This is the full transcript of Hermes Agent Update v0.22 is MASSIVE! (Threshold Release), published on YouTube by Superbash (BoxminingAI). Every paragraph carries the moment it was spoken, so you can click any line to jump straight to that point in the video, search the whole thing for a word, or copy it out.
0:00All right, folks. Hermes' agent just got a big update and the useful part here is how much easier it should be to get everyday work done. With the v0.22 update, you get to bring your apps into the conversation, makes running local models much simpler, and gives you more control when the agent is working in the background. There are also a couple of settings worth checking before you upgrade, including one that can
0:24automatically delete old conversations. So, here's what you need to know if you're using Hermes' agent everyday. Now, first up, app connectors. For me, this is the biggest value with this update. It's functioning more now like Grok Bot. Hermes can now work with apps like Gmail, Notion, and Linear through hosted connectors. You connect an account in your browser and Hermes gets tools to work with that app. So, for example, you can ask it to draft an email in Gmail, read a page in Notion, or create an issue in Linear. That saves the step of taking an answer from the chat and manually moving it into another app. You can also connect multiple
1:08accounts for the same service like a personal Gmail and a work Gmail. My suggestion is to start with one task you already repeat. Maybe that's drafting a weekly update. Connect the app you need, give Hermes the context, and see whether it actually saves you time. I'd say that's a much more useful first test
1:27than connecting every account you own. Now, the second big change is local models. Hermes can now handle the engine installation, model download, and startup for you. It checks your hardware and recommends a model that fits, which removes a lot of the guesswork around what your computer can realistically run. You'll find this under settings, providers, local models. The practical benefit is having an option that runs the model on your own machine. That can be useful for working with local material or reducing your dependency on a paid model service, but keep the distinction clear here. Running the model locally doesn't make every task offline. If you ask Hermes to search the
2:09web or use Gmail, those actions still need outside services. And your hardware still matters. Easier installation doesn't mean a small laptop suddenly performs like a powerful workstation, but I'd try it on a familiar task yet again, like summarizing notes and compare the speed and quality with your usual model. That should tell you where it belongs in your workflow. Now, if you're helping someone get started, there's also a guided desktop setup chat with a one-time free trial model. They can try Hermes before setting up an API key. Just be aware that this trial is for a first desktop install. It isn't unlimited free access across every version of Hermes. Once you found
2:48something useful for Hermes to do, the next question is where you want that workflow to run. Here's how we're handling that for our own channel. If you want Hermes to handle recurring tasks, you need somewhere to keep them running. This is where Hostinger comes in, who we want to thank for sponsoring this video. Right here, I have one of my workflows where I've set up my Hermes on a Hostinger VPS to basically research ideas 24/7 for my faceless YouTube channel Lord Yapper. It's one of those animated history explainer channels. And there are three reasons I'm running it there. First, it gives the agent a separate environment away from the
3:25personal files on my computer, which helps reduce a lot of those risks to my local machine. Secondly, it costs way cheaper than buying a new Mac Mini just for this workflow. And lastly, it can run 24/7 even when my computer is switched off. And here's the workflow. I've given Hermes the channel's niche, the style we're going for, and the topics we've already covered. It searches for potential stories, checks the sources, and prepares a brief with an outline, some title options, a
3:55description, and the supporting links. I've scheduled that for 9:00 every morning with delivery straight to my Telegram. Here's one it produced. I can review the angle, check the evidence, and decide whether it's something interesting to show on YouTube. There's also a separate prompt ready to copy into Higgs Field, and once I'm happy with the research, I hand that over to production for the final video. So, if you're interested, you can check out my
4:21Hostinger link in the description below. I'm using KVM 2 for this workflow. Select the 24-month period, then enter boxmining10 at checkout and click apply for 10% off your purchase. I've also linked our full Hermes setup guide on Hostinger if you want to follow the installation step-by-step. All right, let's get back to the video. Now, the next improvement is being able to see what Hermes is doing while you're
4:45waiting for it to complete its task. When it delegates work to sub agents, those are now visible in the main interface with their task, elapsed time, and live transcript. You can open one, give it a correction, or stop it. For example, let's say you've asked Hermes to research three topics, and one sub agent has misunderstood what you wanted. You can catch that while it's working and steer it back. For me, that's the value here cuz you have a better chance to correct the work before you get a finished answer that misses the point,
5:16and of course, save all those tokens. Now, for browser tasks, we're also seeing two useful improvements. Hermes can use logins from one password, Bitwarden, or its encrypted vault without putting the actual password into the model's conversation. That makes signing into websites less awkward than stopping to handle credentials manually each time. Then, there's bot screen. If your bot runs on a Linux gateway, you can view its desktop from Hermes' desktop, take over for a login or two-factor prompt, and hand control back when you're done. That matters a lot because Hermes would almost always ask
5:51for your permission for the next step. But, with the bot screen being able to handle that step and continue makes the whole workflow more practical. Now, the less visible change is how Hermes installs and updates itself. This is more so for new Hermes users. What's happening right now is that source installs now follow stable releases by default. Desktop packages come signed with their own runtime, and the new package manager checks a replacement
6:17environment before switching over to it. In everyday terms, that should reduce the chance of an interrupted update leaving your installation half finished. Now, there are also other few additions like a catalog of 600 reviewed plugins, expanded model support, and better visibility into subscription limits and the model picker. But, the setting I check first is conversation cleanup. This is more so for the OG Hermes users where you have a lot of data in the old storage. What's happening right now is that ended sessions that have been inactive for 90 days are now automatically deleted at startup. Pinned and open sessions are excluded. So, if you want to keep everything, turn off
6:57sessions.auto_prune in your configuration. Also, chats no longer reset automatically on an idle or daily timer. You can use {slash}new or {slash}reset when you want a fresh start. So, all in all, my take on this release is that Hermes is definitely getting easier to fit into a normal routine, especially with the connectors. This reduced uh copying between apps. And also, if you're on local models, it gives you another way to run Hermes. You have more selections now, and the live controls make it easier to step in when something needs your attention. After updating, you can pick one task you already do every week. For example, for when I use Grok Bot, uh I just
7:38everything that has to do with emails, I use Grok Bot for that. but now with the Hermes connector, it's functioning more and more like that Grokbot. And if you want to level up your Hermes, you have to give them plugins. This is a way for them to get connections to external tools so they can get data, use the tool, help you do your job better. So with the new catalog, they added 600 plugins. They reviewed this, the Hermes team reviewed this. Comes with their own instruction manuals for your agent on how to interact with those tools. So definitely check that out. If you find this video
8:11helpful, smash up that like button, subscribe to the channel. My name is Ron, shoning out.
No line in this video contains that word.
Where these words come from. This is the caption track YouTube holds for this video, written automatically by YouTube rather than by the creator. We read it, tidied the line breaks and laid it out so it can be read. The plain text version is at https://viewrankai.com/tools/youtube-transcript/pAaAgcjNgEs.txt.
All rights in this video belong to Superbash (BoxminingAI). Watch it on YouTube. If this is your video and you would rather this page did not exist, tell us and we will remove it.