OpenClaw: Discord + Parallel Agents are INSANE!

Open claw is the most powerful technology that I've ever tested of all time. And with parallel agents on

time. And with parallel agents on Discord, you can make it even more useful. Here's a problem that most

useful. Here's a problem that most people don't even know they have. So

right now, if you're running Open Claw through Telegram, for example, your AI really, you can only see it working on one thing at a time. You might get occasional updates, but it's not that useful compared to have multiple

parallel agents and teams of agents that can work [music] across multiple channels, which is what you can do with Discord. So, for example, with Telegram,

Discord. So, for example, with Telegram, you might send it a task, you wait, you send another task, you wait, right?

That's like hiring five employees and making them all [music] stand in a single file line to get through one door. It makes no sense. Discord changes

door. It makes no sense. Discord changes

everything because Discord has channels and each channel can become its own separate AI agent working at the same time. So, whilst one agent is handling

time. So, whilst one agent is handling content, for example, another is doing your research, another one is managing your client projects, all running in parallel simultaneously. right now. By

parallel simultaneously. right now. By

the end of this video, your open claw is going to go from doing one thing at a time to running six or seven projects at the same time. [music] Let's run through it together right now. So, you can see Discord running right here. So, we

tagged it in actually adds like a emoji once it's reacted and we can tag it in and then get answered right [music] here. So, for example, I've got two

here. So, for example, I've got two different channels running or three different channels with open claw like this, this and this. And then I can set up multiple different agents running in parallel across multiple [music]

different tasks. Right? So for example

different tasks. Right? So for example here I could say okay openclaw give me some ideas right and you can see it's just waiting online to respond. So, give

me some ideas for marketing my AI automation community. Right? So, we can plug that

community. Right? So, we can plug that in and then you can see it's already looking and responding. And then over here, we can set up another response and say OpenClaw

give me some video ideas for OpenClaw related tutorials, right? And then

you've got multiple different agents working at the same time, responding at the same time, and working on different tasks, right? So this one has already

tasks, right? So this one has already replied whilst this one is working on.

The other thing that I like about this, as you can see, it's responding in real time. The other thing I really like

time. The other thing I really like about this is that you can sort of separate what you're working on. You can

train the agent in different ways, right? So for example, you could have a

right? So for example, you could have a skill MD for video ideas and you could have another skill MD for marketing and then you can get responses in real time as you can see. So it's a really

powerful way to run uh agents directly and it's much easier than running something than say Telegram. If we go over to Telegram example, so for example, if we go over to Telegram here, it's kind of difficult to manage like

multiple different agents um and you've got like one chat and it's kind of linear, right? You could set up multiple

linear, right? You could set up multiple different chats on the left hand side, but that's it's more difficult. It takes

more time, right? Whereas, for example, if you actually go over to Discord, the way that you can set up multiple channels, um I'll show you in a second to create parallel agents working across different tasks is really, really simple

and easy, right? So, all you do is you go over to Discord like this, right? So

let's say you got discord like this already set up with open claw and I'll show you how to set that up in a second as well. So you can just right click the

as well. So you can just right click the channel that you want to run click on duplicate channel rename it something else. So for example, we could say okay

else. So for example, we could say okay let's create one for uh product ideas like this hey create

channel and then we can tag open claw and we can say okay in this channel

focus on ideas around improving the AI profit boardroom school community right what are three ways

we can improve improve school member retention rates. See, it

instantly responds. It's instantly set up. It's really easy to set up these

up. It's really easy to set up these separate channels and um it's incredibly fast, right? So, it's already

fast, right? So, it's already responding. And then we can now work

responding. And then we can now work across three different Discord channels in real time. You could have seven if you want, six or seven or whatever you want. Um and it's quite simple and easy

want. Um and it's quite simple and easy to do. Refer says, "How do we get clawed

to do. Refer says, "How do we get clawed API for free for open claw?" So there's only really one way that I know to do that and you you really get limited tokens with it. So what you can do I

actually got a a tutorial on it yesterday is you can run the new update with open claw using kilo code right once you set up the kilo code gateway you actually get $5 of free credits and

you can use the default API via kilo code gateway which is clawed you get just $5, right? So it's not like a ton of responses but that's the only way that I know you can get a clawed API for

free. meme says, "Do you have a GitHub?"

free. meme says, "Do you have a GitHub?"

I actually don't. I'm not really a developer to be honest. Like, I just I like to show and teach people cool stuff, but I don't really like, you know, publish projects like that. I just

sort of share what I do on Discord, like you can see, uh, sorry, on YouTube, like you can see. So, I just think this is a lot easier to run than something like Telegram. It's a lot more organized. It

Telegram. It's a lot more organized. It

keeps everything together. And I'll run you through like the core problem with Telegram directly. So when you run open

Telegram directly. So when you run open claw through telegram everything is kind of sequential. you can have like

of sequential. you can have like multiple sub agents working together, but again like it doesn't really feel like you're working across different tasks and different projects when you do that, right? Um so usually you just send

that, right? Um so usually you just send a task, wait for it to finish, then send the next one and like all the messages cue as well, right? Same with the gateway. Whereas with this you can have

gateway. Whereas with this you can have um multiple tasks and that means there's no ceiling on your productivity because you've got multiple channels, multiple tasks like you've seen today that work across each other. And this is why I

think Discord wins for parallel workflows. So Discord has channels. Each

workflows. So Discord has channels. Each

channel becomes an independent agent session. So it's the same agent, the

session. So it's the same agent, the same workspace, the same files, but complete separate conversations running simultaneously. So there's no like

simultaneously. So there's no like context sort of merging between each conversation as well. That's the other cool thing. And instead of working on

cool thing. And instead of working on one thing at a time, you can run five to six projects in parallel. Each channel

becomes a specialized agent that remembers everything inside that channel. Right now, if you want to set

channel. Right now, if you want to set it up, we've actually got the instructions right here. Basically, what

you're going to do is just create the app. So you go to your Discord developer

app. So you go to your Discord developer portal, new application, give it a name, configure the bot, set the permissions, and then add that to your server, update the OpenCL config, and restart the

gateway. Right now, another easy way to

gateway. Right now, another easy way to do this if you if you want to skip all all the messing around, and this is something that worked really well for me this morning before I was doing this live stream, is I actually just opened

up Claude Code, gave it the bot token for Discord, and then just kept iterating um typing any errors into claw code until it eventually got fixed. It

took me about 10 or 15 minutes to set up. So, it was pretty chill, pretty

up. So, it was pretty chill, pretty quick and easy. And also claw code can give you the instructions on how to do this step by step if you're struggling with like the actual documentation that's official from open claw. So it's

pretty easy to do. And reefer says is it possible to uh do something like that LLM like claw directly through open claw so that my

cost is lower. So what you can actually do is you can run a local LLM like for example GLM 4.7 flash. I have lots of training inside the AI profitable bottom

on how to do this. So, if you go inside here and then you can see we have a full tutorial on how to run GLM Flash plus O Lama for free, which you can see right here. So,

we have a full video tutorial on it and a step-by-step guide on how to set that up, plus a 30-day plan on how to really master it. But basically, if you do want

master it. But basically, if you do want to get like a free API that's kind of not as good as but right up there and still one of the most powerful models in the world, you can run OAM with GLM 4.7 Flash and then you get a

free API that's plugged in and running locally with OpenClaw. That is a free way to do it. And with that, it's pretty cool because if you've got a Mac Studio or something like that, you can run the API locally. It can run 24/7 on your Mac

API locally. It can run 24/7 on your Mac Studio and it's just running in the background and you can use the bot as much as you want, right? You can use open core as much as you want without running through um credit. So that's one

of the [snorts] other options you've got. Also, for example, you can use uh

got. Also, for example, you can use uh other free setups. So there's there's multiple free methods. So you could use for example miniax a warf and then you get free days on uh seven free days on

the coding plan. You've also got a llama that you can use with kimk 2.5 and then you've got codeexos. So codeex OS is actually free for a limited time. You

can set up open claw with that and then set up a with codeex. That's another

option. And then the final option is um kilo code where you get like $5 credit.

So those are multiple different options that you could run directly to use open core for free. Which AI model are you using? Well, I tend I mean I always show

using? Well, I tend I mean I always show new stuff inside my videos, right? So,

when I'm using Open Claw, basically I'll switch between different APIs uh depending on what the new updates are.

So, for example, like yesterday, Kilo code just came out with open claw. So, I

did a video about setting that up and how to use it. The day before I used Mistra. So, Mistra was a new update that

Mistra. So, Mistra was a new update that just came out this week and you can use that with OpenClaw too. The one that I've seen that's the best for free is just chat GPT codeex. The second best is GLM 4.7 flesh. The third best I would

say is communicating 5, but it did struggle with some local tasks. So, you

know, you you've basically got like a a choice between quality and cost, right?

So, if you want like the best API in the world, like Claude Sonic 4.6 or or Claude Opus 4.6, well, that's going to get more expensive, right? Because

you're using literally like the best AI in the world. But if you want something that runs for free, um, and you just invest in something like a Max Studio, then you can use GLM 4.7 Flash directly

locally. And I've shown you the other

locally. And I've shown you the other options with Miniax and Kim K2.5 as well. Another option is you can get an

well. Another option is you can get an API from Nvidia. Nvidia is another option that you can plug in, but I've heard mixed responses about that. So

that's why um, I don't recommend it normally. All right, so that's basically

normally. All right, so that's basically how you run parallel agents like you see. So we've got multiple agents. You

see. So we've got multiple agents. You

can have multiple personalities, multiple skills. You can work across

multiple skills. You can work across workflows across different ways and you can easily just create a new channel like that. You just click duplicate

like that. You just click duplicate channel using the same options, right?

And if you're struck if you if you're stuck like setting it up or you're struggling with the onboarding or anything like that, just get claw code to help you along the way because that's a really good CLI for just setting stuff

up locally running with open core right and basically you can just open up um the terminal. Let me show you an example

the terminal. Let me show you an example of that. So for example, here you can

of that. So for example, here you can see discord and then we have claw code running in a terminal here. It's just

helped me set up like the server ID and everything else and update the settings.

So that's a good option uh that you can use for setting this up and just getting it working pretty nicely. ID says I just loaded a private LM local on my computer. I have a question maybe you

computer. I have a question maybe you can answer. Can you train these models

can answer. Can you train these models locally on computer? I'm not fully aware of that. Like I think you would just

of that. Like I think you would just train up uh openclaw and then that's the way that you would train up to be custom, right? So you can for example

custom, right? So you can for example set up like a soul MD file or you know like every time openclaw makes a mistake you can actually create a learning MD

file and then tell it to track all the mistakes it has and then use those before it responds anything that you do in the future so that it can learn from its mistakes and just become self

iterating. So you can selftra uh any

iterating. So you can selftra uh any version of open claw using that method instead. Can we build a system in open

instead. Can we build a system in open claw that automatically replies to customer messages because I have an e-commerce business? Yeah, I mean like

e-commerce business? Yeah, I mean like you know you can you can set up a chat you could have like a group WhatsApp chat um you could have a group discord channel even you know between between you and the customers and that sort of thing. So that's a good option for

thing. So that's a good option for setting up chat bots in general with AI fast bots is pretty good. Also end to end is really easy to set these sort of things up as well if you don't want to use like a group discord or a group WhatsApp or that sort of thing as well.

The other option as well is like you could create a separate email for open claw that's like your customer service email inbox and then when customers email you, you could have open claw

respond and just be fully transparent that you're using AI Asian. I've seen

some ecom companies do that where it said it gives me really fast responses using AI, but it will say at the bottom this is a response from an AI agent in training,

right? And that's a really good way to

right? And that's a really good way to to get your customers quick responses, but also to be very transparent about using AI agents to respond to customers.

I can pay whole config, but Julian already explained almost same thing. So

you can get a decent config from any model out there and just ask it to create examples uh like open claw. Yeah,

exactly. So I installed GLM 4.7 flash, but it's lagging on Yeah. So like if it's lagging, you know, using Max Studio, if you don't have the power for it, then you can use like a light local

model. There's loads of models on O Lama

model. There's loads of models on O Lama that you could use. Or the other option is that you can just use the 7-day free coding plan for Miniax or something like that. David says, "Can you please link

that. David says, "Can you please link the video where you created your current AI clone?" Yeah. So, if you go to the AI

AI clone?" Yeah. So, if you go to the AI profit boardroom like so, go inside the classroom here, then go to the playbooks, you can see our full process on how to automate AI avatar videos like

this. So, if you want to get the full

this. So, if you want to get the full system for that with the video tutorial, the step-by-step operating procedure, etc., This is the same system we give out to our team to train editors on how to create this and you can see the whole process right here. So it's inside the

AI profit boarding link in the comments in description or just go to AI profitable boarding. Can I use a VPS

profitable boarding. Can I use a VPS server? I wouldn't recommend that

server? I wouldn't recommend that because you know like it's not that secure. Um you know there are people

secure. Um you know there are people like trying to get access to gateways with VPSs and that sort of thing. So I

never recommend using a VPS. The only

VPS that I'd recommend is Malt worker because it's it's powered by Cloudflare, right? And Cloudflare is quite a secure

right? And Cloudflare is quite a secure system to run open core on. So you can use molt worker. I've got a full video tutorial and step-by-step guide right here. So let's run through some examples

here. So let's run through some examples of what I could do personally, right? So

people always ask like what the real use cases for, you know, running open claw, especially when it comes to parallel agents. So here's how you could

agents. So here's how you could structure your channels. This is

something that I actually use or I could use as an agency owner, right? So I have an SEO agency. I could have one channel for SEO content and it could be an agent focused on like content briefs and keyword research. I could have another

keyword research. I could have another for link building. So I could have a channel called link building and then it could be an agent managing our outreach and campaigns. We could have one for the

and campaigns. We could have one for the AI profit boardroom handling community updates and member questions and one for all sorts of like research and that sort of thing. So each channel reads the same

of thing. So each channel reads the same workspace files and has the same tools but stays laser focused on its own topic. So when to use each? Well, if

topic. So when to use each? Well, if

you're using Telegram or WhatsApp, you know, quick back and forth, simple one-off tasks or on mobile is great, but Discord is better for parallel project workflows, team collaboration, anything requiring dedicated persistent context per project. And also, you can use that

per project. And also, you can use that on your phone as well. All right, so the bottom line here is like if you're running OpenC purely through Telegram, you're kind of like putting a ceiling on your own productivity, right? Because

it's like a single channel and Discord turns one agent into a multi- project powerhouse. It's really easy to set up.

powerhouse. It's really easy to set up.

So you set up once, structure your channels by project or skill, and then you've got parallel workflows running like you've seen today directly on Discord. So you can see that here we've

Discord. So you can see that here we've got marketing, video ideas, product ideas. I'm actually going to develop

ideas. I'm actually going to develop this more as we go along. So if you want me to share more stuff about that, happy to. Yeah, happy to help. So glad that

to. Yeah, happy to help. So glad that helps as well. So that's basically it.

Thanks so much for watching. If you want to learn more about AI automation using OpenClor, etc., you can check out my AI automation community. The AI profit

automation community. The AI profit boardroom. It's available link in the

boardroom. It's available link in the comments description or go to apiprofitboard.com. And inside this

apiprofitboard.com. And inside this community, you can ask questions, get help, get support, whatever you need to.

There's always people online. Like, you

see the 67 people right online right now. So, it's very active community. You

now. So, it's very active community. You

can post questions, etc. You can see that it's not just me winning with this stuff. So, there's loads of people

stuff. So, there's loads of people awesome results from our school communities like you can see right here.

So, for example, Russell um landed a client with avatar videos. You've got,

for example, Eric who's on boarding a new client. You've got Dan who landed

new client. You've got Dan who landed his first customer in two weeks. So,

there's some pretty awesome stuff that's happening inside there right now. And if

you want to get access, you you know how to get it. Also, inside the calendar, we do four weekly video coaching calls team uh jump on the calls live in real time, ask questions, share your screen, meet the community, etc. Inside the map, you

can actually zoom into your local area and then DM people in your local area and say, "Hey, do you want to jump on a Zoom call or do you want to meet up in real life, etc." And inside the classroom, we have a full six week AR automation master class that takes you

from beginner to expert. Plus, it shows you how to build your first AI agent in under 5 minutes. If you like OpenClaw, then we actually publish daily updates as you can see right here inside the SOP

update section. And we actually have a

update section. And we actually have a six-hour openclaw course. We have

step-by-step guides on how to set up everything. And then also another 3-hour

everything. And then also another 3-hour course right here that shows you how to run OpenClaw for free and that sort of thing. If you want to get the full guide

thing. If you want to get the full guide on Discord Parallel Agents, we've got it here. Uh Mistral, the new update that

here. Uh Mistral, the new update that just came out. We update this literally every single day and we add like anywhere between like three to four guides each day. We basically just look at all the updates that came out and then we're like, "Right, these are the

most useful for people who watch videos like this." Spartan says, "I've been

like this." Spartan says, "I've been looking for a Discord walkthrough for Open Claw in so long." Well, there you go. And you know, like it is really

go. And you know, like it is really powerful. It's something I I was

powerful. It's something I I was honestly using Telegram before cuz it was pretty simple and easy, but between claw code and between Discord, it it is relatively easy to set this all up. It

only took me about 20 minutes. I mean, I do this every day, so I would give yourself an hour just set up. But it is super powerful and useful and I can just come back to these channels at any time.