mission control without touching the task, which is literally what we're doing right now.
And then the final step is the operating system delivers the result, right? So if we click into this task, for example, we can see the completed run history here, and we also have the markdown file of the complete task, right? So you can see that right here, we open it up in a new tab, it's written in markdown, so I'm going to open up in a previewer, you can see that it's come up with the core topics for our keyword research, right? And so that is all set up inside here, which is pretty amazing.
So every agent closes with a structured handoff, what changed, how it was verified, what the agent needs to know next, etc. And you can just see everything that's been completed over here, which is absolutely amazing, right? Same for this section, you can see the document over here, you can save it, copy it, you can plug that into the markdown live preview, if you want to check out, and you've got like all the keywords, the long tail variations for AI automation, open core Hermes, the full structure, all the keywords, and this is split into clusters, right? So for example, you got Hermes agent architecture, and then relevant keywords underneath that, building and maintaining custom skills for Hermes, and then relevant keywords underneath that, right?
And so it's planned out the website step by step in terms of exactly how we're going to do the keyword research. And we can just preview and manage that inside our Kanban board. So it's a really powerful stack. And this isn't about prompting better and file that is about building an operating system that implements 24 seven, even whilst you're not there, right?
And so you want to think of it like it's like the agent OS is like a company, and you are the CEO, you set goals, you review results, you give direction, and the mission control dashboard is your boardroom, right? So you see everything across your agent teams, tasks in triage, tasks in progress, tasks blocked, waiting on your input, etc. So that's basically it, that's how you can set it up. Now, some people say I need to manage every AI task myself to get quality output.
As you've seen, this all worked in the background with a team of AI agents, the keyword research was amazing. And so the right belief really is like the orchestrator agent manages your entire team, you just set the goal, right? So if you're still manually assigning tasks to agents, one by one, you're operating the old way, right? The orchestrator profile exists specifically.
So you never have to do that. You write one goal, and the orchestrator reads your team profiles, builds a task graph and routes everything, right? Your job is just to set the direction, not manage the implementation, right? And you can see the tasks has been completed.
It's inside the ready section, we can see all the tasks ticked off here. These were the subtasks that were built on the side, the Kanban board, and we've got everything ready to go. Pretty amazing. Other people say an agent OS is too complex to set up.
Honestly, I can understand that. What you can do is you can build it once and after that it runs itself, right? So every time a new update comes out like this, I can just stack it into my system, right? I can set up a new section of SEO.
I can, if grok 4.3 comes out with images and video and speech, I can stack them right here, right? If you want just my system, if you want files that I've used to create the agent operating system, we've got that inside the AirProfit board in here, as you can see, and literally you can get all the prompts that I use to create it, plus you can get the zip file with all the documents for actually creating it, right? So you can actually get that zip file and then just give it to your agents and be like, hey, create this, right? Based on these screenshots.
So it's pretty cool. So other people say AI agents are unreliable, they break and have to fix everything manually. The right belief here is the agent operating system has a self-healing fixer and improver, right? So if a task fails repeatedly, it also blocks with the errors and clearly locked.
You see what failed when and why, and that'll be inside the block section here, as you can see, right? And so you can add a comment with instructions, you click on block, and the operating system just picks it up on the next tick. That's basically it. So just to recap on this, you've got a really powerful way to delegate tasks and orchestrate them across all of your agents using the Kanban board.
If you plug in the memory as well, it's going to be 10 times better, simply because it understands you, your business, what you work on, and it's going to have way more context. When you give it tasks, it breaks it down into subtasks, and then it goes off and organizes it on the Kanban board, as you can see, right? And if you want the full system from me, all the prompts, etc, you've got that inside the AR Profit Boarding. I'm actually going to add a new section inside the AR SEO automation part here.
If you want the actual prompt that I used to create this, they'll be inside here too. So that's it. So that's basically my automation community, you can post questions inside there, you can connect with me personally, we do four weekly coaching courses as well, you can jump in the classroom, get all my best trainings, and that's all inside here. Also, if you want a free one-to-one SEO strategy session, that shows you how to get more leads, traffic from Google rank number one inside AI and search engines, then you can book a free SEO strategy session that's available at Goldie.agency, you'll get a free SEO domination plan, so you'll get a custom tailored link building plan to generate more leads and sales from your website, you'll discover the secrets SEO link building, you'll be able to ask any questions on the call one-to-one, and you'll learn the best link building strategies for your website.
Link in the comments description or go to Goldie.agency. So let's answer some of the questions from the community as well. So Jean says, where are the NA10 templates? And they're all right here.
Jose says, Julian is going viral. Let's take this, let's have a look at this. Oh, look at that. Wow.
That's pretty cool. That's pretty cool. Nice. Thanks for sharing Jose.
Let's see what else we got here. Now this is cool. What I like about this is people are sharing all sorts of interesting stuff inside here as well. So you can see this example, where people are sharing the MCPs and the GitHub open source projects are working on.
It's pretty amazing. This is pretty interesting. Surf says, you finally got Hermes working. So I use Manus every day.
Hermes is not as good. This is interesting. I would say the outputs here that I see probably depend on the quality of the API you use, right? So for example, if you're using Manus, and you've got probably Claude inside there creating content, but then you've got like a subpart API inside Hermes, the content is not going to be as good.
But one thing that I'd recommend is that you train Hermes and give it a skill for creating content so that it understands exactly how you want your content created. So for example, if we go back to the Asian OS here, and we'll go to the SEO section, history, and we'll see some examples of content that we've created. This content is written exactly how I want it to be written based on the case studies I give it, because I've trained it up on the skill. So usually if I'm trying to train Hermes agent to create better content, I'll give it the exact prompts and the skills and the context it needs to, and then I'll just keep tweaking and improving over time.
But yeah, it takes a little bit of time to do that. But I think the main thing is here, like whether you use Manus or whether you use Hermes, it's all about finding what's comfortable for you, right? So if you prefer Manus, there's nothing wrong with that. And you can just keep going that we've got some awesome wins as well.
But yeah, I think that's all the questions answered. If you want me to answer your questions, just like we've done today with a video tutorial and a summary, if you don't have time to actually watch the video tutorial, like you can see here, I just break it down into a quick answer as well for you, then you can post your questions inside the community and I'm always happy to help. Let's see what we got in the questions. So Oflow says, Hi, from Paris, how are you?
I'm doing great. Thanks for joining. Bonjour. Trad says, I hope I get Hermes set up and rolling.
Just get it done today, man. It sounds good. We've got a good question here, which is with all the difficulty of setting up open, if you find it's difficult to set up open claw or Hermes, I would just go into something like Max Hermes or Max Claw and you can just get them deployed in like one click. So if you find this stuff difficult to set up, you can always go over to something where it's cloud hosted and they just do all the hard stuff for you.
That might be a good option. And Brian says Kanban. Yeah, absolutely. So we've just covered a full session on how to use Kanban to automate basically anything using these teams.
And once you give it a load of tasks and they're all marked as completed here, right? So you can see all the tasks right here. We can view the history inside this agent OS system and it's super easy and simple to set up. So that's basically it for the training today.
Thanks for watching. I'll see you on the next one. Cheers. Bye-bye.
More episodes