Hermes Agent V0.19 Just Changed AI Agents Forever!

**Julian Goldie** (0:00)
Hermes has a brand new update, Hermes Agent V 0.19, the Quicksilver release. I'm gonna walk you through exactly how it works, what it means, how to use it, and how to get the most out of it. Plus, there's two new pre-models that you can use with Hermes Agent. And I'm gonna walk you through the Quicksilver Command Center and Hermes V 0.19.
This is the brand new release that just dropped, so that Hermes now replies, actually 80% faster agents finish jobs even if the app crashes. And one gateway now runs a whole fleet of different agents. I updated mine the day it shipped, and I'm gonna walk you through the whole walkthrough today and how it works directly. So, how this works is basically we have the new Quicksilver Core and V 0.19, so the first word in has a response time in less than one second. You can have multiple agents work walking through. Now, what is the problem here? Well, basically with Hermes Agent previously, imagine you had or hired the fastest assistant in the world, then making them sit in a waiting room before every single task. Well, that's what running AI agents felt like before this week. So, before you would ask a question and stare at a blank screen, sometimes for like four to six seconds. I remember testing out Grop 4.5 recently, and it was super slow to reply. So, you would hand an agent a big job, close the laptop, come back to find the job didn't even work, and you would run one agent for everything, so that your agent is super slow. And every command also needed to be clicked, proved like you were just babysitting an AI agent. And adding this all up, it made agents super slow and also not very useful. So, the quick silver command center now breaks that cycle for good. And you might say, well, this sounds like another update. I'll need a weekend to learn. The whole upgrade is super simple and easy. So, what you can do is you can go over to your agent OS, the link in the comments description will go to the AI Profile Board if you want to get mine. And then from here, you can go over to the manage section. And then literally what you do is just click update Hermes right there, and you can get instant access to this.
Now, if you're wondering what Hermes is, it's an open source AI agent from News Research. Lives on your computer, talks to any AI model you point at, does real work. And v0.19 is called the quick silver release. And the name is literal because the team rebuilt how fast it wakes up, how fast it answers, and how safely it can run. Many agents at once. So there's three ideas that cover the whole release. Number one, it answers instantly. So the time from pressing answer to seeing the first word dropped by about 80%.
So cold start went from 4.3 seconds to 0.9. You never lose your work. So background jobs and final replies are now written into a database. So if the app or gateway crashes mid job, it finishes and delivers when it comes back. And you can command a fleet, not a single agent. So one gateway can now route different channels to completely different agent profiles, each with its own model memory skills and secrets. So if you want to indicate how this works, well, you can type a message anywhere. It could be in the desktop app, could be in the AgentAware, so wherever you want. The gateway checks who should answer. The call wakes up in under a second and cold start initialization dropped from 4.3 seconds to 0.9 seconds. So if the model thinks before answering, you now watch it actually thinking in real time instead of just staring at nothing as well, which is useful. Also, it can delegate tasks to sub-agents now in a much more smarter way. And everything lands in your records for sessions as well, which makes it better than ever. Now, if you want to indicate what the release notes, what are the main points here, this is not like some fancy new update where, you know, there's a bunch of new features. It's really focused on one thing, which is speed. And also there's two new free updates for the model so you can plug in to this. So if you look at all of these updates, there's nothing like massively changing here in terms of what it can do. It's really just about speeding up the agent, which I think is what the people want, to be honest with you. So this has become much better at delegating to sub-agents. And then the main feature here is it replies, land 80% faster, which is fantastic. Also has smart approval, so there's a second brain that checks like the risky stuff. So if you've ever like left an agent running, and then you're worried about what it's going to do when it's running in the background. Well, basically, you can switch it on by default so that every basically flagged command gets approved by you manually, or you can switch on full auto. So you get a choice between full auto mode or getting every approval, every command getting flagged by hand. So you can choose between manual and auto approvals. And there's also a third way now, which is it's on by default. When a command gets flagged, a separate AI independently assesses it before it runs, which is great as well. And you can also set up your own deny rules as well, that block stuff in auto mode as well. So if the agent wants to run a command, there's a smart gate where a second AI assesses it, or it can run or it can ask you as well.