Okay, so this is really sick. I just set

up a Hermes agent on this Nvidia DGX

Spark completely powered by a local

model that's running on it. I now have

an AI agent, a 24/7 AI employee working

for me completely on this local model on

this device, completely private, secure,

and best of all, free. Just the power of

the energy going into this device. And

look how cool this is. Isn't this the

coolest looking computer ever? In this

video, I'm going to show you how to do

the exact same thing. I'm going to show

you how to set up Hermes agent, have it

set up on a local model that will be

running on this DGX Spark. You can put

it on any device you want. For this, I'm

using the DGX Spark. I'll show you how

easy that is. And I'll show you how you

can have your full-time 24/7 AI employee

that's doing work for you at all times.

I'll even show you some really awesome

use cases that can be really helpful.

Whether you've never loaded up a local

model in your life, you're an expert, or

you're just curious in how all of this

stuff works, you're going to learn a ton

in this video. Now, let's lock in and

get into it. So, this is going to be

really fun. I'm going to show you like

the most cutting-edge technology ever.

Quick shout-out. Shout-out Nvidia.

Thanks so much for sponsoring this

video. If you watch this channel, you

know I don't take a lot of sponsorships.

99.9% of my videos are not sponsored. I

try to only work with companies that are

really amazing that I use daily. When

Nvidia reached out, I was super pumped.

One of the greatest companies of all

time. Shout-out Nvidia. Thanks for

sponsoring this. Let's go into the

strengths of local models and why a

local model with Hermes agent is so

sick. even send me a computer. I bought

this one myself months ago. So, that's

why this is really easy for me to do.

But, let's get into this. Let's talk

about local models, what makes them so

amazing, and why it's so powerful with

Hermes agent. Then, we'll quickly jump

into setting it up and having a lot of

fun in setting up experiments and doing

a lot of cool things. If you need to,

chapters down below, jump around if you

need to. Local models, I believe are the

future. I believe very soon everyone

will have their own super intelligence

on their desk. Why is that? Well, local

models are free. You can download them

online, load them onto your computer,

and they're basically free. Just you're

paying the cost of electricity, right?

So, the power powering the computer,

running local models does use up more

electricity. So, you're paying for the

power, you pay for the computer itself,

but unlike AI models running on the

cloud, you're not paying for every token

you use. You're not paying for

subscription plans, nothing like that.

Just the power going into the computer

and the computer itself. So, that's one,

unlimited usage of super intelligence,

pretty incredible. Next up, and this is

a big one, it's completely private. When

you use cloud models, right? So, when

you go to any of the AI websites and you

start chatting with their models, your

prompts and all your chat logs are

stored in the cloud. They're stored on

servers somewhere else. And well, I

don't think any of the employees are

going and reading your chat logs or

anything like that. Having your most

important conversations being private is

really important. So, you get the

privacy. If I unplug the internet from

this DGX Spark right now, I can still

keep using it. I can still chat with the

model and use my Hermes agent cuz it's

all local and private and secure. So,

everything stays local, everything stays

in your computer, no one can snoop on

your chat messages. That's big. It's

customizable. So, the models themselves

are customizable. You can install these

things called LoRAs, which are basically

plugins for the model, which allow you

to customize them really any way you

want. If you want them to really learn

your voice, if you want them to learn of

how to make specific types of images,

you can customize these models any way

you want. I can literally customize

right now so it sounds like me. So, all

of its output sounds like me, which is

amazing.

So, customization, that's big. It's

educational. You're learning about AI,

right? When you install it and start

using it locally, you learn how it

works. And I think this is the most

important technology of all time. I

think it's super important you learn it.

So, when you download it onto your

computer, you start using you learn

about the most important technology I

think humanity has ever invented. That's

huge. It's just straight up fun. It's

fun doing this. It's fun looking at your

desk, seeing a DGX Spark there, any

computer you have, and knowing there's

like superintelligence running on that

device. It's really, really fun. You're

allowed to have fun. You're allowed to

do things for the fun of it. Not

everything needs to be a hyper-optimized

dollar spend. You're allowed to just

have fun. And the last one here before

we get into actually promise we're about

to have a little fun here, use cases.

When you can run a model 24/7 for just

the cost of power, it unlocks so many

use cases. And I'll go into those use

cases in a second, but you really get a

powerful 24/7 AI employee. So, real

quick, why the DGX Spark? Everyone knows

I got a whole bunch of devices. I got

Macs, got everything under the sun here.

Why the DGX Spark though? Why do I like

this device? It is super easy. It makes

this whole process super easy. You buy

it, you plug it in, and it just works.

It turns on and you can immediately

start putting models on it. It's also It

also has the full suite of Nvidia

developer tools on it, so you can do all

the customization that's really only

possible on Nvidia chips. So, if you're

into customization as well, this is like

the best way to do it. So, the Spark is

a great way to run these models and run

Hermes on it. And I'll show you in a

second how easy it is for the agents you

have on your computer to actually manage

this device and install different models

on it. And by the way, that's not even

part of the sponsorship. I just threw

that in cuz I've been using this device

for a long time, and I use it every day,

and I love it. So, let's go into setting

it up. Let's get the device set up. I'll

speed through actually setting up the

computer, and we'll spend most of our

time setting up the agent and the model

itself. Feel free again jump around the

different chapters below if you want to

hear about any specific subject. First

thing you need to do though is actually

plug the computer in. Unlike other

computers out there, you do not need a

monitor for this computer. You can run

this in what is called headless mode.

Headless mode basically means there's no

monitor, you don't have to control it,

you don't have to touch it. It's just

powered on and you're able to use it

basically as a tool for whatever your

main computer is. So, once you have it

plugged into the wall, into the power,

it just turns on and it starts up a

local network. If you look at your

instruction manual with the DGX Spark,

it has actually like network information

on it. You just connect to that network

on your main device. Now your main

device can control that DGX Spark. So,

what you're going to want to do for this

is you're going to want to install

Hermes Agent on your main computer.

Powered by a cloud model at first, we're

going to have the local Hermes Agent

work side-by-side with our cloud agent

so we can get this all set up. Once you

have Hermes Agent set up, you have it

connected to a cloud model, whether it's

a ChatGPT account or using the Anthropic

API, whatever you want, we can now get

to work in setting up our local models

and setting up our new DGX Spark. Now

that we got the Spark plugged in, this

is the prompt I'm going to give to my

Hermes Agent. I'll put this down below

if you're setting this up with me. I

just purchased a new DGX Spark and want

to set it up. I want to run it headless

and I want you to be able to control it.

I'd like you to walk through setting up

the Spark then install Tailscale on it

so you can control it from any device.

So, a whole bunch of things going on

here, let me explain. Number one,

remember as I explained before headless,

that means no monitor or other devices

are going to control it. And number two

is Tailscale. For those not familiar,

Tailscale is like one of the most

incredible free pieces of software out

there. It's basically going to allow all

your devices to create its own private

network. So, your Spark and whatever

other computers you're using will be

able to be on the same kind of virtual

network so they can talk to each other

and control each other. So, what this

prompt will do is have your Hermes agent

walk through setting up your Spark.

It'll have you connect to the network on

the manual. It'll walk through

step-by-step. And then it'll go on there

once you're connected, set up the

computer itself, and then install

Tailscale so that moving forward it's a

lot easier for your Hermes agent to

actually control that device and use it.

It's going to allow your Hermes agent to

go on it, install local models on it,

set them up, run it, and then talk to

the local model anywhere you are in the

world cuz it'll be on that same private

network. So, you hit that, you're going

to be all set up. It's super easy.

That's the beauty of these AI agents.

That's the beauty of using Hermes agent

here is that it makes it super easy to

do whatever you want. Before this, it

would be super intimidating to do

something like this. It'd be super

intimidating to set up a supercomputer

on your desk. People would go, "Oh."

They throw their hands and go, "Oh, I'm

not tech. I'm a non-technical person.

This isn't for me." If you're watching

this video right now and you're

thinking, "Oh, this is not for me. I'm

super non-technical." Throw the word

non-technical in the garbage can. Take

it out of your lexicon. Non-technical

does not exist anymore. Everyone is a

technical person. As long as you got an

AI model, as long as you got a Hermes

agent, it doesn't matter what it is. You

go to your agent, you say, "This is what

I want to accomplish. You accomplish it

for me." And it will figure it out.

Everyone is technical now. I don't want

you to limit yourself by calling

yourself a non-technical person. Every

time someone in my life says, "Oh, I'm a

non-technical person." I I tell them to

kick rocks. Figure it out. Download an

AI model and figure it out. You got

this. I promise you. So, you got the DGX

Spark set up. Next, we're going to

download and install a local model. For

this video, we're going to use Qwen 3.6

27B. Qwen 3.6 is the latest set of

models from Qwen. This is, in my opinion

the strongest local model there is. It

is super fast, it is super efficient,

and at the same time, it is super

intelligent. It is up there with a lot

of the frontier models from the last few

months. So, it is very, very good. So,

here's what we're going to do now that

you have the Spark set up. Say this, "I

want to download and install Qwen 3.6

27B onto the DGX Spark so I can use it

to power AI agents and chat with it."

27B just means it's 27 billion

parameters, which is kind of on the

medium-to-smaller side, but through

tests I've done and through a lot of

tests other people have done, it's still

incredibly smart, but you're going to

get the benefit of really, really good

speeds. So, I'm going to put this down

below as well, so feel free to steal

that. That will go and have your agent

go into your DGX Spark, search the web

on there for that model Qwen 3.6 27B,

find the version that's appropriate for

the Spark, download it, which might take

a little bit. These models are pretty

chunky, but it'll fit onto your new

computer, and then actually load it into

memory. Right, these models, the way

they work is they're loaded onto your

memory, and then you can start using it

and communicating with it. This took

about 20 minutes or so on my computer.

It might take about the same for you

depending on your internet speeds and

all that. So, do that. Feel free to

pause here if you're doing this

alongside of me and come on back. So, if

this is your first time doing this and

you just set it up and it downloaded

Qwen and then got it loaded into the

memory, first of all, congratulations to

you. You just did something that like

0.000001%

of the world has ever done, which is

host your own super intelligence on your

desk. Do yourself a favor, look down at

your computer and go, "Holy crap, I have

super intelligence on my desk. I am a

sovereign individual. Nobody can take

this away from me. No one can cut off my

AI service. No one can take this away.

Even if the internet goes out forever, I

will have super intelligence to get me

through it. So, shout out you, this is

your first time. Congratulations. Now,

let's get into actually using that super

intelligence. So, here's the plan what

we're going to do here. We're going to

set it up inside Hermes agent. I'm going

to show you how to do that. And then I'm

going to show you three really awesome

use cases for Hermes running on local

models. I just like to do one fun thing

real quick before we get into that. And

that is set up a front end chat

interface for our model. This allows you

to just test it real quick and kind of

feel that magical moment of, "Oh my god,

super intelligence is talking to me

locally." So, do this. Please build a

front end for our new local model. Make

it a chat interface and test that it

works. I will put that down below as

well, so you can steal that if need be.

And when you hit enter in that, your

Hermes agent's going to go. It's going

to actually build out a really quick

front end interface, or really simple

code it out, and then it will connect it

to your local model. So, mine is all

built out here. Look at this. This is

awesome. Spark 1 3627B. Let's give it a

test. Hey, are you there? And enter and

now let's see how we do. Is our local

model going to work here? Is it going to

chat with us? Super intelligence, are

you working? Oh, there it is. Yes, I'm

here. How can I help you today? That's

amazing. That came straight from our

computer on our desk. If you're not

geeking out with me, then you're not

alive. That's amazing. So, we've

confirmed it's working. We confirmed our

local model is talking to us. We

confirmed the super intelligence is

doing its thing. Now, let's actually

plug this into Hermes agent so we can

have our local working for us. So, what

makes Hermes agent

really cool is it is built in to be

multi-agent. What that means is it is

just one command away from Hermes being

able to create an entirely new Hermes

agent that can run side by side with

your current one. So, you can do really

cool workflows with multiple agents.

What we're going to do is we're going to

have it set up a second Hermes agent for

us and we're going to say plug it in to

our new Qwen 36 model that's running on

our DGX Spark. So, let's do this. I just

So, I'm going to use this prompt. I'll

put this down below as well if you want

to take it. I just installed Qwen 36 27B

on our DGX Spark. Please set up a new

Hermes profile that is plugged into that

local model and name it Qwen. I like

that name Qwen. So, I'm going to name

our new Hermes agent Qwen. Let's hit

enter on that. Now our Hermes agent's

going to go and it's going to create

this new Hermes profile which is

basically just a second Hermes agent and

it's going to plug it into that model

running on our Spark which is going to

be sick.

All right, here we go. It found the

llama server where the uh local model is

running. It's going to ask us for

permission. Let's give it the okay on

this and it's going to start getting to

work building out that new Hermes agent.

So, we'll have two Hermes agents working

for us, two full-time employees, one

that's going to be completely free run

on the Spark and I'll walk through all

of the different use cases and how do

you do the two models together after

this. All right, looks like it's all

set. All right, so to run this all we

need to do is hermie p Qwen. That's

going to run the Hermes profile for Qwen

which they just set up. So, let's get

this popping. Let's do this. So, I'm

going to open up my terminal here. I'm

going to paste in hermes -p Qwen. I'm

going to hit enter and boom, there we

go. Look at that Hermes agent. Oh,

that's incredible. There it is Qwen 36.

That's the model we got running locally.

We now have an AI employee 24/7 running

locally on our desk on our computer.

Let's say hey here. Let's see how it's

doing. Hey, are you there, Qwen? Come

on, here we go. Hey, I'm here. I'm

actually Hermes agent not Qwen but I'm

ready to help. What do you need? Okay,

so it doesn't know its name but it's

here. It's working. We can talk to it. I

guess when you set up a profile it

doesn't give it its name. So, let's give

it its name now. By the way, your name

is Quen. And that will save it to

memory. So, moving forward, we can refer

to our new AI agent employee by its

proper name, Quen. Got it. Quen it is.

I'll go by that from now on. And then

you can see this is a cool part about

Hermes. It tells you every tool call,

every memory updates. Updating the

memory, the user refers the assistant as

Quen. Really, really cool. Now, let's go

into the three use cases for our local

agent. We have a local 24/7 AI employee.

What do you do with it? It's probably

the number one question I get. It's

like, "Oh, what are the use cases you

do?" Let's go through three. I'm going

to give you a beginner one, a moderate

one, and an advanced one. So, no matter

where you are, you're going to have a

use case that works for you. Let's start

with the beginner one. I strongly

believe everyone should be investing. I

don't think that's a crazy take. I do a

lot of investing myself. My favorite way

to do investing and learn about

investing and get better at it is by

using my AI agents. And so, what we're

going to do is we're going to have Quen

set up a daily report for us that is

going to investigate AI stocks and AI

companies cuz I personally, not

financial advice, think that's a great

place to be investing. So, let's do

this. All right. So, here's the prompt

I'm going to do for this. Please, every

morning at 9:00 a.m. research AI stocks

for me. These are stocks for companies

that stand to benefit from a 10-year AI

build-out. Give me a report on the top

companies that have great moats and

great businesses and tell me why they

have moats. I want to invest in

companies that don't really have

competition or are the very best in

their field. So, that's why I want to

know about these moats. What's great

about Hermes agent is you can have these

scheduled tasks. So, every day at

specific times, it does things for me.

So, I do this. I have this built out.

It's going to now send this to me every

morning at 9:00 a.m. Every morning at

9:00 a.m. it will deliver me this report

so I can read it, see if there's

businesses I want to invest in, make

sure I spend my money wisely, and make

sure I can grow in the future. Now, my

AI employee is going to be a stock

researcher for me. This is really

beginner. Anyone can do this and you're

going to get value out of doing this.

What this is doing in the background is

just scheduling a cron job. For those

that don't know, cron jobs are basically

just scheduled task for computers. In

this case, it is for our agent. And now

every day at 9:00 a.m., the cron job

will fire. It'll tell our agent, which

is running on our local model, to go and

do the research for us and give us that

report. And boom, look at this. Done.

The cron job is scheduled here. The

details, daily AI stock report, every

day at 9:00 a.m. Next run's going to be

tomorrow, 9:00 a.m. And it is all set.

Each morning will search for the current

stock prices and news. That is sick. So,

now we have our daily research reports

up. Let's get into the intermediate use

case for a local AI agent.

And again, this is completely free other

than the cost of the electricity going

into your computer. So, you basically

have your own personal local free AI

researcher, which is amazing. Let's get

into the moderate use case here. This

next one is really cool. It's something

I do often and that is repurpose

content. What I'm going to do is I'm

going to give our local agent a link to

a YouTube video. Even if you're not a

content creator, this will be helpful

for you. What I like to do is I give it

links to my own YouTube videos and say

repurpose this into a newsletter for me.

But what you can do is say, "Hey, check

out this video link, get the transcript,

and tell me what lessons we can learn

from it." You can do it on this video

right now that we have here. So, let's

do this. I'm going to grab one of my

past videos and I'm going to say, "Get

me the transcript of this video, then

repurpose it into a newsletter." And I

paste it in and I'm going to hit enter

and it will be off to the races. It's

going to go And one thing Hermes does

really well is just figure out how to do

things. It'll figure out how to get that

transcript, download it, and then

repurpose it new content for us. Again,

for you, you can have it get the

transcript and then take learnings from

it, right? Say, "Hey, Hermes agent,

check this out and see what we can learn

from this video. Apply new skills. Get

me the top lessons." and it'll do it for

you. So, so many things you guys can do

with this as well. If you want to do as

long as I mean, just take the link to

this video and give it to your agent.

And this is so cool. Look at this. It's

figuring out the skill to get YouTube

content. It's building its own skills.

Got the full transcript. Here's the

newsletter version and boom, it's

writing out the newsletter for me. It

just like so amazing to think about how

this is happening all locally, all just

on the computer on my desk. This is all

just getting figured out. I mean, we

really live in the most amazing time

ever. So incredible. And here's a little

twist on this use case because it's

happening locally, right? Because I'm

not paying for tokens on a cloud, maybe

I schedule this so every hour it

searches YouTube for a new AI video,

downloads it, gets the transcript, gets

the lessons, and now my Hermes agent is

automated improving itself every hour

searching for AI videos, getting lessons

from it, self-improving itself, and

doing that non-stop. That's incredible

to think about. That's something you

can't really do with cloud models cuz

you'd be paying for every single cycle,

for every single token. Because this

just costs the electricity going in

computer, you can just have this go all

day, right? Just constantly downloading

new videos and self-improving itself. It

really is amazing the possibilities to

think about with the Hermes and the

local model. All right, here we go.

Let's get into the advanced use case and

that is vibe coding. We are going to

have our Hermes agent vibe code a to-do

list app for us. So, very simple, just

for demonstration purposes. You can now

have your local models do vibe coding

for you. This is one of the most

compute-intensive activities people do.

Some people spend thousands of dollars a

month on these different vibe coding

tools. You can now get it just for the

cost of electricity. So, let's have Qwen

vibe code for us. Really is amazing

thing about this, you just now have

unlimited vibe coding, but that's one of

the benefits. That's how you get an ROI

on these investments in these computers

is doing things like this. So, let's do

this. Please build me a to-do list app

where I can add new task with priorities

and dates for those tasks. Make the app

beautiful and clean, and we're going to

hit enter. And now our 24/7 AI employee

is going to be building out us out an

entire app. All done locally. Really,

really cool. And what's great is you can

even go and you can plug this local

model in to the many various vibe coding

tools out there. There are ways to plug

it into Claude code, Codex, open code,

whatever you want to use. You can now

leverage this local model in any vibe

coding tool you want, but Hermes is a

great coder. So, using Hermes to build

things out works totally fine. All

right, look at this. It is all complete.

Open in your browser. All right, we're

going to open this up. Here's what it

does. Add task, mark complete, delete

tasks, filter between all active, done.

Wow, it's got a lot in it. This is

amazing. Let's go. And this is all built

locally. How incredible is that? All

right, let's do this. Let's take the

command, and we are going to run it so

we can use our new app. And it opens up.

Let's see what we got. Let's see what

was built 100%

locally. Look at this. This is actually

sick. This looks like it was made by

like a cloud top-of-the-line frontier

model. how good Qwen 3.6 is. Let's do

this.

Edit the DGX Spark Hermes video. Let's

add this task. Boom, there it is. And

you see that nice animation? That

animation was so sick. Let's do I want

to see that again. Make another video.

Add the task. See, oh, that is nice.

That is nice. I love that. Now you have

your free unlimited vibe coding tool

ready to go. You can build whatever you

want. No limitations at all. That is the

beauty of this. This is amazing.

Anything you want to build, you now can

build it. All done locally on your desk.

Really, really well done. If you learned

anything at all, make sure to leave a

like down below. Subscribe. Turn on

notifications. All I do is make amazing

videos about AI. I do live bootcamps on

vibe coding and building things every

single Friday in the vibe coding

academy. Link down below for that. So,

check that out as well. Shout out Nvidia

for partnering up on this. I've had the

DGX Spark forever. I've been running AI

agents on it forever. So, for them to

reach out was amazing. So, thank you so

much for that. Also, another thing, make

sure to hit the link down below for more

information on the Nvidia DGX Spark. An

incredible device to be hosting your own

models on. Appreciate you guys watching.

Let me know down below in the comments

what you want to see next. Do you want

to see more use cases for Hermes agents?

want to see more advanced workflows? I'd

love to hear your opinion. Thank you for

watching. It truly means the world and

I'll see you in the next video.