OpenClaw vs Hermes: Who Wins?

**Julian Goldie** (0:00)
So today, we're going to be comparing OpenClaw vs Hermes to see who wins and what the differences are. Now, this is based on a question I got inside the AI Profit Boardroom. So Sophia was asking about OpenClaw, what's the difference? I think a lot of people are confused which one is the best to use and also what makes them different, why they're similar, what tasks are they best for, right? So basically, to summarize quickly, what I've found is that Hermes is a lot smoother and easier to use, and that OpenClaw tends to break a lot more. The other thing that I've found is that because Hermes was engineered by one of the biggest AI labs in the world, News Research, because of that, they've basically created something that evolves over time, it gets better and better, right? Whereas for example, if you look at OpenClaw, it's become worse over time. Let me show you an example right here. So if we go over to OpenRouter, and you can see the general consensus, so you can see how these agents are performing. If you look at the most popular agents in the world, Hermes and OpenClaw are right at the top on OpenRouter. OpenRouter is a place where people use APIs, so you can get a feel for what is being used the most. Now this year, Hermes launched in February, so it's newer than OpenClaw, and Hermes has actually gradually overtaken OpenClaw directly. So for example, if you look here, you can see that the subjectory of Hermes is that more and more people are using it over time. The graph is going up, the trend is going up. If you have a look at OpenClaw, it has declined.
Like over the last 30 days, but I've seen it drop in really since the release of Hermes. And that's because it's just a much smoother experience. Now, when you're using them side by side, recently, I have noticed that OpenClaw has become smoother to use, it's less buggy, it breaks less. But I tend to find myself using Hermes more and more. And also they're releasing some interesting stuff that's actually useful. So, for example, OpenClaw, they've released a bunch of interesting features recently inside their releases, right? So, for example, they have brought out stuff like being able to talk to your agent inside Google Chat or Google Me. They've actually got an app out as well. But I think most people are not using this stuff, right? The main use case I see for using AI agents is that you have them working as a team, you're doing some work on your laptop, and you can start using OpenClaw or Hermes directly on your laptop. Whereas, for example, I don't know many people who actually use OpenClaw inside Discord or Slack. It tends to be like they want to go to one centralized hub where all of the agents are living together, like Claude and Oberclaw, like we've got inside this agentic OS system right here. Now, when we actually look at these systems as well, they're very similar in terms of the models they can use. If you actually look at what are the most popular models that people use inside Hermes, you can see the top three right here. If you have a look at OpenClaw, it's similar, right? It's not far off at all. Now, they pretty much have similar functionality in terms of the models they can use. They can both use free APIs as well. So they can both use stuff like our alpha as well, which is a free API on OpenRouter. And also one big difference between them all is that Hermes, I've seen they've organized their knowledge a lot better. So let me show you an example of that. If we go to the news research portal here, you can see that they've organized their knowledge better. Whereas for example, if you want to get the same thing for OpenClaw and the documentation on OpenClaw, they do have an OpenClaw docs set up here, but it's just it doesn't feel as easy to navigate around. It's not as useful as I find myself coming back to Hermes agent and learning new things all the time. So you can see, for example, they have a Kanban set up, which you can't get with OpenClaw. They have the persistent goal set up, which is their take on RalphLoop, which is a way of just using agents autonomously as well. So there's some feature differences too in terms of Kanban boards, persistent goals, they've added MCP catalogs as well. Some interesting features that I don't see anywhere else. Also, when I look at what OpenClaw is releasing versus Hermes agent, I feel like Hermes agent is actually listening to what people want, paying a lot more attention to what are the problems and then how can they solve them. Let me give you another example. So today they actually released something called Hermes Desktop. Hermes Desktop is a free application that you can use and you can plug your Hermes agent too. Now, why did they do that? They did it because there was no gateway for Hermes agent privacy. You couldn't really chat to it unless you create your own agent operating system. As they knew that if they wanted to keep growing beyond the current trajectory of where they're at, they had to release something that was user friendly and easy to use rather than the terminal, right? Like how many people are just going to open up the terminal and then start coding, essentially coding inside a terminal command line inference like this. You're not going to do that, right? So that's why they've released Hermes Desktop. You don't really see that for OpenClaw so much. You don't really see them looking and paying attention. I see a lot of the releases, they sometimes they bring out shiny objects or flashy features, but these are features that 99% of people are not going to use.