Hermes Agent Bot Mode Guide (Setup and Workflows)

0:00All right, folks, we have a new feature for Hermes Agent. This is bot mode. This is Hermes Agent's response to the Grok Bot released 2 days ago where we said that Grok Bot has a bigger advantage over Hermes Agent in terms of computer and browser use cuz each Grok Bot has its own cloud computer. Now, with this new bot mode, does this mean Hermes Agent is on par with Grok Bot? Well, not really, but the bot mode is a very big improvement in terms of how you communicate with your Hermes Agent. So, in this video, we're going to show you guys how to quickly set this up. We're

0:36going to play around with it a bit and let you guys know what we think about bot mode. [music] All right, so how are we going to use Hermes bot mode is on the desktop app. But, in order to do this, we need to install the Hermes bot mode plugin from their GitHub. So, just take the link and instruct your agent to install this. All right, so that took about 5 minutes and as you can see, there is a change in the UI on the top left, you can see the bots panel and on the right side, you can see the cron jobs list as well for your

1:11bots. And what's interesting is it already spawns a chat session with your new bot here. This is the default provided by Hermes Agent and you can go to the bot section here and create new agent bots. So, one of the simplest workflow that you can do is a research and report writing uh workflow. So, we're going to create two bots. One will be for research, so research bot. So, we gave it a description to research anything on the web based on my query and then send the research report to the other bot which we named uh report writer. So, we're going to create this agent first. Uh

1:46that's the simplest one that you can do and as you can see, it spawns a new session here for that agent. All right, so even with that very simple description, uh Hermes Agent is smart enough to know what the requirements are and gives a formal description of what its roles, what its responsibilities, so it can browse the web, run code, delegate tasks. So, this is a very nice

2:08setup, especially for everyday users. So, this is what it looks like on a new setup. As you can see, my desktop app looks very blank here because we don't use this to work. But, if you have been working with your Hermes Agent on the desktop app, what you can do is you can, uh, instead of creating a new bot from scratch, you can actually go to advanced and clone from the profile, the existing profiles that you have. So, that's one of the amazing things that Hermes Agent has is you can separate your workspace and your agents uh, by giving them profiles. So, if you have any existing

2:42profiles here, you can click that, clone, and create a new agent. So, we're going to go ahead and finish the second bot, the writer, and there we have it. We got report writer, the second bot. So, now we're going to talk about the biggest improvement that I find with the bot mode is before, when you had to communicate with um, different profiles that you've set up for your Hermes Agent, you had to individually go into the session and talk to them. Uh, but with the bot mode, what you can do is you have a whole interface here for you to just chat, just click and chat. You

3:14don't have to do extra CLIs uh, or commands to enter the session with that different profile. And what's even more interesting is you can actually talk uh, sorry, get your bot to talk to the other bots, which is very similar to the Grok Bot feature. So, we're going to do this right now. We're going to ask our research bot to find out what's happening in the commodities and equities market, and then report your findings to the report writer bot. Uh, let me actually add report writer bot there, so he can summarize and finalize a formal research report. So, we're going to see this in motion. What what

3:50want to see is one of the command lines here where it's going to initiate some sort of agent in box to the report writer profile. So, there we go. We have the task. Four out of five of them will be on its own doing the research. And then finally, when we get to task five, send findings to report writer bot, we want to see that it's running the commands to talk to the report bot. And we should see um a chat pop up here that is not initiated by us, but in fact by the research bot. So, we're going to wait for that to um finish and then

4:23we'll see how that goes. All right. So, there we go. Actually, even on task one earlier, um in one of the commands here, it found the profile of the report writer uh with the cat command. And then soon after, yep, uh found the handoff mechanism where agents message each other via Hermes P uh bot chat. So, this is really good, right? And you can see there's a lot of value in this. In fact, the bigger value here is instead of having you to initiate the task for your agent, what you can do is create a cron job. Tell the research report to research on any aspect or any field, whatever, every

5:06hour or so and then hand that off to the report writer to write uh every hour, right? So, this is a very um straightforward uh workflow that you can do. There's tons of ways that you can go about it. It doesn't have to be research and report writing. This can span to code writing or drafts of code writing. Uh you can even include uh if you have multiple harnesses or agents in your box, all right? You can create that handoff uh dot MD to hand it off to Claude code, right? Or Kimiko 3 or GLM to do the coding job. So, there's so much value here. You can see it's very

5:43convenient. It's on It's GUI. You don't have to go to command lines. you don't have to type so many things, and you don't have to go to different sessions every time or have team ups on to talk to those different profiles. You just have it all here in one desktop app. Now, I previously have said that desktop app comparing desktop app with terminal in terms of how you communicate with your agent coding based tasks are much stronger in terminal because they filter the token streams much more efficiently compared to desktop app. So, still our current philosophy is to still stick to easy tasks such as web search, report

6:23writing, sentiment analysis, but anything to do specifically with coding itself straightforward building apps or websites, you'd have better chances to do that in terminal compared to desktop app. But, we're seeing much more value now with the desktop app because of this single feature of bot mode. So, it's very very helpful. You guys can check that out. Tech name did say that this is still on beta. So, if you do find bugs, make sure that you submit that findings to the news research team so that they will address the feedback and then update. So, I'm definitely very excited to see because this is already working pretty well so far during public

7:02beta and I cannot wait to see if there are more improvements from this. And this is one of the things that I also really like about Hermes agent is they're very quick to respond to what's going on in the world. They know that Grokbot has a very significant advantage in terms of computer and browser use, right? Because Hermes agent exposes browser automation as one of its tool sets alongside file operations, terminal, memory, and delegation, but it runs those tools on whatever back end you configured, whether if you're on local or if you're on Docker, if you're on VPS, or a modal sandbox. There isn't a baked-in persistent DAS style deck

7:41that you get for free, which is why Grok Bot is very expensive. It's like how much again? 300 to use Grok Bot. So, you either run Hermes where you already have compute such as your local device or your VPS, or you opt into a sandbox back end using their tool gateway. So, you can see that there are a lot of steps here. Grok Bot instantly does all of that for you because every Grok Bot has access to the same signed-in browser running on the persistent VM of XAI, which is why their computer task, their browser use tasks are really strong and

8:19very seamless. But still, you're locked into the Grok ecosystem. Whereas Hermes agent, you have bring your own key. There's so much flexibility. It's just all about setting it up. That's the hard part. But when you get through that, it's very, very easy and very, very fun. All right. So, it did all of the research findings, and

8:37now it's messaging the report writer. Uh, there we go. So, message from research bot. Research task complete. Um, we want to see the full thing. Okay. So, so all of these are the findings. This is what we need. So, now we're going to wait for the report writer to write up the report, uh, and then we're

8:54going to see how that goes. All right. So, it's been about 10 minutes and it's stuck. I'm asking my report writer, is it done? Uh, it says it don't have any report in progress. I think this is because I was switching model midway, maybe. So, what we can do now is go back to the research uh, and just say tell him to write it up

9:12ASAP. All right. There we go. Oh, I've [laughter] I've reached my usage limit. I think that is why it was stuck. Uh, okay. But all in all, you get the gist. You can see that there's a lot of value here for using bot mode in the Hermes desktop app because you have one interface to communicate with all of your Hermes agent rather than you having to enter a separate session to communicate with them one by one. And it's even more convenient that you can get them to talk to each other or even set up a cron job to automate that process for them to talk to each other. But in terms of

9:47computer and browser use, your Hermes agent bot can only be as good as the back-end that you've given it, right? So, it's a lot of setup there, a lot of maintenance. But if you don't want that barrier, then yeah, you'd have to pay 300 to use Grok bot for that purpose. So, I hope you find this video helpful. Give us a like. Let us know what you think about bot mode. This is certainly very fun feature. I think this is one of the most meaningful updates I've seen also so far from Hermes agent, right? I remember back in the days, was it 6 7 months ago when we were using MiniMax,

10:22right? We had to do a whole different type of architecture just for our agents to talk to one another. It wasn't even Hermes agent at that time. It was Open Claw. But now, right? There's a whole one interface for you to get them to talk to each other and you can automate that process as well. So, yeah. Subscribe to the channel to follow for more updates, guides, and news. My name

10:43is Ron. Signing out.