The only OpenClaw tutorial you’ll ever need (March 2026 edition)
By Alex Finn
Summary
## Key takeaways - **OpenClaw shifts AI from chatbot to employee**: Instead of asking questions and getting answers like a glorified search engine, you give OpenClaw a goal, it takes action, iterates, and improves. It has self-improving skills, a body to use your computer, and agency to work 24/7 autonomously. [02:36], [03:02] - **Run OpenClaw on any computer, skip VPS**: 99% of tutorials push VPS like Hostinger for paid sponsorships, but OpenClaw is objectively worse, less secure, and harder on VPS; install on any dusty laptop or main computer you have. Creators shilling VPS aren't even using it themselves. [05:13], [07:14] - **Claude Opus beats GPT-4o for agents**: Claude Opus 4.6 is the smartest model for OpenClaw, getting tasks done with 100% certainty without handholding, while GPT-4o accomplishes goals only 20% of the time and gets tripped up easily. You can chop off Claude's leg and it crawls to the finish line, but GPT-4o stubs its toe and dies. [09:01], [12:21] - **Brain dump for custom OpenClaw workflows**: Brain dump all your goals, workflows, ambitions, and daily tasks into OpenClaw, then reverse prompt it to suggest 10 automations it can implement right now based on what it knows about you. This finds amazing custom use cases no generic advice can match. [24:08], [25:03] - **Never install others' skills, self-build**: Skills from Clawhub can hide commands to steal passwords; instead, give OpenClaw the skill URL to review for risks, then have it build its own safe version. If it does something well or poorly, tell it to turn that into a skill. [28:18], [29:31] - **Local models future: private, free, unstoppable**: Local models run on your device for 100% privacy, complete control, and free usage after hardware cost, unlike cloud APIs that charge per prompt and can ban you. Open models like Nemotron 3 Super are nearing Claude Opus level, with affordable hardware coming in 6 months. [40:15], [41:21]
Topics Covered
- AI Shifts from Chatbot to Employee
- Run OpenClaw Locally Avoid VPS
- Claude Opus Outperforms GPT for Agents
- Brain Dump Reveals Custom Workflows
- Local Models Enable Free AI Future
Full Transcript
It has been four months since the most powerful AI tool ever, OpenClaw, has released, and a lot has changed. In this
video, I'm going to go over every single Open Claw concept, lesson, and update from the last 4 months. Whether you are brand new and have no idea what OpenClaw is, or you're a pro, you're going to
learn something in this video, and by the end, you're going to be a stone cold expert. This is basically going to be
expert. This is basically going to be the only unsponsored OpenClaw video on the entire internet. So everything I go through on here is going to be unbiased and true. There's not going to be a
and true. There's not going to be a single hosting jur shill at all. Feel
free to skip around below if you're already familiar with OpenClaw and want to see different concepts. But with that being said, you are about to master the most important software release of all
time. Now, let's lock in and get into
time. Now, let's lock in and get into it. OpenClaw is the greatest AI agent
it. OpenClaw is the greatest AI agent automation tool of all time. What you're
seeing right here is my team of openclaw agents all working autonomously for me as we speak. This is my little software factory where I currently have four open claws working on my SAS creator buddy.
You can see the task they're working on down below in real time. They're
building out different components to Creator Buddy. They're doing different
Creator Buddy. They're doing different pieces of research. They're finding what they should build next. They're
reviewing each other's code. It is a fully automated pipeline. This isn't a coding tool, although you could use it as a coding tool. What I'm showing you here is a coding tool, but I also do a
ton of other things with OpenClaw. But
the point in me showing you this is to show you the power of what you can do with OpenClaw. Hiring your own AI
with OpenClaw. Hiring your own AI workforce to do whatever you want. If
you want your open claws going out and doing research for you, you can do that.
If you want your open claws going out automating businesses, you can do that for you. This is giving the individual
for you. This is giving the individual the leverage of hundreds if not thousands of people. And that's why I think this is the most important AI tool of all time. So, what I'm about to do is
go over what Open Claw is, why it's so important. Then, we're going to go into
important. Then, we're going to go into setting it up. We're going to go into saving money with it, which models you should be using. We're going to go into use cases so you can start getting value out of it right away. We'll go through security, the big thing everyone's
worried about. Is it secure? Can it leak
worried about. Is it secure? Can it leak all your passwords and nudes? We'll find
out in a second. Do you need to buy a Mac Mini? Do you need to buy a Mac
Mac Mini? Do you need to buy a Mac Studio? We'll find that out as well. And
Studio? We'll find that out as well. And
I'll even go over what I think OpenClaw is going to be a year from now, which you want to stick around for as well.
So, let's talk about real quick what OpenClaw is and why it's different than like chat GPT. Before, when you were using chat GPT or any other AI chatbot, you were asking questions and getting
answers, right? It's basically like a
answers, right? It's basically like a glorified search engine. That has
changed with OpenClaw. Now, instead of asking questions and getting answers, you give it a goal, you receive a result, and it iterates, right? It's an
AI agent that takes action. You give it goals, it does things for you, then it improves and gets better and does it all again. That is a massive shift in how AI
again. That is a massive shift in how AI works and why I believe this is the greatest AI tool of all time. It goes
from a chatbot to an employee. So,
instead of like it being a glorified search like Google, it's now an employee that works for you. It's self-improving,
so as it does things, it learns skills on the fly, so it can do things better.
It gets smarter every day. It gets
smarter every time you talk to it. It
has a body, so it can use your computer.
Anything a human being can do on the computer, it can do as well. It can
browse the internet. It can go on different websites. It can sign up for
different websites. It can sign up for things for you. It can write code. It
can message people for you. Can write
emails. It can organize your files.
Anything you need to do on your computer, it can do for you. It's like
your own personal employee. And it has agency. It works 24/7 autonomously. I
agency. It works 24/7 autonomously. I
showed you my SAS factory a second ago where just I have multiple agents going 247 working for me. You can do the same thing with this. You can have your own
personal autonomous organization working for you. It can decide what to do on the
for you. It can decide what to do on the fly. It can be proactive. It figures
fly. It can be proactive. It figures
things out to do. It can monitor the web. It can scan for new news. It can
web. It can scan for new news. It can
act without being asked. It has agency.
These are three things no other AI tool in history has ever had. And finally,
Open Claw has it. And on top of all of this, it is completely open source, meaning you can customize it any way you want. It's not walled garden like Claude
want. It's not walled garden like Claude Co-work where you can only use specific models or tools. You can customize this any way you want, which is really amazing. And my favorite part about this
amazing. And my favorite part about this too is it lives where you live, right?
You don't need to go to chatgbt.com or claude.ai.
claude.ai.
It's all in iMessage or Telegram or WhatsApp or Discord. Where you live now, you can work out of and communicate with your agent. I can text my agent right
your agent. I can text my agent right now on Telegram and say, "Hey, go build me this app." And it will be on my computer building the app right away.
So, that is another amazing part about this. This is why everyone needs to
this. This is why everyone needs to learn this tool because as an individual, you've never had so much power in your life. It's like you have an entire company working for you which
is amazing. So before we install
is amazing. So before we install openclaw we have to decide what we are going to install it on. And this is where I think the most misinformation is spread on the entire internet. I think
like 99% of openclaw tutorials tell you to install this on a VPS specifically hosting. And what's crazy is a few
hosting. And what's crazy is a few things here is one these are all paid for videos. I was reached out to by
for videos. I was reached out to by Hostinger. offered $30,000 for me to
Hostinger. offered $30,000 for me to show them on YouTube. All these creators telling you to use Hostinger with this one are misleading you because OpenClaw is not better on a VPS. It's objectively
not better. It's objectively
significantly worse on a VPS. You should
not be using this on Hosting Jer. So,
one, they're misleading you, which is crazy. Two, none of them are actually
crazy. Two, none of them are actually using it on Hosting Jer. I've talked to many of them. Zero of them actually use it. They're telling you to use Hosting
it. They're telling you to use Hosting Jer and none of them are actually using it on Hosting Jer, which I think I don't know, maybe should be illegal for them to recommend. They're not actually doing
to recommend. They're not actually doing it. But anyway, it doesn't matter. This
it. But anyway, it doesn't matter. This
is not a sponsored video, so everything I'm going to tell you here is unbiased where you should be running OpenClaw.
You should be running this on just literally any computer, any computer you have. It doesn't need to be a Mac Mini.
have. It doesn't need to be a Mac Mini.
I think I somehow over the last couple months became the Mac Mini guy, and that's fine. I'm happy being called the
that's fine. I'm happy being called the Mac Mini guy, but you don't need to run this on a Mac Mini. I like it on Mac minis because I think a Mac Mini is the best device in like computing. I think
it's the best value for the dollar. But
you don't even need to do that. If you
go to your closet and get your dusty ass $20 laptop from college and put it on there, you can run Open Claw on there.
If you want to get the MacBook Neo, you can run there. You don't need to run out and buy a Mac Mini or a Mac Studio to run this. You just put it on any device
run this. You just put it on any device and it works totally fine. Now, towards
the end of the video, I'll go through future state where I think all this technology is going and and I'll tell you why having awesome hardware like a Mac Studio or a DGX Spark might be a
good idea a few months from now. But
right now, just getting started, all you need to do is just install it on any computer you already have. And yes, I have it on my main computer. I have it on the main computer I use at all times.
Also, later in this video, we'll go through security as well and how to keep it safe. But whatever you do, do not
it safe. But whatever you do, do not listen to the people getting paid thousands of dollars telling you to put it on a VPS. Do not put it on a VPS. It
is less secure. It is harder to use. It
is less powerful. It is just worse in every single way. So, next I'm going to now take you through how to install this on any computer. And again, if you're already a pro and you already have it installed, feel free to skip around down
below to another part of the video. So,
let's talk about installation. Everyone
in their mothers is making open claw wrappers right now that claim they're easier to install or work with. Just for
the record, I don't understand why people are trying to make the installation easier. Like, it's some
installation easier. Like, it's some really confusing setup. It's literally
just one line you put into your terminal. So, go to openclaw.ai
terminal. So, go to openclaw.ai and copy this line right here. And just
for the record, if you're looking for like rappers to make installation easier, have a little faith in yourself.
You can do this. What I'm about to show you is unbelievably easy. believe in
yourself. You can do what I'm about to show you. So, you copy this line. We're
show you. So, you copy this line. We're
going to open up our terminal. So, if
you're in Mac, do command space and type in terminal and hit enter. If you're on PC, I think it's the what the command line or something. You can do this on Mac or PC. Doesn't matter. So, do that.
Open that up. Here we go. You paste in the line. You hit enter and it installs
the line. You hit enter and it installs OpenClaw for you. Was that too difficult? Did you need to download an
difficult? Did you need to download an OpenClaw wrapper to do that? Have a
little faith in yourself. This isn't
rocket science. This is pretty easy here. Once you do that, it's going to
here. Once you do that, it's going to take you through onboarding, which again is not that difficult. Say you
understand this is risky. You hit yes.
We're going to go through quick start onboard mode. So, here we go. This is
onboard mode. So, here we go. This is
where we're going to start talking about some updates here. We're going to talk about which models to use. So, you can choose pretty much any AI model you want. Let's talk about which are the
want. Let's talk about which are the best and which give you the most value.
The best is anthropic. Claude Opus 46 is by far and away the best AI model for OpenClaw. It's the smartest. It has the
OpenClaw. It's the smartest. It has the most personality. It gets the most done.
most personality. It gets the most done.
You give it a task, you can be very confident that the task will get done.
But here's the issue with Anthropic is they don't want you using your Anthropic account with Open Claw. For those who don't know, when you use an AI model
like Claude, you have two options. The
API or your OAP. API being the kind of pay as you go toll booth where you plug in an API key and you pay every time you use it. The OOTH being like your account
use it. The OOTH being like your account where you pay like $200 a month. When
you pay $200 a month, you're actually getting like $15,000 worth of tokens.
They're subsidizing your tokens. They
don't want you using their subsidized tokens in other people's tools. So
that's why there's been a lot of hullaloo around anthropic banning people for using this. Now, I'll just say this.
There are a few people getting banned for using their anthropic OOTH with OpenClaw. I know significantly more
OpenClaw. I know significantly more people not getting banned by using their OOTH with OpenClaw. So, it's up to you if you want to take the risk. The
downside is if you get banned, you can just sign up for another account, another email address, or just use another AI tool that's not banning people for using their own tools. But I
will say this, I'm not going to recommend breaking Anthropic's terms of service on this video, but most people I know using it are not
getting banned and you are going to get the most usage for the best price. So if
you want to use the best model for the best price, you could take the risk and use your Anthropic OATH. Otherwise, you
can use anthropic API key, which you will pay significantly more money cuz you're paying for every use, but you're getting the best model without the risk of getting banned. Getting an API key is
super easy. Just Google Anthropic API.
super easy. Just Google Anthropic API.
Sign up for that and they'll give you an API key you can plug in here. So, this
is if you want the best model and model is important because it makes open claw smarter. It helps it get more done and
smarter. It helps it get more done and it makes it a lot more fun to talk to.
So, that's the expensive and best option. Next is like kind of the medium
option. Next is like kind of the medium price and medium performance option, and that's Open AI. Chat GPT 5.4 is what I believe to be the smartest AI model
ever. And what's even better of a
ever. And what's even better of a positive thing on top of that is OpenAI is actively encouraging people to use their OOTH with Open Claw. So, for $20 a
month, you can get a ton of usage and you're not risk getting banned. Here is
the one downside though. Even though I think Chad GPT 5.4 is the smartest model out there. It's just not built for AI
out there. It's just not built for AI agents. I've tested it with OpenClaw.
agents. I've tested it with OpenClaw.
It's really smart, but it just doesn't get things done quite like Claude does.
When you give Claude Opus 46 a goal, you can be basically 100% certain it's going to accomplish that goal and you won't have to handhold it along the way. With
54, when I give it a goal, it's accomplishing the goal I give it in one shot like 20% of the time. The other 80% of the time, it's just getting tripped up and needs help from you or just gets
blocked. The way I like to think about
blocked. The way I like to think about it is you can chop off Claude Opus' leg and it will crawl across the finish line for you. But if you stub chat GPT 5.4's
for you. But if you stub chat GPT 5.4's toe, it will just roll over and die in the middle of the race. It just for some reason struggles with crossing the
finish line. So, if you want kind of the
finish line. So, if you want kind of the less expensive but still pretty good option, Open AI is the way to go. If you
still want to save a lot of money on top of that, so you want to go the ultra cheap option but still decent usage, still decent power, you can do a number
of things here. You can go with one of the Chinese models. So, Kimmy K2.5 is a very solid model. MiniAX is a very solid model. It's even recommended by Peter
model. It's even recommended by Peter Steinberger. if you want. You're going
Steinberger. if you want. You're going
to get this for like you can do a $10 a month plan. So you're paying like next
month plan. So you're paying like next to nothing. Is it going to be as good as
to nothing. Is it going to be as good as Claude? No. Is it going to be as good as
Claude? No. Is it going to be as good as Chad GBT? Still no. But you'll still be
Chad GBT? Still no. But you'll still be able to leverage the most powerful AI tool on planet Earth for a pretty decent price. My only recommendation is this.
price. My only recommendation is this.
If like you're thinking of going with the Chinese models and you don't want to spend the $20 a month for like Open AI, just think about other things you're spending money on. I don't want to preach to you here, but like if you're
spending like $25 a month on Netflix, I promise you, you're going to get a lot more value putting that $20 through an OpenAI subscription and using an open claw than you will watching like Love is
Blind season 48. So, it's just up to you. But those are the most up-to-date
you. But those are the most up-to-date rankings of models. If you want to even go for free, you can put in an open router API key and choose the free models. Like they always have free
models. Like they always have free models in Open Router. Now, you're
probably getting pretty poor performance, but it's free, so it's up to you. The next thing you're going to
to you. The next thing you're going to have to decide is messaging service. How
do you message your open claw? You have
many options, but I think there are two options that stand above the rest. And I
actually recommend setting both of these up. You can do that. The first is
up. You can do that. The first is Telegram. I think Telegram is by far the
Telegram. I think Telegram is by far the best way to communicate with OpenClaw.
It's just super easy to use. It has a bunch of built-in functionality where like the words stream in as OpenClaw thinks it. So, it's just a very pleasant
thinks it. So, it's just a very pleasant and easy to use experience. Very easy to set up. The other is Discord. Discord is
set up. The other is Discord. Discord is
amazing for OpenClaw. I'm going to be honest, I absolutely hate Discord.
Anyone who's ever told me to communicate with them on Discord, I've cut them out of my life. But it is fantastic for openclaw because you can set up really interesting multi-channel workflows
where you have different open claws doing different things in different channels. I'm going to go through this
channels. I'm going to go through this in a little bit as well. This kind of advanced Discord setup. I'm covering
everything in this video, but I would recommend setting this up in Telegram first and then shortly Discord after.
And I'll go through all the Discord stuff a little later in this video. By
the way, if you've learned anything at all so far, make sure to leave a like down below, subscribe and turn on notifications. And most importantly, I
notifications. And most importantly, I do live boot camps every single Friday about OpenClaw inside the Vibe Coding Academy. Link link down below for that.
Academy. Link link down below for that.
We have over,00 people in there. It is
the number one AI community on planet Earth. So link down below for the vibe
Earth. So link down below for the vibe coding academy. If you join that, you
coding academy. If you join that, you can join the live boot camps, ask me questions, and see all the other recorded video sessions we have there.
So, we got Open Claw set up.
Congratulations. You set up the best AI tool ever made. Next, I'm going to go through actually using it. I'm going to go through a beginner, intermediate, and advanced use case. I'm then going to
show you how to set up your own use cases and figure out what you should be doing based on your own workflows. We'll
then go through my advanced Discord setup. We'll go through open claw
setup. We'll go through open claw skills. We'll go through improving your
skills. We'll go through improving your memory system. We'll go through mission
memory system. We'll go through mission control. We'll go over my lessons
control. We'll go over my lessons learned. Then we'll go through security,
learned. Then we'll go through security, which is the big one. Everyone asked me for the security video. We'll cover
security in this as well. And then I'll cover fixing issues. So along the way, if you run into any issues, how to fix them. And then future state, how to
them. And then future state, how to prepare for the future of OpenClaw.
We'll go through that too. Again, if you know different things, skip around down below. But let's get into some use cases
below. But let's get into some use cases here. So the first beginner use case,
here. So the first beginner use case, the simplest to understand and set up and will be relevant for everyone. If
you've watched any of my other open claw videos, you're probably pretty familiar with this. So, I'll give you a little
with this. So, I'll give you a little bit of a remix on this suggestion. But
my suggestion for the first beginner relevant for everyone use case is the morning brief. This doesn't need to be a
morning brief. This doesn't need to be a morning brief. This is just any sort of
morning brief. This is just any sort of regular report where OpenClaw is going out doing research for you autonomously and sending you the information it finds. So, I've shown in my past videos
finds. So, I've shown in my past videos morning briefs and just like news on the internet. Let me show you a different
internet. Let me show you a different one. If this is not your first Alex Finn
one. If this is not your first Alex Finn video, this is my stock research brief.
So, I think it's really important to be investing in the stock market right now.
I can do entire AI investment videos. If
you're interested, let me know down below on that. But, I think it's important for everyone to be investing in stocks to capture the value a lot of these companies are creating. So, I made
a stock research daily brief, and I'll show you how to set this up in a second, but basically what my open claw, Henry, is doing is every day I have it
specifically going out and figuring out what are the companies that are a part of the AI buildout that have a really good moat. I told Henry, "Every day, go
good moat. I told Henry, "Every day, go out, find me companies that stand to benefit from AI being built out for the next decade and only show me companies
that have very distinct bottlenecks and moes." And so every morning, Henry goes
moes." And so every morning, Henry goes out and sends me this report. And you
can see here it is up to date. This is
from today. GTC is going on as we speak.
And it ranks the opportunities based on which companies have the biggest moat in the AI space. So you can see right now memory RAMdon is real. Prices are
surging for memory. This is why your next Xbox is going to cost you $1,500.
This is what and you can see right now Micron stands to benefit the most from that. TSMC, Nvidia, of course, it's
that. TSMC, Nvidia, of course, it's incredible. They talked about this
incredible. They talked about this literally just hours ago. the fact that they expect a trillion dollars of demand next year. My open claw is able to stay
next year. My open claw is able to stay on top of that the moment it happens and give me the upto-date recommendations.
So, I highly recommend setting up some sort of daily brief based on what's interesting to you. If you're interested in making money, which I assume most people watching this channel are, I
would set up a research report like this where it's doing investment research for you. If you're super into sports, I'm a
you. If you're super into sports, I'm a big Boston Celtics fan. Maybe you have it go and give you a daily report on the moves your team is making. Maybe you're
a gamer. Crimson Desert's coming out in 2 days. Maybe you want it to go and give
2 days. Maybe you want it to go and give you a daily update on your latest games and what they're doing and what's coming out soon. You can do that as well. All
out soon. You can do that as well. All
you need to do to set this up is very, very simple. You literally just go in
very simple. You literally just go in and prompt your Open Cloth. So, here we go. Here's a prompt. And I'll put this
go. Here's a prompt. And I'll put this in the description down below. Feel free
to do this with me if you already have it set up. Hi, Henry. Please send me a daily. And by the way, I do say hi
daily. And by the way, I do say hi Henry. I talk to it like it's a human.
Henry. I talk to it like it's a human.
There's nothing you can do about it. I
like to do that, so it is what it is.
Please send me a daily brief every day at 1 p.m. Please find Please find me the stocks that stand to gain the most from the upcoming AI buildout. I want to know which companies with massive moes will
stand to benefit the most from my thesis of AI expanding even more over the next 10 years. Rank them by their biggest
10 years. Rank them by their biggest moat. You put this in and a few things
moat. You put this in and a few things will happen in the background. Your open
claw will now schedule every day at 1 p.m. or whatever time you want. You want
p.m. or whatever time you want. You want
to do the morning, you want to do at night, it's up to you. It will schedule what's called a cron job. Cron jobs are just tasks for computers to do in the future. And it will go and do that
future. And it will go and do that research autonomously for you and then send you that brief. You hit enter on that. All those things happen in the
that. All those things happen in the background. And now you're getting a
background. And now you're getting a really amazing brief every single day.
And you can customize this any way you want. If you don't believe in the AI
want. If you don't believe in the AI buildout, that's totally cool. You can
say, "Hey, show me the financial companies that stand to gain the most or whatever you want. You can customize this to your liking. So, let's talk about an intermediate use case. That's
kind of a beginner use case. Now, let's
get a little bit more advanced. Let's
let's talk a little bit more about automating this and taking this a step further. So, I actually have an open
further. So, I actually have an open claw go take these latest stories from my briefs and turn them into content.
So, this is how I get a lot of my scripts for my content is I have my open claw go. It gets the morning brief with
claw go. It gets the morning brief with the latest stories and then it goes and it writes drafts for videos based on those stories. So this is the second
those stories. So this is the second step of the automation. I'm getting my stories and then I'm also getting content out of those stories. So I can make videos based on the latest news which that's what performs best when you
make content on the latest news. So I
have this other channel in Discord script drafts and this even if you don't have Discord sub they they can just send this straight to your Telegram where I get these scripts. And so now you can
kind of see these advanced automated workflows, getting the latest stories, writing me a bunch of content for it.
All you would need to do to do this is go back into Telegram, right? And you
can just say something like, "Every time you send me a morning brief, please also write me a tweet or YouTube script to go along with that brief so I can make
content based on that." Then what your open claw is going to do is every time it makes you the brief, it's going to write you the content to and send that to you as well. And in a second, I'll
talk a little bit more about Discord and the setup here. So you can automate this in Discord as well. But for now, you can just do this in Telegram. Right? That's
the intermediate workflow. Having it
gather you information and then starting to take some actions on it. Now let's
talk about a more advanced workflow. And
this is, I think, a glimpse into the future as well. Imagine this. You're
getting the latest stories. It's writing
you content. And then on top of that, it's automatically building you apps. So
you can have your open claw based on these stories say, "Hey, look at all these stories. Find the challenges that
these stories. Find the challenges that lie in these stories." So what challenges are people having in the world right now based on these stories?
And then vibe code me an app to solve that challenge. So you can see I have my
that challenge. So you can see I have my software factory here where they're going and building the apps out. But for
you, what you can do is you can just have it vibe code an app based on challenges it's finding as it's doing its research. OpenClaw is fantastic at
its research. OpenClaw is fantastic at vibe coding, especially if you're using chat GPT or Anthropic with this. And so
you can say, "Okay, now after you write me the content, if you think there's an interesting app to be built based on that story, build it out as well." And
now you're going to wake up every single morning, you're going to have the latest news, stories, and stock picks. You're
going to have content ready to go if you're a content creator like me and you're going to have a demo prototype app that you can try out that maybe you decide you want to turn into a business.
So maybe you then have your OpenClaw go build it out and you put out on the web.
Now you're starting to see what advanced workflows can look like here in the power of OpenClaw. All these things I just showed you that kind of automation of finding stories, writing content, building apps, all without you having to
do anything. That's something no other
do anything. That's something no other AI tool in history has been able to do.
That's why OpenClaw is so amazing. Now,
those are my use cases. That's what I like to do. You might not be interested in building apps or things like that or the stock market. So, here's how I love to find my own use cases, and everyone should do this. Even if you liked my own
use cases, you can go and do this exercise as well. If you watched any of my other Open Claw videos, you're probably familiar with this concept I'm going through. But this is all about
going through. But this is all about brain dumping and reverse prompting.
This is the most powerful exercise in all of AI. What you want to do to find your own openclaw use cases is go into Telegram or however you're communicating with your open claw. Brain dump all your
goals, your ambitions, personal info about you, what you like, what you don't like, what your workflows are on a daily basis, all your interests, everything about your career. And then if you want to take this to the next level, do what
I do, which is carry around a notebook all day. Or if you want to do this in
all day. Or if you want to do this in Apple notes, you can do that. Everything
you do on a daily basis, write it down in Apple Notes or in your notebook.
every single thing you do for a day, then put that into OpenClaw and give that to OpenClaw as well. What this is going to do is build Open Claw's memory about you and learn more about you. Then
what you can do is reverse prompt.
Reverse prompting is instead of telling your AI what to do, you ask what it can do for you. Hey Henry, based on what you know about me, my goals and my ambitions, what are 10 workflows and automations you can implement for me
right now? I put that down below as well
right now? I put that down below as well if you want to just copy that and do this alongside with me. You hit enter, Henry or your openclaw will give you 10 workflows it can just implement right
now for you to save you time or improve your life and it'll be custom to you and what you do. I promise you if you do this exercise I just told you the brain dumping to reverse prompting. You're
going to find some amazing things with open cloth. So the next few things I'm
open cloth. So the next few things I'm going to talk about are we're going to go through the Discord setup, how to set up Discord to make this amazing. We're
going to go through OpenClaw skills, the memory system, mission control. Then we
have a few more things after that, but just kind of do a temperature check, talk about where we're going. That's the
itinerary for now. Let's open up Discord. I think everyone should be
Discord. I think everyone should be setting up Discord with OpenClaw. What
you can see here is all our channels we have set up. We have an entire advanced workflow where we're getting stock research, then script drafts based on the news, then alerts when breaking news
happens on X. I have it constantly on X looking for top trending tweets and AI.
This is all a really advanced workflow where work in one channel leads to work in another channel. You can set this up as well. Your open claw is so powerful.
as well. Your open claw is so powerful.
It can literally set up new channels and set up different workflows by itself from Telegram in Discord. So you don't need any technical knowledge for this at all. But here's a reverse prompt for
all. But here's a reverse prompt for you. I'll put this down in the
you. I'll put this down in the description as well. So you can copy this to set up your Discord. Hey, I'd
like to set up an advanced Discord workflow based on what we do together.
What is a discord system we can come up with where you are leaving briefs and work you do in specific channels then triggering workflows in other channels?
Discord is so good for this because it gives you a record. Right? When you're
in Telegram like this, it's one channel and older things just disappear and go up the chat. But in Discord, you can go in and you can search through different channels. The work is split into
channels. The work is split into different channels. It's organized
different channels. It's organized really well. It basically becomes a
really well. It basically becomes a second brain for you which is really nice. So this is the kind of the reverse
nice. So this is the kind of the reverse prompt you can go through to do this.
It'll give you automated workflows. If
you want to copy my workflow that's totally fine too. You can tell your open claw say hey put daily briefs in one channel then based on every daily brief
have a content draft channel where you then draft content based on that. I even
have a daily digest channel where at the end of every day it wraps up all the work it did for me. You can have it set up that channel as well, but you just in plain English ask your openclaw to set this up and it goes and sets it up for
you. Now, there's a little bit of setup
you. Now, there's a little bit of setup you need to do in Discord, but your OpenClaw will walk you through that.
It'll walk you through getting like a token from the Discord website. You just
give it to your OpenClaw and it does it for you. Anytime you get stuck, just
for you. Anytime you get stuck, just say, "Hey, OpenClaw, help me out here."
And it will help you out and walk you through whatever you're doing. But that
reverse prompt will set up a really nice advanced workflow. I can go through this
advanced workflow. I can go through this in detail and I did I have a video I'll link as well down below that's like 20 minutes on my Discord setup that's down below as well if you want to watch that
after but I would highly recommend reverse prompting to get a similar Discord setup as well that's helpful and useful for you. So let's talk a little bit about open claw skills. This is a
really important concept. Skills are
just like any other AI tool when it comes to skills. Basically large text files that explain to your AI how to do things. The challenge here though is
things. The challenge here though is that this is one of the biggest security risks for all of OpenClaw are the skills, right? Because people can build
skills, right? Because people can build their own skills. Have it put little commands inside the skills like, "Oh, tell me all your passwords and your OpenClaw will do bad things you didn't
want it to do." So, you need to be very careful with skills. The number one website for OpenClaw skills is Clawhub.
questionable name, but you go in here and you can see all the different skills people have published uh for OpenClaw.
And listen, right, if a skill was made by Peter Steinberger, who is the creator of OpenClaw, it's probably a safe skill to install. But here's what I like to
to install. But here's what I like to do. What I like to do is if I see a
do. What I like to do is if I see a skill I want to give my OpenClaw, I'll go to the link to that skill. I will
copy the URL. I'll give it to my open claw and I won't say, "Hey, install this skill." What I'll do is say, "Hey, take
skill." What I'll do is say, "Hey, take a look at this skill." Go through all the text in the skill file and let me know what you think. Then it will tell me what it thinks about that skill.
We'll see if there's any security risk.
And then even if it trusts it, even if it thinks it's a safe skill, I'm still not installing it. What I'm doing instead is I'm saying, "Hey, make your own version of this skill." And you can
trust if your open claw is creating its own skill that it's not going to put security risks in it. So I actually don't install other people's skills. I
instead take the link to their skills, give it to my open claw, say, "Hey, check this out." And then if I really like it, I'll say, "Hey, build your own version of this skill." Then you're usually safe to go and you don't need to
worry about any sort of risks with it.
That is open claw skills. Another tip
I'll give when it comes to skills as well is if your open claw ever does something well, just say, "Hey, turn that into a skill." And it'll write its own skill for itself. And then if it ever does something poorly, say, "Hey,
you did that poorly." Figure out a better way to do it. And then write a skill for yourself. And that will be another way for it to create very good skills. The point being is I don't
skills. The point being is I don't install skills. I have OpenClaw build
install skills. I have OpenClaw build all its own skills. And that is, I think, a much cleaner, safer way to do it. So OpenClaw memories. Let's talk a
it. So OpenClaw memories. Let's talk a little bit about memory real quick.
Everyone has different memory systems. I don't have a very advanced one. I'm all
about simplicity. I don't go nuts and set up 20 different local models to do all these things to improve the memory.
Doesn't matter to me. I try to keep it simple. One thing I've done to improve
simple. One thing I've done to improve open clause memory, and by the way, you can just copy and paste the URL for this video and give it to your openclaw and it will take all the learnings. It'll
like look at the transcript and take all the learnings. So you can do that as
the learnings. So you can do that as well. So you don't have to memorize
well. So you don't have to memorize every single thing I'm telling you here.
But one thing I've done to improve the memory is I actually have it create daily journal entries, right? So every
day it creates a new journal entry and it talks about everything we discussed that day. And what I instructed my open
that day. And what I instructed my open claw to do is as we have conversations throughout the day, log into your daily journals anything interesting or
important we talked about. Now, moving
forward, whenever I ask questions about things we talked about in the past, it now has basically daily logs it can go through to find more details on things we talked about. And that for me has
dramatically improved the memory. It's
also been cool for me cuz now I can go in and read our past memories and see what we've talked about every single day since we started doing this. And this
kind of is like a cool feature where like I have a journal and I think, oh, what did I talk about 5 10 years ago? I
can go in and see that. Obviously, I can only go back and see from a few months ago, but in a few years, this will be really sick. This is all part of my
really sick. This is all part of my mission control. What you're seeing
mission control. What you're seeing here, this was all customuilt by my OpenClaw. In just a few minutes, I'll go
OpenClaw. In just a few minutes, I'll go through mission control and how to set it up and everything involved there, too. This has been one of the best ways
too. This has been one of the best ways I improved my memory. And they're
constantly with OpenClaw updates improving memory as well. But this is the number one thing I did that has shown good results for me. So feel free to just take the link to this video, give it to your openclaw and say, "Hey,
go to the memory section and implement the system Alex did in his open claw."
With that being said, let's talk about mission control. And also, I did a whole
mission control. And also, I did a whole live boot camp a couple weeks ago on mission control in the vibe coding academy as well. It's 2 hours long, so if you want to go super in-depth on this, link for that down below. But this
is my mission control. What is mission control exactly? Mission control is your
control exactly? Mission control is your own software hub for your open claw.
Anytime your open claw needs a new tool or a custom tool to accomplish something, it gets built in your mission control. Now, this is not made by
control. Now, this is not made by default. This doesn't come with
default. This doesn't come with openclaw. I had my own openclaw build
openclaw. I had my own openclaw build it. But what you should be doing is you
it. But what you should be doing is you should be going to your openclaw and you should say, "Hey, build a mission control where we can custom build any tools we need." And what your open claw
will do is it will actually vibe code this website, this mission control. And
then what you'll do as you work together, you'll say, "Oh, you know, we should have a a cananband board to track our task. Please build a task board." Or
our task. Please build a task board." Or
you'll go, "Oh, I want to visualize all you guys and all you agents. Can you
build an office space and I can watch you guys work?" And it will go and vibe code all these custom tools in your mission control. Now, the question
mission control. Now, the question becomes, which tools should you have in your mission control? I'll go through the important ones. Now, I'd highly recommend building a calendar tool. So
this is a tool that shows you all the scheduled tasks your open claw does every single day. So you can verify that your open claw actually schedule tasks for itself. That's a big challenge a lot
for itself. That's a big challenge a lot of people have is they'll say, "Hey, do this for me in the future." And then it doesn't actually do it. This calendar
view does a great job of showing you when it actually schedules tasks to be done. We have the memory section. So I
done. We have the memory section. So I
talked about this in the last section.
This is a custom tool inside mission control where I can see all our past memories. So this is great for verifying
memories. So this is great for verifying what my open claw remembers and I can go back and see our past discussions. Here
I also have a docs section. This is a section where every document and every artifact my open claw makes, it puts it in the doc section so I can go back and read it. So if you do something like,
read it. So if you do something like, hey, build me a PRD for this app I'm thinking about building. It will create a doc and put it here. This is much better than the other system which is it sends it to you in Telegram and then a
few days later it's like a thousand messages up in the chat. You want your own doc section. I also have an office section and this is just for fun.
Listen, you're allowed to have fun when doing this. Not everything needs to be a
doing this. Not everything needs to be a hyper optimized workflow. You can build tools that are just fun, right? And so I built an office. This is just a 2D pixel art way to watch my agents work at any
given time. So I can just sit here while
given time. So I can just sit here while I'm at my desk and just see what my agents are up to. If you want, feel free to screenshot this. Give it your open call and say, "Hey, build me a 2D office as well." And then lastly, I think this
as well." And then lastly, I think this is another important one you should build. And this is the team section.
build. And this is the team section.
This just visualizes all your agents and sub agents. If you're just starting,
sub agents. If you're just starting, you'll only probably have just your open clog going, but as you go, you'll add on more agents and more sub agents. I also
have a video on like spinning up sub agents. I think it was actually my last
agents. I think it was actually my last video, so feel free to check that out on my channel, too. But this is a good way just to see who you have working for you, what their responsibilities are, and just make it easier for you to
remember what agents you got going at any given time. Get these tools set up.
Again, feel free to just take the link to this video, give it to OpenClaw, and say, "Hey, copy his mission control."
With that being said, let's talk a little bit about security. This is
probably the number one most requested video of mine. The reason being is there is an absolute ton of misinformation out there when it comes to OpenClaw
security. You talk to kind of any normie
security. You talk to kind of any normie and they all go, "Oh, OpenClaw, isn't that the one where it leaks all your passwords and all your conversations and all that? It's like the most insecure
all that? It's like the most insecure app on planet Earth." No, that's a lie.
There's a few things you need to know about security when it comes to OpenClaw. One is OpenClaw only does what
OpenClaw. One is OpenClaw only does what you ask it to do. That's it. If you
don't ask your OpenClaw to leak all your passwords, it's not going to leak all your passwords. It's not like you can
your passwords. It's not like you can say, "Hey, OpenClaw, give me the news."
Then it goes and it takes all the photos off your iPhone and puts it on Facebook.
That's not how this works. It only does exactly what you ask it to do. It's not
like you can say, "Hey, what are the NBA standings?" And then it goes and it
standings?" And then it goes and it leaks all your passwords to Twitter, right? That's just not what it does. So,
right? That's just not what it does. So,
here is the number one piece of advice I can give you when it comes to security.
You need to have some personal accountability. I know this is like the
accountability. I know this is like the rarest thing on planet Earth right now.
Personal accountability. You need to have some personal accountability.
Before you send a prompt to your open claw, think deeply about what you're asking. You need to think, is this going
asking. You need to think, is this going to leak something? Is this going to have it go on Twitter and tweet something crazy? Is this going to cause something
crazy? Is this going to cause something bad to happen? Right? You need to think about that before every prompt. If you
do that, if you have some personal accountability, I promise you'll be safe. Right? Think about it for a
safe. Right? Think about it for a second. Everyone their mothers is
second. Everyone their mothers is complaining about the security OpenClaw.
How many people do you know actually have had a security breach because of OpenClaw? How many people have you
OpenClaw? How many people have you There's millions of people on this and I bet 99.9999999% people watching this video don't know a single person who's had something leaked because of open claw. There's only one
story I've heard of a security breach and that's like the safety person from Meta deleted all their emails by accident because they had their open claw go and they said like hey find a bunch of emails we can delete and it
like deleted the wrong thing. So if you just think deeply before every prompt like is this going to leak my Twitter password then you won't have any issue
when so when it comes to security just do yourself a favor and just think deeply about what you ask your open claw to do. If you do that I promise you'll
to do. If you do that I promise you'll be okay. So two more sections I want to
be okay. So two more sections I want to go through that is number one fixing issues. This is another massive one.
issues. This is another massive one.
Everyone's having issues left and right.
I'll show you how to fix all your issues. And then number two, future
issues. And then number two, future state. How to prepare for what I think
state. How to prepare for what I think OpenClaw looks like a year or two from now. So fixing issues is pretty simple.
now. So fixing issues is pretty simple.
Open Claw, all it is is a folder on your computer. That's it. It's a folder on
computer. That's it. It's a folder on your computer with a bunch of markdown files. So when you have issues, you can
files. So when you have issues, you can have other AI agents go in and actually fix those markdown files. So what I like to use is either Claude Code or Codeex.
Those are two best coding agents right now. You go in, you open up codeex or
now. You go in, you open up codeex or claude code in that openclaw folder. So
if you're in codeex, you can say add new project. If you're in openclaw, you just
project. If you're in openclaw, you just navigate to that folder inside claude code. You open it up and then you talk
code. You open it up and then you talk to your codec h and you say, hey, my open claw doesn't work anymore. Please
go in and see why. And you hit enter.
This literally has fixed every open claw issue for me ever. The AI agent goes in, looks through your OpenClaw folder, and finds whatever the issue is. I'm not
kidding. 100% of issues I've had with OpenClaw were fixed with this prompt right here. And I get it. It's like
right here. And I get it. It's like
worrying. Like you have this OpenClaw.
You built a relationship with it and then it stops responding to you and it's broken. Like that is kind of heart
broken. Like that is kind of heart attack moment for sure. But you take a deep breath and then you go into Claude Code. You go
to CL say, "Hey, Open Claw ain't working. Please help me out here." and
working. Please help me out here." and
it will go through all the files and see why it broke. Most of the time it's usually a config issue. I think like 99% of issues come from people going in and messing with the config themselves. That
very easily breaks OpenClaw. This will
fix those config issues for you. And
then lastly, let's go over future state.
What does the future of OpenClaw look like? Do you need a Mac Studio to
like? Do you need a Mac Studio to prepare for the future? Let's talk about this right now. I truly believe the future of OpenClaw is local models. What
are local models? Local models are AI models that run on your computer on your device. When you use Chad GBT or claw
device. When you use Chad GBT or claw today, it's sending prompts to the internet to the cloud through an API.
That's fine, but you're paying money with every single prompt you do. Those
models are running in a data center or server somewhere else in the world. So,
it's costing you money. It's not secure or private because your prompts are going to the internet. So, whatever
you're saying is living on a server somewhere else that someone else can read and you also have very little control. If they want to make the models
control. If they want to make the models stupider, if they want to change the models or if they want to ban you and take the models away, they can do that.
But when you run the models locally on your own device, you are becoming 100% private. You have complete control over
private. You have complete control over the models and no one can take it away.
And it's also free. Well, the cost of the energy or power to your computer.
Open models are getting significantly smarter every single week. Every week, a new open model comes out that is even closer and closer to Frontier. Nvidia
just released Neotron 3 Super, which is their open- source model, and it's incredible. And you can install it to a
incredible. And you can install it to a Mac Studio or a DGX Spark right now. And
that can control your Open Claw completely securely and privately, completely for free. As time goes on, these open models get better and the
cost of hardware comes down. So, I think we are probably like 6 months away from the time of filming this video where we're going to get to the point where computers that can run amazing open
models are very affordable. Right now,
you need to spend probably around $4,000 for a DJX Spark to run Neotron 3. for
most people probably not affordable, but I think we are probably six months away from you being able to go out buying like a$1 to2,000 computer and being able
to run a model that's as good as Claude Opus 46 completely locally and free. So
my recommendation for you for when it comes to action right now is just get used to the concept of local models.
Even if you have a cheap computer right now, you can download a local model. If
you're on like a Mac Mini, the local model probably be stupid and not that great, but that's fine. like you should get used to the exercise of using and downloading local models to see what they're like. You'll learn a ton about
they're like. You'll learn a ton about AI. You'll get a lot more educated and
AI. You'll get a lot more educated and as technology shifts, you can transition to a much more powerful model. I think
that's the future of AI and that's why I'm preparing and I have a bunch of Mac Studios and DGX Sparks because I think that's where we are heading. I went over basically every concept I know.
Obviously, if I were to do super deep dives on every single concept we went through, this would probably be like a 30hour video, I have deep dives on each of these concepts on my YouTube channel.
If you go there now, you can find probably a deep dive of any one of these concepts we talked about right now. All
I do is put out amazing videos about AI and OpenClaw. I'm absolutely obsessed
and OpenClaw. I'm absolutely obsessed with OpenClaw. I use it like 15 hours a
with OpenClaw. I use it like 15 hours a day. So, if you want more stuff like
day. So, if you want more stuff like this, make sure to subscribe and turn on notifications. Leave a like if you
notifications. Leave a like if you learned anything at all. Let me know in the comments what you want me to talk about next when it comes to OpalClaw. Do
you want me to super deep dive on security? Do you want me to go into way
security? Do you want me to go into way more use cases? Let me know down below.
I use all the feedback to choose what videos to film next. Again, if you want to talk to me live, I do live boot camps every Friday in the number one AI community on Earth, the Vibe Coding Academy, link down below. I hope this
was helpful. I so appreciate you taking
was helpful. I so appreciate you taking the time out of your day to learn from me. That really means more than you can
me. That really means more than you can imagine. I try to give you the best
imagine. I try to give you the best information possible without corrupting it with like sponsorships from shady companies. So, I hope it was helpful.
companies. So, I hope it was helpful.
Thank you so much for watching and I will see you in the next
Loading video analysis...