We are talking open claw today. What can

people expect if they watch to the end

of this conversation? [music]

>> We'll cover the basics. What's openclaw?

The prerequisites.

>> Do you think the comparison [music] of

openclaw to Jarvis from Iron Man is an

apt comparison?

>> We'll find out when we look at the names

of my Hner servers.

>> I feel [clears throat] like a proud

parent right now. And how profound is

OpenClaw in terms of what it offers to

people? The thing with OpenClaw that

made it catch fire was

CJ Hes. Welcome to the show.

>> Alex Lieberman, thank you for having me.

>> Okay, so we are talking Open Claw today.

What can people expect if they watch to

the end of this conversation?

>> Yeah. So, we've prepped kind of a guide

here that we're going to walk through

today. Um, we'll cover the basics,

what's openclaw, the prerequisites. Then

really, we want to go zero to one with

having an openclaw setup without

spending a ton of money on a Mac Mini or

wasting a bunch of time trying to do it

through some provider. This is going to

be the bare metal guide to do it safely

and securely on a VPS through Hzner for

as little as five bucks a month. Um then

we're going to get an open cloth set up

and we're going to start to play around

with it. And then hopefully at the end

um we'll have time to show a couple of

cool use cases that I've personally done

as well as some that I've seen online

and we can kind of discuss how you would

go about setting up things like that. So

prerex basically what we're going to

need VPS provider cloud but there's a

ton of providers out there. I personally

like I've used it forever. They're a

European company. So when you get

invoiced, they send you an email that

says, "Stir CJ, your bill is here." Um,

then we're going to set up this private

network that's going to be tail scale.

Um, this essentially gives you a way to

connect to each of the machines, but

without having to use the open internet

as a way to connect. Essentially makes a

walled garden of the computers that you

want to be able to connect to. We're

going to set up messaging. OpenCloud is

great because it has a ton of messaging

providers. I've been dabbling with

trying to actually transfer over to

Discord, but right now Telegram's my

main driver for my agents, so we're

going to set it up that way. It's also

the fastest and easiest. And then the

last thing is we're going to need an API

key from a popular provider.

Fortunately, Anthropic shut off the the

Claude Code subscription token, but for

this we're actually going to use a

Codeex token. So, I'll log in with my

Codeex account and then we'll have that

all set up.

>> Cool.

>> Sweet. First thing we're going to do,

going to try and speedun through this

part. Going to click here to add a

server. Really not much extra you have

to do here. You can pick from any of

these costs optimized or regular

performance. One thing to note is that

you do need a decent amount of RAM to

run OpenClaw. Four should be fine, but I

usually air on the side of caution and

grab eight. And then if you're in the

US, you're going to want to choose US

East or US West. If you're somewhere

else, you're going to want to choose

essentially the closest one to you.

These change depending on where Henner

wants to serve exactly where the box is

from. But in this case, I'm going to

choose US East. And then there's not

much more you really have to touch here.

Ubuntu is set up. You want both these

public networking ports. These are

essentially the doors for how we're

going to get into the box. And then the

only other thing that is mentioned in

this guide on how to set it up that I've

already set up is the SSH key. This is

basically the password to log into your

virtual server. Volumes, firewalls,

backups, nothing to worry about here.

Um, and sorry, just one question on SSH

key. Is that effectively the is that the

key that

>> other applications are using to get into

the server? Like who like who uses the

SSH key in this context? Yeah. So the

SSH key is basically a password between

your computer and the virtual server.

>> I see.

>> And in theory when you're open to the

open internet anything can happen. But

in theory that makes it secure so that

you can SSH and actually like log into

and then have a terminal that's inside

that virtual server and only you can do

that. Now

>> keeping anything on the open internet is

scary. So, we're going to want to lock

it down, but this is kind of that first

security measure to give you a password

to try and block people from logging

into your machine.

>> Cool.

>> Volumes, firewalls, backups, placement

groups, labels, all those we can skip

down to here. Alex, do you have a cool

name for this open claw?

>> Um,

I can't use Jarvis. Let's use Ron.

>> Love it. Ron open claw.

Sweet.

So now that we have him set up, I'm

missing something I can tell. Oh, I

didn't click up here. Great. We have

that selected. And so basically for you

to you you paid for the second tier. You

said that the user could pay for the

first tier. That'd probably be enough

RAM, but you just were you played it

safe, paid for the second one. And

that's like what? A little over 10 bucks

a month.

>> Yeah. I mean, this this will come out to

literally $9.99. You can up it. You can

go bigger, but most use cases, you

really don't need it. Um, it actually

looks like I've run out of my IPs. So,

luckily, I do have a server that is

already set up that um I have not

provisioned yet, and that is Jarvis. So,

we're going to use Jarvis here. Um, I'm

actually going to hop in here.

Again, note all these IPs should

probably get blurred, but I'm going to

hop in here, copy these IPs, and then

I'm going to hop to this claude. So, one

thing that I would recommend when you're

setting this stuff up is to actually

have a documentation folder on your

local computer. Your local computer is

going to be how you're connecting to

everything with OpenClaw through SSH.

And again, that's just logging into your

server so that you can run commands on

it. If you're not familiar with cloud

code, this might be a great time to try.

I know we already have a bunch of

resources on using Cloud Code, but I'm

going to use Cloud Code here to actually

help me do the setup. This guide that we

have back here, I've already copy and

pasted and I've given it to Claude to

say,

you know, Claude's complimenting the

chef, I guess, but um I gave it all the

instructions and what we want to do. So,

I'm going to paste in that information

about um my IP address and the name of

the box. And I'm going to say great, set

this up and tell me when we

are able to SSH into the VPS. This is

essentially saying, "All right, Claude,

I need you to log into this. Here are

the credentials. You're going to go in.

You're going to lock it down and it's

going to set up all of those security

protocols that were already um listed

above. Great.

>> And and just a question here, how does

it know cuz I basically see okay, you

set up your VPS, you you have the IP,

you have the information about like I

see you have the name, you have the IP,

you give that to Claude and you say tell

me when we were able to SSH into the

VPS. Basically, when can my computer get

access to the virtual computer? How does

it know here? Where did you give it the

direction to like start hardening? Uh is

is that just using the walk through?

Like basically the walkthrough you

provided it. It's just going on to the

next step of the walk through.

>> Yep. So Claude didn't exactly do what I

wanted it to here, but that's fine.

Right now, all it did was it tried to

make sure that we're able to SSH into

the machine. It looks like we're good to

the box is live. It's here. We can

access it. Then the next thing it's

asking me is if we want to set up tail

scale. Tail scale basically gives you

your own private internet. Um it's

secure and it allows you to connect to

different machines without being

accessible from the public internet.

This is kind of stage one of security

where we really want to isolate that VPS

and your computer so that no one just

out in the wild can actually access your

server or any of the data on it with

your open claw.

love it. So, I'm going to ask it to set

up tail scale. It's actually going to

log into that VPS, install tail scale,

and then it will shoot me back a link

where I just have to confirm it. Um, one

confirmation step after this is

installed. And then we'll essentially

have tail scale on the machine. Next

couple steps are going to cut it off

from the open internet, set it up so

that we're only connecting with tail

scale and then add some further security

measures. At that point, we'll be able

to log in and start actually setting up

open claw.

>> And so just so I understand like when

you're using tail scale, what I almost

my visual here is that it basically just

creates this like safety bubble around

your around the open claw. What what

does that mean in practice? uh does it

basically just mean that like maybe you

have a good example for what tail scale

actually does that like in context is

helpful from a security perspective?

>> Mhm. So essentially the way that SSH

works is it lets us log into the machine

with that password. But when we're doing

that, we're doing that across the open

internet, which means that you know this

this virtual private server with its IP

address is accessible from the open

internet. That being said, unless you

have the password, you can't get in. But

I'm certain if we looked at that box,

it's probably getting bombarded by bots.

The second you spin something up, bots

try to attack it and they try to get in

and I've even had VPS's in the past with

nothing sensitive on them get kind of

compromised in this way. Um, I had a dev

server running once and someone put a

crypto miner on it and it it became this

whole thing but eventually I was able to

just turn it off. Um but tail scale kind

of from this diagram basically what it

does is it makes your virtual server

invisible to the public internet. So it

blocks it out and it makes sure that

only whatever devices you have on your

tail scale network can access it and

nothing else can really reach it. Plus

it's encrypted and it kind of saves you

from all just the lowhanging fruit of

security issues by having something on

the open internet.

>> Cool.

Sweet. Great. Cloud's giving me the off

link. I'm going to open this real quick.

This is probably another place to blur.

Great. We're connected. Oh, amazing. And

Claude's just taking it away. Great. So

now what's happening here is we actually

get a new IP address for Jarvis

that is on the tail scale network as

opposed to the open internet. So

essentially it's we had an address that

was open to everyone. Think of it as the

address of your house, but it's on

Google Maps.

Now Jarvis is in a gated community and

no one else knows who he where he lives

except for the other people in the

community. No one in the public knows

this address or can even access it.

>> Love it. Great. It's like that gated

community in the Bahamas where all the

celebrities live and no one else knows

who's there other than the other

celebrities.

>> Exactly. Jarvis. Jarvis is a celebrity

now.

Cool. Going to try and log into him

here. In the meantime, kind of want to

hop to some of these other concepts and

just like best practices to harden the

server and what we're going to do here.

I don't think we're going to go through

all of these just to avoid all the

config and setup and hoops to jump

through, but unfortunately, you do kind

of have to take those steps to actually

make this truly secure. All of these

steps are kind of classic cyber security

principles. And one thing I even like to

do occasionally is I run a codeex and I

say codeex

red team these boxes. Try and get in.

Try and break in. Find any security flaw

you can. And it's actually been really

good at saying hey I wasn't able to get

in but here's a place where you have a

lot of attack surface and here's how you

can harden it. Um and I'm no, you know,

expert in the security stuff. I know the

basics, but it's been really helpful in

giving me confidence that we've fully

secured this VPS and that no one else

can get it.

>> So, we're going to show the share this

in the show notes, but just because I

know you're going you don't want to go

deep in all of the security hardening

stuff for someone like I saw one of the

things is automatic security updates.

Another one is firewall. How should

people be setting this up since you're

not going through each one right now? Is

it literally as simple as just like if

they tell Claude to ingest this guide

and do it, Claude will take them through

all the steps?

>> Yeah. Yeah. I think um we can probably

go live basically right here. But um as

far as these security things go, we

won't go through line by line, nor am I

even going to have Claude set these up

on the one we just set up. But overall,

if you give these kind of guides and

descriptions, Claude will actually set

this up. And this includes a lot of

hardening. So you can kind of see in

this diagram here that we have your VPS,

this tail scale network, it's going to

block it off from the public internet.

This firewall is even going to further

block any public inbound traffic. We're

going to lock it down with SSH

hardening. This is going to give you

keys only. It's going to make sure that

traffic can only go from your computer

to the machine. Other machines can't

access each other. We're going to have

some intrusion prevention. Basically, if

some public address tries to get in, um,

we're going to use this package that

automatically bans IP addresses. And

we're going to set up auto updates on

the box so that if any new operating

system updates come out, it just

automatically stays up to date. Um,

classic security principle. No, no

single layer is bulletproof. So, we're

just going to layer them on and make

sure that this is fully secure. And

hopefully these instructions are easy

enough to just lob to Claude and have it

set everything up for you. And then you

kind of come out the end with a VPS

that's fully provisioned and secure that

only you can access.

>> Love it.

Now we're on to the fun stuff.

[laughter]

Um, it's time for Ron.

>> It's time for Ron. We're going to bring

Ron online. So what I usually do in this

step is I'll have Claude handle

installing OpenClaw, installing the

dependencies, making sure that we have

node on the um server just so that we

have everything to go or everything that

we need to actually run openclaw. And

then I'm going to log in and do the

onboarding.

>> And uh my my my friend was wondering

what is node? Uh, node is just um it's

just a fancy thing that runs code that

openclaw needs. Um, it's just a

javascript javascript runtime that um

lets you use javascript [clears throat]

code. Um, openclaw is written in

typescript. So, it needs something on

that server. Think of this computer that

we just bought this VPS as a brand new

computer that just came into this world

and it has nothing on it. So, we have to

install all the dependencies first.

Cool. I'm going to say skip the security

for now. Log in and set up node and open

claw.

Don't run setup or onboarding commands.

Just install it. Great. And then over

here, I'm going to SSH into Jarvis. This

is actually going to give us a terminal

inside Jarvis. So, we're actually on

that machine, which is pretty cool. In

here, we can run commands. We can work

with OpenClaw. We're actually going to

run the OpenClaw kind of CLI interface

and chat with the bot as we get it set

up. And this will take a second over

here. There's a nice little install

script from OpenClaw that automatically

installs all the dependencies. Um,

>> love it.

>> I have that link.

>> And for that, like did you did you have

to feed from your doc into Claude what

these install scripts were or is Claude

able to just find them? This started as

a battle of me going online and trying

to figure everything out. And then you

can see at the top, if I scroll all the

way up to the top of this chat, I

actually dumped the whole guide into it.

So, it knows all these steps from this

guide. Um, and that's been super helpful

to me in just

>> spinning up new ones quickly.

>> But it I will say the first time I set

it up, it was a bit of a maze. And

that's what we're here for today. We

want to get everyone watching set up as

fast as possible.

>> Awesome.

Sweet. So, we're in. We can now hatch

Ron/Jarvis.

>> I was going to say Ron's going through

an identity crisis ever since you had to

switch uh switch who he was in Tail

Scale, but I'm going to keep calling him

Ron.

>> Yeah, we'll go we'll go with Ron. We can

actually you get to name him. So,

um, there is a thing. It's probably not

going to show here, but they refer to it

as being hatched.

Yeah, it doesn't say on that setup

command, but when you first install

OpenClaw, it says, "How do you want to

hatch your claw?"

>> I love it.

>> So, we're going to hatch him. There's a

quick command called OpenClaw onboard.

I'm actually just going to zoom in here.

It's going to ask us a couple questions.

It's going to get us set up with our

account and then we're going to be off

to the races.

>> And just so for someone who's setting up

OpenClaw in lockep with you, do they

have to go do they have to basically go

in their terminal and do this command,

this OpenClaw onboard command within the

VPS or could you just have Claude do

this?

>> You have to do it in the VPS. So, you're

going to want to install it on the on

the VPS, log into it, and then run this.

This does take a second to boot up. And

then the reason you're going to want to

run this on the VPS and then go through

the steps there is because you need to

log into a few things. You need to make

sure that everything is set up. And

honestly, the onboarding is kind of fun.

You can just make config files and throw

them on the machine and it'll work. But

I actually quite like the experience of

going through and chatting with it the

first time to make sure it understands

who it is, who I am, and what we're

going to set up.

>> Yep.

>> Cool. Always recommend onboarding mode,

quick start. Then you've got a ton of

model providers to choose from. Um, I've

been clawed through and through, but

recently with the codeex ooth and the

codeex rate limits, like the actual

token limits that you get, I've been

preferring those for my open clause just

because I can get so much more usage.

Um, so I'm going to log in here with a

codeex token. Essentially, all of these

are going to look roughly like this

where you get some

URL to go to and then you're going to

log into whatever account you chose.

Then you're going to come back here and

paste some token. So, I'm going to hop

to this link. Once you're logged in,

you'll get a link. This is going to be a

redirect link and it's not going to work

on your computer because it's trying to

redirect to the link that actually came

from the other machine and your browser

and your computer can't exactly talk

between each other like that. So, we're

just going to copy this link and we're

actually going to give this back to the

open claw here. It's going to take that

redirect and get everything set up. You

can choose a model.

>> Is there a model that you recommend?

Yeah, I mean 5.3 codeex seems to work

really well in this open claw harness. I

did set one up with an anthropic API key

and opus I think is Opus is the goat in

agentic environments, but the problem

right now is just price. You can't use

your cloud code subscription and you're

paying a massive API price. So for me,

5.3 codeex does fine. And it's I I can't

hit my codeex usage if I want to.

>> Cool.

>> Now we get to set up our first channel.

We're going to do Telegram.

And then what's really cool is you get

some steps here. You're going to open

Telegram and actually message this

account called BotFather. And BotFather

is going to get you set up with the

Telegram integration.

So, I'm going to hop over to my

Telegram, and I've already got a

conversation with BotFather,

and I'm going to say slashnewbot.

Is Botfather owned by OpenClaw, or is

this like a

uh like a third-party

uh Telegram bot that just like people

use as a service,

>> you know? I'm not sure. I in my head it

was open. it was open claw. But thinking

about it now, I don't know if that makes

sense. I don't know if there's

>> honestly I was Yeah, that's why that's

why I was curious. I was I saw 5.9

million uh monthly users and I was

curious if those are all OpenClaw or

other other people.

>> Yeah, I I have no idea. I mean, this

this has enough functionality that I bet

it might just be a Telegram thing.

>> Yeah, cool.

>> So, we're choosing a name for the bot.

>> Ron,

>> it's going to be Ron.

Now you have to pick a like a username

for it that ends in bot. I found that

most of these are taken. So we're going

to do

>> Ron Ron Johnbot. Ron Johnbot.

>> I Ron John bot's got to be taken.

>> Wow. We got Ron John. [laughter]

>> Awesome. Cool. So now we get this token.

And this token is what actually lets us

use this to hook up our open claw to

Telegram. So I'm going to take this

token and I'm going to put it here.

That's our bot token.

And it looks like it worked fine there.

Great.

Now we're on to some of the config

stuff. So skills are super super helpful

in OpenClaw environments. Um, I kind of

skip them only because Peter has set up

so [snorts] many of these integrations

with either pieces of software that he's

worked closely with or exist or that

he's built himself. And a lot of these I

personally don't use or don't need.

Sometimes I set up the GitHub one. One

time I set up the Nano Banana one just

because I wanted it to send me more

pictures. But generally, I skip this

because you can always come back to it

and add the skills when you want and how

you want them.

>> And just to clearly break this down,

these skills that um the creator of Open

Claw created, like how should someone be

thinking about these? Like for example,

the Nano Banana Pro skill, what does

that actually do in practice? What does

that allow you to do? Yeah, any of these

skill files are just a markdown file

that kind of lay out

a a specific workflow or how to use a

tool and

kind of give the the agent guidance.

Almost think of it as like

for a human it would be a set of

instructions or a YouTube tutorial. For

the agent, it's just like a markdown

file. So there's all these like

different integrations that it can use

and work with and do things for you with

these tools. I much prefer rolling with

no skills at the start and then per claw

depending on what we're doing. I kind of

walk through a manual workflow. You

know, let's say we're we're working on

some project and we're doing a GitHub

PR.

I I would walk it through. Okay, you're

gonna make your changes. You're going to

push them. Then that's actually going to

show up in a preview deployment just

like a a preview environment. It's not

deployed on production, but it's

somewhere where you can test. Then you

can use, you know, this browser use

skill to go test it. And then once it's

tested, you can ship that to production.

So that flow I would do manually a few

times. And then I would say based on

this workflow, make a repeatable skill

for yourself. So you know this flow in

the future. And I've just found that be

much more customizable and functional

for what I want to be doing with Open

Club.

>> Cool. Awesome.

>> Cool. So we're going to skip that. Um

there's a bunch of API keys you can dump

into this to give you access to Google

Places, Nano Banana, Notion, OpenAI,

OpenAI, [laughter] 11 Labs. You know,

there's a ton of a ton of things you can

set up. Um, the 11 Labs one is kind of

cool because you can have your agent

actually talk to you. Um, but a lot of

these things are more, you know, when

you really want to get deep into the use

cases and play around with it. For our

setup, we definitely don't need these.

Now, hooks. I like to enable all of

these hooks. These hooks are very

helpful because what these do is they

automatically load context and

automatically save things like commands

or sessions that you're in to memory.

One of the other big things that

OpenClaw I think has um found a novel

approach to is long-term memory in these

agents and I think it's done a great

job. So we want to enable all of these.

>> And can you just explain what a hook is

exactly? Yeah, a hook, if you think of a

skill as a set of instructions that the

agent uses to do something, a hook is a

little piece of code. I don't have a

great analogy actually. It's a little

piece of code that runs when the agent

does something very specific. So, if we

look at boot, there's a boot.md file and

this runs when you start your claw. So

maybe every time you start it or it, you

know, kicks back into action, you want

it to do something right away when it

boots. You would put instructions in

boot empty. I see.

>> You can also have additional

instructions that come in right away.

You can also have this one that logs all

command events to a central file so that

as it's doing stuff, it's recorded

somewhere automatically. And then the

claw can actually go back and reference

that if it needs to see, hey, how did we

do this before? It can actually

reference that audit file um and then

make use of it for future.

>> I don't know. I don't know if this is

accurate or the right analogy, but

almost like when you're saying this,

what it comes off as to me is like these

are almost like

um

conditional or situational skills

relevant to like your actual computer or

like relevant to the the machine like

almost like I think of like skills in

the way you were showing them before are

relating to other applications and how

your opening claw basically acts with

and around these other applications.

Whereas like these are specific to the

open claw itself and they're conditional

based on certain events.

>> Yeah. Yeah. So these just get [snorts]

fired when things happen and sometimes

they're injecting context and sometimes

they're just saving what the claw does

so that later it can reference it.

>> Cool.

Cool. So we're in everything is set up.

What it's doing now is installing the

gateway. The gateway is essentially this

other core piece of open claw that gives

the agents a channel they can talk

through. Think of it as a big group chat

and in that group chat it can be your

telegram, your discord, your slack, but

it can also be the main agent fires off

a sub agent and the sub agent has to ask

it a question and it talks back. It's

basically this main bus where all the

communications between your main agent,

your sub agents, and you go through. And

that's what's running 24/7, kicking off

those heartbeat runs, keeping it alive,

and you know, really having it work

while you sleep. And

>> And why is that setup important? Like

why why why does that make the specific

like Open Clause specifically more

powerful? This kind of like team of

agents that are talking to each other.

So this is I think a lot of the products

and tools right now are trying to solve

this like agent collaboration

communication problem. The thing that I

love about the gateway is it's so

simple. Um I think there's actually a

diagram over here of it working. Maybe

not. I might have hallucinated that. Um

but basically this gateway is so simple

where you have your inputs. Your inputs

are either you writing from Telegram,

Discord, whatever messaging setup you've

connected. The inputs can also be other

agents and that flows through a queue.

That Q then gets picked up by the main

agent. It does something based on the

message and then it can send messages

back out. Whether that's a reply to you

on Telegram or a reply to you um on

Discord or a reply to a sub agent, it's

just this bus that lets anything come in

and then the main agent orchestrate

anything going out any of the other

channels.

>> Cool.

Okay, it's time to hatch run.

>> I do like to hatch in the the TUI. The

web UI I think could use some work. So,

if you're familiar with cloud code in

your terminal, this is going to feel

very familiar. We're going to hatch.

All that hatching does is text the bot,

wake up, my friend. [laughter]

>> I I feel I feel like a proud parent

right now.

>> I know Ron Ron is here. Um, so there's

kind of a flow that's set up in the very

beginning.

We're not going to go through like deep

deep detailed prompts in this right now,

but we have gone from zero to one and we

have an open clause setup. Um, and

following all those steps, we'd have it

on a secure box that we can only access.

And now we just kind of get to play with

our new open claw. So,

real quick, I'm just going to say

you are Ron.

You are my

personal art finder.

You are to go online

and find photos of art and send me them

and make a comprehensive profile on what

I like and don't like.

And then I'm gonna say I am actually two

people.

Oh my god.

CJ and Alex,

you should

store our art preferences

in your memory

and learn more and more over time what

we like.

And then what what vibe do you think,

Alex? [snorts]

>> I like photorealistic.

Okay. Alex likes photorealistic,

but what vibe do you want Ron to have in

his personality?

>> Oh, um,

sassy.

>> You are very sassy. [laughter]

Um, there's also a great post by Peter

on Twitter about how to get its soul up

and running. Um, and I like to say you

are very opinionated

and you act on those opinions.

Love it. Great. And we're going to kick

him off. He's going to take this, ingest

it, and hopefully we can actually get

him to run through one of these tasks.

One other thing that's always good to

test once you have this set up is

hopping over to Telegram,

clicking on Ron Johnbot. We're going to

hit start and we're going to say hi.

It is going to give you this pairing

code.

>> So just so I understand here, can you

you talking to Ron in Telegram, is that

the same as you talking to Ron in your

terminal?

>> Yeah. So that's that's the whole beauty

of the gateway. All of the messages can

come from different places and then all

of them can go out in different places

>> and Ron can actually choose it.

>> So Ron is online. Save this memory

already. He's sassy.

>> So we can give a starter profile. But

first I'm actually just going to say,

can you make sure you are configured for

Telegram?

Because now he's set up, he can run

commands.

He can actually manage himself, his own

workspace, his configuration.

>> So giving him all this information,

>> he should be able to just set this up

for us.

>> Love it.

>> Sweet. So we're good on Telegram. I'm

going to hop over here and say, "Hey,

Ron, can you hear me from Telegram?" And

he's typing. And here he is. Same bot,

different channel. And we've done it.

We've set up Ron end to end. Um, from

here, you know, my my only real tip in

your in building this up is actually

just dumping as much information about

you as you can. Really try to get it to

understand who you are, what your

preferences are. Um, and the bots are

really good at managing their internal

state to try and get you to a point

where um, it's performing better, it's

doing more for you, it's actually acting

on your behalf.

Love it. And one last question for you.

Um, is if I'm a busy professional or

executive watching this and I I've now

set up my run my open claw and I'm like,

okay, what do I do now? Like what is

actually a useful way that I can use

openclaw? Just based on you playing

around with openclaw a lot and just

understanding like what other knowledge

workers may like? like what do you think

is the lowest hanging fruit of how Ron

or their open claw could be

exceptionally helpful that let's just

say they start with one use case what is

the one use case that you think people

could feel the power of openclaw with if

they get it set up the right way

>> yeah so I like to think of it as if you

had a perfect assistant what would they

do for you and in any of those scenarios

it's going to be very custom that's what

openclaw is strong in doing these like

custom workflows. You know, the the

basics of monitoring your email,

monitoring Slack, helping you reply to

things, helping you stay on top of

stuff. I can actually show you my

personal assistant, Alfred. I had him

send me an essay that I think is about

laundry to just push our message history

up. But with Alfred, he is monitoring

kind of my Slack channels and my emails

and making sure to keep me up to date on

my to-dos. And he even has a little task

tracker of his own where I kind of made

some of this fake data. But he actually

puts things on here, messages me, keeps

me on top of things. So whatever the

simplest kind of task management way

that you can think about handling tasks,

organizing them, keeping yourself on top

of stuff, trying to offload as much of

that to the bot as possible, and then

finding where it breaks. There are

places where it's not perfect, but it

can do a lot more than I think people

give it credit if you just spend the

time to play around with it. And when it

does something wrong, ask it why it did

it wrong and what we can do to make it

do it right in the future. And the

amount of things it can self-correct is

still surprising me.

>> Love it. I feel like we we're going to

want to do a part two of this. That's is

that's all around like, okay, now that

we have OpenClaw set up, how do we

actually like if I want to configure it

to be my perfect personal assistant?

like we could probably spend like

another 45 minutes just getting it to

like be my personal assistant through my

emails, my Slacks, my texts, whatever.

So, I think that would be a good part

two for us to do.

>> Yeah, there's tons of that's kind of the

tricky part is like setting this up is

half the battle, but I think a lot of

people get to that point, have it set up

and go, now what? And there's a lot of

like places you can push it, but at the

end of the day, it is this. kind of the

first technology where it's always on

and you can customize it to whatever use

case you want, but there's a ton of

lowhanging fruit that I think is easy to

set up that can be immensely valuable to

anyone out there who's able to set this

up.

>> Love it. Anything else you want to share

before we uh finish things up?

>> No, I think that's it. I hope this is uh

helpful and I hope more and more claws

are hatched as a result of this

walkthrough.

>> Love it. CJ, thank you so much. For

those who are watching, again, look at

this as like a almost like a reference

guide. Feel free to watch the video, but

rewatch it as you're working through the

process of setting up your open claw.

We're also going to include the guide

that CJ created in the show notes, so

that one, you can reference it, but also

you can feed it to Claw to help you set

up a safe and secure OpenClaw. And then

more videos to come on specific use

cases and applications in your work uh

once you have your version of Ron set

up. Thanks so much for joining everyone

and we'll see you again soon. Later.