OpenHuman VS Hermes AI VS OpenClaw: Best AI Agent?

Today, we're going to be testing our OpenHuman, a new free open source AI agent compared to OpenClaw, compared to Hermes. It's trending, it's absolutely growing like crazy at the minute. You can see over the last 24 hours, it's got 3,000 stars, total 8,000 stars on GitHub, open source project. It was featured by GitHub as a trending repository of the day, and it's described as your personal AI super intelligence.

Private, simple, extremely powerful. So we're going to be testing out today, and you can see it's available at tinyhumans.ai forward slash OpenHuman, so we can download it from here. The other cool thing I like about this is it comes straight out of the box as a desktop app, which I think everyone wants. Why has that not been done so much before?

You do have stuff like, for example, you've got Hermes desktop, you've got Atomic Chat, which can actually have OpenClaw inside a desktop app. You've got AonUI as well. But the thing is, why do none of these tools just launch their own desktop app? You know, I think 99% of people don't want to go into the terminal.

I just, maybe it's just me, but I feel like 99% of people don't want to go into the terminal and start using an AI agent like that. Like that's something most people do not understand, right? Having said that, I do, I'm a big fan of Hermes, so, you know, don't get me wrong here. So we're going to test out OpenHuman now, and we'll compare it side by side in a minute as well versus other stuff, for example, like Hermes agent.

Let's see what we got in the comments here. You are, you are, who says OpenHuman name is promising. Yes, I would agree. And you can see here, it says it needs a running core to operate.

I want to, let's see if we can just run that locally. Let's see how it goes. So we'll go local, see if this could be set up here. Then we've got approve the settings.

Ah, so it has computer use in. That's why it's asking for access. By the way, peeps, I'm installing this, right? And I'm installing this AI agent.

I tested stuff out for you, so you don't have to. All right, so just to be clear, like I'm downloading this AI agent. If you don't feel comfortable testing out this AI agent yourself, don't do it, right? Don't link it to anything you don't want to.

I'm actually linking out to a different Gmail, for example. You know, like these AI agents are open source projects, and you don't know what you're getting with open source projects. So just be careful with this sort of stuff. I tested out for you, so you don't have to.

Just see that you're aware of this. So you can see it's logging in right here in the background. Let's see what we've got. Tiffany says, interesting.

We've got maybe Steve Jobs. He says, hello. And then we've got Milad, who says, so it's free, and all it is is a download instead of high technical stuff. Exactly, right?

That's what it's designed to be. Having said that, it's struggling to log in. So let's see if we can get it logged in here. There we go.

There we go. All right, we're back in the game now. So we've set that up. So we just have to log in to Gmail in the background to get that set up.

You can run it on a local as well. Sorry, on a cloud-based. And then it's in early beta. As you can see, they've actually got a Discord here that you can join, which is always promising.

And then we can start opening this up. So let's click on Let's Start. Then we can connect our Gmail. Let's just test it out.

Again, do I recommend connecting your AI agents to Gmail? No, right? Like, no. You can set it up with a Gmail that you don't really use, which I tend to prefer.

But even then, just be careful with this stuff. Because you don't know what these AI agents can do and all that sort of stuff. So what you can do here is you can change the permissions. If you want to be careful with using AI agents, then what you wouldn't do is give it, for example, write permissions or admin permissions, right?

But you can if you want to. So it depends on what you're comfortable with, essentially. That's what I'm telling you. And just be careful.

Anything you're not sure of, don't do when you're using these AI agents. So it says, let's get started. Your device is connected. Keep the app running to keep the connection alive.

Message your agent with the button below. It hasn't asked me to add an API yet, but I'm sure it will. I mean, it's super easy to use, isn't it? Look how easy this is.

You can connect to all these different apps, which is pretty cool. I mean, compare this to the onboard. Have you ever seen the onboarding? I mean, I'm talking to many people here, but if you've ever seen the onboarding for using Hermes, right?

So if you go to Hermes setup, this is the onboarding over here for connected to apps. This is the onboarding for OpenHuman. Which one is easier? Obviously, it is OpenHuman, just so that you're aware here.

Now, this is a brand new app. I've never tested it out, but we can connect it to Telegram and that sort of thing, kind of like OpenClaw. You've got your assistant's brain over here. So this is where your assistant learns to remember.

So you can set up the intelligence for it. Then you can make it yours inside the settings. And then we can go from here. Let's go.

All right, it did mention something about a subscription a second ago. So let's just have a look inside the settings. Yeah, so settings at the bottom you can switch to here. And let's just see what the settings are like here.

So it does have a subscription plan, I think. Even though it's open source, which is strange. I think if you, hmm, I wonder why that is. Let's see what we got here.

Prince Brian says, OpenHuman seems interesting. So here you go, right? So you can use free local models. You've got unlimited connections, unlimited apps, unlimited devices, right?

And then you can use like a free credit. So it looks like you do need to set up a plan, really. They do have this code that you can use. And then you can switch here.

You get 10 times more usage. And pro, you get 20 times more usage, right? So just be aware of that. But I'm pretty sure you could change the model that you're using inside your settings here.

And then you should be able to use that with like a free API. But I haven't tested this out. It's the first time I'm using it, right? We're testing out together.

That's the whole point of the journey. Let's see what else we got here. A-Man says it's not free. So it does have a free plan.

Like, let's be honest here. It has a free plan. So just being 100% honest with you there. It does have tiers as well, but it has a free plan.

So just being 100% honest. What else we got here? DJI says, I love Hermes. Me too.

Big fan of Hermes. A-Man says, so you can use the free plan, right? Just coming back to that, A-Man, right? Just being 100% clear.

So some people are asking, is it free? You can use the free plan. And then it's got an upgrade and a pro plan there. And then we got Robert who says, hello, Julian, Hermes.

And Prince says, is there a beginner's tutorial on Hermes? Yeah, absolutely. So if you go to the AI Profit Boardroom, link in the comments description, and go to AIprofitboardroom.com, then you're going to type in Hermes setup here. And we've got a full course on how to use Hermes here.

And we also have loads of different tutorials, depending on what you want to learn. But if you just want to get started from scratch, you can start with that guide right there. And that's all inside here. And also, if you ask questions, I answer questions daily inside the community.

So for example, this morning, Ted asked his question, and then I created a video for him. Gave him the bullet points in case he wants a TLDR. But then I gave him the YouTube video right there. Milad says, is Atomic chat free?

Or is the credits needed? So if you use local models, it's free. Or if you plug in your Alarm API, then it's free. All right, let's get back to OpenHuman.

So we're going to test this out. We're going to type in message, OpenHuman. And one thing I want to try first is just switching over the API to a free one. So I'm going to go to OpenRouter here.

So you can see you've got different providers, right? Now, if you go to OpenRouter, we should be able to plug in an API key. So I'm going to go to OpenRouter and grab Al-Alpha, which is a free API available for anything. That can work with Hermes Agent.

It can work with OpenHuman, whatever you want. You've got the details right there. So you can then change the model you use for everything. So for reasoning, for agentic, for coding, for summarization.

So I'm just going to change all of them to Al-Alpha, as you can see. And then we need to get an API key, which I can get in a second. And then we're going to go over here and we shall go to my settings, workspaces, API key. And we just need to grab a free API key from OpenRouter, right?

So it's using Al-Alpha, just being 100% honest with you. Al-Alpha is a free model you can use for OpenRouter. You could also use Olama. So you can use local models with this.

And the other option you have here is that you could use custom, right? So you could use, for example, like NewsPortal. NewsPortal has DeepSeek v4 Flash and also it has Step 3.5 Fun available for free as well. So we're going to select those.

We've plugged in the API key here. It's saved in the details. Let's hope it works in a second. Let's just check it works.

We'll put HeyHey inside the chat here. So, I mean, it's super easy to set up, right? Like it started working straight away, which is awesome. And you can see it's replied to me using the free API from OpenRouter, right?

Which is really cool. Now, OpenHuman is free to download. You've got a free API plugged into there. I don't know if it's going to ask me to upgrade, but so far, so good.

So we've also got this option here. So we've got human and we can actually speak to our AI agent inside the chat, right? So this is OpenHuman and we can speak to it and then it should be able to speak back to us. So let's just try this now.

It's going to ask for access to our microphone. We'll say yes. Hey, are you working right now? Are you working right now?

And then let's try it again. And it says, yeah, I'm ready to go. Okay. So it transcribes.

Let's test that again. Hey, are you working? Let's see if the transcription works here. And it says, yes, online.

I'm ready to help. So it can actually talk to you right there, right? And it's replied to me. So the text-to-speech option is super easy.

As you can see, that was really, really easy. easy and then also we've got the chat over here right so so far everything is working pretty smoothly so the chat works here and then the human text-to- speech works perfectly as well by the way try and set up with open claw mate if you try and set this sort of stuff up with open claw it can work I've done tutorials on it but is it messy does it work first time round is it as easy as that no not at all right so just be aware of that and you can see here how many credits we've used right so you can ask it to do stuff and then it gives you your details right here and so it does give you like free credits to start using it and then you could just test out see what you like now so let's just check this out so if we're at 0.92 here let's just do a little test and we'll see how much a message costs so I posted this message here and it's not used up any of the 0.92 as far as I can see which is good now also got connections so if we wanted to for example connect this to our calendar or to Google Drive or to Google Docs we can do that over here right so let me show an example that like we can connect to Google Docs here it's gonna pull up our account and I'm gonna connect it to my other gmail account as you can see then we're gonna select all click continue and we've connected that like so right now it's gonna ask you for permissions so if you just wanted to you know not allow it to edit anything you can just have it read right and you can switch off if you want like full access in terms of like being able to share in terms of being able to delete admin access then you would click on that right I think for me read and write is totally fine for you it might just be read right if you want to just play it safe with this stuff so if we go inside the chat here and then I'm gonna take one of my favorite prompts for AI SEO which you can see all right so this is an example prompt that I use for creating content with AI and it works pretty well on my other websites like let me show you some examples of this show you some examples of like our websites and the ones that are doing well etc and you can see here for example this is a website that's doing pretty good so far and it's growing nicely that's growing from like literally zero clicks a day all the way up to here and you can see the trajectory as well it's pretty nice too right here's another example as you can see this another example we've got one here as well they're all sort of following the same trajectory right and I'm actually using Hermes AI SEO for that but look at that trajectory right it's awesome so if we go back inside the chat here I've plugged in my prompt one thing that seems slightly annoying is that if you paste in the long prompt like this you can't really see it properly if that makes sense right so it might not be ideal I don't know if there's another I can't see any other views you can't switch between like work briefing etc so you can set up a new thread just for work if that makes sense and then we can paste in this but it's not that easy so that's one sort of thing that's not ideal of it right there's like a long prompt in here but I literally can't see it whereas for example if we go into let's say you're using Hermes desktop for this or Aon UI, Aon UI is pretty good for managing all your agents with OpenCore and Hermes and everything else when you're doing that like if you paste in a prompt you'll see the full prompt right here it's pretty easy to see what's going on so I'm gonna type in a keyword here so we'll put for example Hermes web UI plug in the prompt see what happens next a man says after this after the credits my app stopped midway even though I was on a local model you'll see how even though you use a free API the credits will deplete I haven't seen that so far like it's still at 0.92 even though I've done a long prompt here but maybe it will update in a second like this is the first time I'm using it but I'm not seeing a problem here as you can see like it still stayed at 0.92 even though we've put in a long prompt and I've tried to test it as you can see right here right and let's see what else we've got here how much memory do you have so I'm on a Mac studio right if you want to see the details here let me click on about this Mac you can see so we have 36 gigabyte memory but I don't really run local models right I just run AI agents and then a plug-in API through them so even on 36 gigabytes like I don't run local models most of the time is there a comprehensive guide of hardware requirements other than M4 Mac mini to run local models for the agents yeah I would check Olamis documentation and I'd also open up LM studio right so if you go to LM studio it will tell you which models are the best for your setup and the way that works is LM studio is a free program you can download and then you can see your options in terms of what you can use and if it's too large or too powerful for your local setup it will say likely too large right so that's how you can figure out what works for your specific setup and then also you can just plug in the documentation ask Olam or something like that you see his research what it is features benefits is it okay all right so you see now I can see my actual prompt here but that wasn't ideal so what I'm gonna do over here is say create an optimized article for this and then we're gonna go inside the chat and put it in Google Docs and then I'm gonna take a case study this is what I do whenever I'm creating you know content for my websites I will take an example case study like this and just grab that case study and then plug it into the AI right so it's got context on what we do now you can see it has used some of the 0.92 so use 0.01 but again like we're doing a full prompt here for AI SEO so it's pretty good as a test jumpers says I've tried so many agents and models and local models it's unreal none of which I've managed to get running properly or seem to crash not to ask so I would check out the the AI problem boarding like we've got loads of training loads of people getting awesome results for this so I'd recommend going out is there anything similar to an easy setup of some sort of agent that can browse the web on Android I think actually Google are releasing something soon they've actually they've talked about it recently where they're basically offering it's like a Chrome feature that's coming out soon let me try and find the details on this yeah here's an example of what it's gonna look like right it's coming to the US first so it's called auto browse with Gemini and Chrome and that's from Google and they're basically releasing this which looks really cool now this is not working so well in the chat so as you can see here it doesn't seem to deal with long messages very well right so we've given it an actual task there if I gave that exact same task to Hermes is it's gonna crash that task right let me show you an example of this so let's say for example I want it I want Hermes to create an SEO optimized article for this and then we open up Hermes over here and we type in Hermes and we're just gonna set this up so I mean from what I'm seeing so far like open human is great for non-technical people but it doesn't seem great for like word tasks for example like creating a blog post right it seems to struggle with that so I'm gonna plug in the exact same prompt here into Hermes and say do this and then you can go from there so let's see what Hermes is doing yeah so it's going off and it just handles that task perfectly right like it's just going off and creating it I'll say like don't worry about Google Docs but just create the post and as a local markdown file right so I posted that inside the chat and you just see like the quality of outputs are much better from Hermes than they have been from open human right like open human really failed for that task like I tried to give it a chance and it replied with this I gave it another shot it got confused and struggled again and then you can see it struggled again here so it's cool on the outside again this first time I'm using it it's pretty cool on the outside but for actually being autonomous for being a really powerful agent that can actually just handle anything that you throw at it I think that Hermes is way better right and you can see an example right here like it's already written the content it's told me exactly where it's creating the content if you're gonna do AI SEO Hermes seems to have crushed that task whereas for example if you're using open human it seems to struggle with it completely now Milad says is there a free fast setup of Hermes absolutely right so what you want to do since it's pretty simple so you just you use a quick install instruction here on Hermes GitHub right that's free to set up and then if you want to set up free just use Hermes with our alpha right Hermes with our alpha if you need help setting up we've got a full guide here as you can see inside here but basically you just use our alpha with Hermes agent and then you've got a free API that's running right and you can install Hermes in one click like I've just shown you so that's how I'd handle both of those things alright so Hermes completed the task in one shot and open human totally failed at that task right it was pretty cool at setting up the messaging it seems pretty useful for that not great for actual work tasks now I really like the idea of the text-to-speech option as well I think that's pretty cool but bear in mind you do need to keep open human running on your computer to use it elsewhere right to keep it connected they actually told me that during the onboarding if you saw earlier now if we go inside the chat as well yeah it's I mean we've got that free usage here and it's it seems to have it seems like we've only used 9% of it so far out of everything that we've done which is not too bad if you go inside connections we can connect our gmail we can connect it to Google calendar we can connect this to air table Google Docs etc you see how it's connected to gmail here so if we say okay send an email to Julian Goldie at gmail.com being hello just as a little test we'll see if that actually works and we'll see what the tool usage is like right so it looks i like the ui i like how simple it is it's non-technical to set up let's see if we can actually use tools and send the email properly right so it says i'll send that email for you but has it done it that's a question we all want to know so i'm going to go inside the chat here and say have you sent it now also bear in mind like the quality of the outputs you get from this might be because it depends on the ai that you plug in right so i changed the default models i changed the settings maybe it would be better with like the default inside um for example using not like an api but using the recommended settings from open human so that's something to bear in mind as well let's just check if this works i've asked it three times now so i've said send an email have you sent it did it work let's see if it finally works if we yeah let's let's check this out i think it might be because we're not using the default settings from open human right so if we go back in the sense here and we change the model details to lm provider and then we switch to open human as you can see we'll save that and then we'll just test it again right i'll do this in the new thread so we'll say okay send an email to this account saying hello so it's now using yeah so look it looks it works much better if you're using the open human connection details but obviously bear in mind like if you use the open human ai it's going to run out of credits faster if you have any questions as we're moving along here feel free to ask all right so it says done sent the email let's have a look here see if it's actually done so if we go to my account on emails there we go right so it was really easy to connect that if you're using the default settings from open human it actually works really well and that was super easy right far easier than for example setting up hermes with email so again i think if you want something that's just fast efficient doesn't do anything too fancy is non-technical and easy to set up go with open human but if you want something that's fully autonomous customizable open source and doesn't require you know you can use it 100 free bear in mind hermes is totally free then i would go hermes so those are the biggest differences so you can have more connections here you can set up the memory we can actually i mean you see it's actually connected to obsidian so we can sync this over to obsidian which is good that's where i basically store all of my context for air agents you can see an example right here and this basically organizes all of my notes and ai agents write new notes for me it's a nicely organized system i called it the infinite context engine and basically what this does inside obsidian is i've got omi taking notes on me all day in terms of what i work on and then it writes it into a memories vault inside obsidian and then obsidian links all my agents right and that's how they understand what i'm doing who i am my memory etc and uh yeah it's pretty powerful stuff here's an example of omi what it looks like um all the memories that it tracks etc and that's pretty cool right really easy to use so if we go to open human here let's turn there you go right if you want more training on that stuff go to this tutorial right here and it's called the infinite context engine it's a full step-by-step process for setting up a second brain a full memory for your ai agent so i could just plug that into open human and then it'd be super powerful to do that milad says what's the best free desktop app that is the most powerful so i would recommend using um homie's desktop it's pretty good i really like homie's desktop it's a great tool it simplifies things you know instead of using the terminal you can just use homie's desktop and everything looks clean and easy and then we've also got alerts here so you can set up notifications that sort of thing um it's kind of gamified and that's basically it that is how to use this setup now one thing i wanted to check as well because it's important like if you're doing for example aiso with an agent you want to be able to schedule content you want to be able to schedule tasks daily right now if we use for example hermes it's really powerful for this i can just say let's schedule a task uh do a blog for me daily 5am every day publish it to my website it's good to go right if you go inside this agent let's just see if it can do that so i'm going to say set up a daily task where you post a blog for me each day see if it works and then it says which platform okay i'm gonna say just like replying you can post it in the chat here you write the post 5am each day almost style about ai automation is there an easy setup for pica claw on android um that's super niche but not as far as i know i mean like you can plug in a free api into pica and then i think you can connect it to android but i've never tried that but yeah that's probably the way that would do it oh look at that so it doesn't have scheduling tools here this isn't great so like for example even open claw has scheduling tools and then you've got for example um you got hermes so it can easily schedule tasks right if i if i go into hermes here and i say do this right every day at 5am or whatever it's going to just schedule that task in no questions asked right super easy and simple to do so just go off and do that for me and you can see here it just schedules the task right no questions asked so in that sense you know the fact that open human can't schedule tasks but hermes can open claw can again i think hermes is winning here by a long way versus everything else and that's basically it that's basically it so thanks so much for watching that is my guide on open human versus hermes agent versus open claw which one is best as you can see today i would say that hermes agent still wins if you want to get all my best trainings on hermes and open core you can get that inside the air profitable link in the comments description or go to the air profitable.com we have courses on open claw and hermes and you can see that we add new daily trainings and update this stuff with all the best stuff right so if you want to learn ai automation if you want to learn how to use ai agents if you want me to answer your questions with video tutorials every single day which is what i do inside the air profitable then i would recommend joining here and you can post questions inside the chat i mean here's an example so for example here you know we got this question from ted and we answered it with a video and broke it down in a step-by-step guide for him right and that's what we do every single day that's what i personally do we've also got all of my best trainings inside the classroom so for example you can see here that we've got a beginner to expert guide we've got a full course on how to get more clients we have my playbook on how i for example generate ai avatar videos x automations how automate shorts etc and you can also learn how to do ai seo you can watch back the coaching calls if you want to jump on live coaching calls you can jump on these coaching calls they're four times a week on ai automation you can ask questions about your hermes setup about open core about ai agents in general you can connect with other members inside the community here in your city and local area from the map here as you can see and then that's all available right here also if you join also if you join you'll get all of my best trainings on like all sorts of cool stuff around ai agents hermes setup kits etc so thanks so much for watching i appreciate as always i'll see you on the next one also if you want to get a free one-to-one seo strategy session that shows you how to get more leads more traffic more sales with seo to your website then feel free to get this it's available at goldie.agency link in the comments description you can get a free seo domination plan so you get a custom tailored link building game plan so you can generate more leads and sales you can discover the secrets seo link building we can answer any questions you have one-to-one on that call you can learn the best link building strategies for your website plus how to outrank your competitors with link building that's all available goldie.agency get a free seo strategy session i'll see you on the next one cheers