Here are five boring AI automations that

you could sell today for 1,500 bucks a

pop or more. These are not flashy chat

bots. They're not AI agents. What they

are are unsexy but very simple

straightline automations that can add

value to virtually any business. I sold

systems just like this when I scaled my

automation agency to $72,000 per month.

They're going to be in a blend of

make.com and n and I'll show you exactly

how to build them over the course of the

next few minutes. And by the end of the

video, you'll have added five more great

systems to your arsenal. Let's get

started. The first system is a search

intent scraping system. What the system

does is it finds lists of high quality

companies that are hiring for a specific

position. Then it scrapes those

companies, gets CEOs, decision makers,

and founders at those businesses, and

then adds them to a cold email sequence

using a bunch of automation. I'll run

you through how all of this stuff works,

but first, for context, let's say you're

a company that wants to hire somebody.

Where do you go? Well, you basically

have two major options, at least in the

United States. The first is a website

called Indeed and the second is

LinkedIn. So what companies do when

they're hiring for a role like an SDR

sales development representative is

they'll create a post on LinkedIn

looking for people to fill. Floor coding

warehouse just did this. Fulcrum just

did this. Blanco Technology Group just

did this. Right now what we can do is we

could take this page aka this URL and we

can actually pump it into a scraper

created on a service called Appify. If

we do this, what we can do is we can

actually go through the previous page

automatically and scrape the company

name, the company website, the

description of the job, everything that

they're looking for, their compensation,

and we could add it to a big

spreadsheet. Okay. Once we have it in a

big spreadsheet, we can fire it off to

one of many email verification services.

The system uses one called any

mailfinder. And then on the back end of

the system, we can do some additional

things like research the decision maker,

give us some more context about them

autonomously before finally adding them

to a cold email service. This video is

going to be using instantly, but you can

also swap with whatever you want. Um,

smart lead is a big one. There are a few

other ones as well. So, in terms of how

the system works under the hood, as you

can see, the very first thing we do is

we run an actor. Now, I'm putting in a

hard-coded URL here, but I'm sure you

guys can imagine you guys could swap out

this URL however you like. That URL is

equivalent to the URL of the LinkedIn

job, the post that we just showed you a

minute ago. Okay. From there, we will

get the data set items of that post, aka

this run just occurred. It deposited a

bunch of data into this data set on

Appify. And what we're doing is we're

just extracting it here. The end result

is we have a giant list of companies.

Okay, Air Garage. We have a company

LinkedIn URLs, company logos, location,

salary info, and so on and so forth,

just repeated for all of the jobs that

we've scraped. So in this case, I think

we scraped 30 or 33 or something. From

there, I have a bunch of filters set.

These filters just check to see does the

website exist because we need the

website to proceed. Is the website

linkedin.com? Sometimes LinkedIn posts

ad posts that go to careers.linked, so

this just filters them out. Are the

number of employees at the business less

than 150? This might be useful if you're

looking to work with small businesses

like I am. Assuming that they are okay,

what we do is we jump into a database.

Now, I'm using Google Sheets as my

database. You guys can use whatever you

want. What this Google Sheets database

basically does is it just checks to see,

have we added the same company name to

the database before? If so, don't

proceed. What we want to do is we only

want to pitch companies once. We don't

want to continuously pitch people over

and over and over again every time the

scraper runs. And this is a very simple

design pattern that allows us to do

that. So, assuming that the job we're

scraping doesn't already exist in this

big database with the company name and

the title and the tracking ID and so on

and so forth, then what we do is we

actually add them to the database right

over here. Okay? So, mapping all of the

fields from that app search. And as you

can see at the end of it, we get a ton

of data. Finally, I'm using an open AI

or an artificial intelligence module,

GPT4 mini in this case, but you guys

could swap this out with whatever model

you guys want to basically filter out

jobs that don't contain or include

things that I'm looking for. So, I left

this very vague here, but essentially,

you can use this filter to determine

whether or not a job is relevant to us.

Now, think about the logic here. We're

only filtering jobs that aren't already

in our database. So we're significantly

reducing the token count that we're

feeding into the artificial intelligence

and we're ensuring that we're only

filtering jobs that are new. Now once we

filter the job, what we finally do is we

search for a decision maker using a

service that I really like called any

mailfinder. You guys could swap this out

with any API call to any similar

enrichment service. What we're doing is

we're looking specifically for CEOs that

operate at the domain. And then I'm also

passing some additional information, row

number, company name, and campaign

status. Then I'm also adding what's

called a web hook URL. This just allows

us to set up a web hook somewhere else

and then watch and wait to see all the

results roll in. You don't need to do

this, but I do it because I like making

sure these systems are pretty airtight.

Okay. Now, once we're done with that,

what we do is we go over to this second

scenario here that's waiting with a web

hook. If I run this, we'll catch all the

data currently in the queue. And this

sends a node right off to Perplexity,

which is an artificial intelligence

service that allows us to browse the

web. Very similar to GPT web search, but

this is sort of the first iteration of

it. And assuming that we get some

information on who the prospect is. So,

what I'm doing is I'm actually just

feeding in a bunch of messages about who

the person is and seeing if we can look

them up. What we receive as a result is

we get a bunch of data on who that

person is. In this case, Molina, where

she's based out of, how long she's been

working, and so on and so forth. And we

do is we actually feed this information

into an AI module that generates a

customized icebreaker that allows us to

reach out to her. Instead of just

saying, "Hey, Melina, how's it going?

I'd really like to sell you something."

What we do is we use AI to take that web

search data and then build us an

extraordinarily customized icebreaker.

So the end result of this is hey Molina

admire how Propel Champions diversity

and digital talent because diversity is

important to her. Also love your thought

leadership on the future of tech because

I think she published some podcast or

something. From there we add that to the

database. We actually update the row

that we had earlier with a bunch more

information aka full name all of their

data basically over here. And then

underneath what we do is we add an

icebreaker and then a link. We do an

HTTP request instantly, which in our

case is just going to a test campaign.

This test campaign is already set up and

basically is just queued up to send the

icebreaker as well as a couple of other

points of information. So the end result

is we now have an email sequence that is

crafted, ready to go and automatically

already sending emails completely

autonomously. All we need to do is just

add a LinkedIn search URL to the end of

the system. This is going to say

something like, "Hey Molina, admire how

you do X, Y, and Z. I really love to get

in touch with you because I saw that you

were hiring for a certain position. I

think that I have a solution that might

be able to solve your problem for

significantly less money and make it

really, really easy for you to take care

of this. Here's what the solution would

look like. The second system is an AI

podcast repurposing engine. Now, the way

that this works, and I built this out on

my channel just a couple weeks ago, so

if you guys are already familiar with

this, this is why. You enter the URL of

a podcast. Once you enter the URL of a

podcast, it will then transcribe the

entire podcast, extract anything of

relevance, and then use that to develop

highquality social media posts that you

guys could publish on Instagram,

LinkedIn, or Facebook. So, let's give

this a go. I'm going to go over here and

go Nick Sarrive Jack Roberts YouTube

podcast because we had a podcast

published on his channel just a little

while ago. I'm going to copy this URL

and then I'm going to paste this into my

system. Okay, what this is going to do

is if we go back here, we're now getting

the transcript via an Aify actor. Ampify

being the same scraping service that I

was using before. And you'll notice that

it comes up quite often because I'm a

big fan of Apify. I think it makes my

life a lot easier. As you can see, the

one we're using is called YouTube

Transcript Ninja. So, it's actually now

extracting the transcript for that video

specifically. Because this is a little

bit longer of a video, I think this one

was 40 minutes. You might have to wait a

minute or two. After this is done,

you'll see we've now finished with the

scrape. You now have a bunch of data

over here which if I make JSON, one of

the items is a long transcript from

start to finish where Jack and I talk

about, you know, AI and automation and

so on and so forth. From there, I feed

this transcript into artificial

intelligence open AI with a prompt like

you take as input a long meandering

transcript and you identify the 10 most

interesting engaging points. Then you

generate a bunch of JSON containing

these points in the following structure.

Okay, I give it some rules. I give it a

quick little data dump and then it goes

and it actually generates me a list of

10 engaging points. Once it's completed,

we have a list of these 10 points on the

right hand side. So there's a transcript

of the paragraph of interest. There's

some context and then there's some

feedback that I have the model actually

give me on how we could make it better.

There's a deep explanation of that

section. Then there's a short image

description because I want to generate

some images on this later on. From

there, we use a split out node. This

just turns our data from one item into

10 items, which allows us to loop over

them and iterate them. Then from here,

what we do is we pass them through three

separate nodes. The first is an

Instagram post generator. The second is

the LinkedIn post generator. And the

third is a Facebook post generator. So

I'll only open the Instagram one for

brevity, but essentially we take as

input a section of a transcript right

over here. We tell AI, hey, I want you

to write it like an Instagram post. Then

it goes out and does it. After that, we

generate an Instagram image. Okay. Now,

the way we're doing this is with Dolly.

This is by no means the best AI image

generator on the market. In fact, I

think as of right now, it's probably

kind of somewhere near the worst. But

the end result is we get some cute

looking image like this. In our case, I

think we were talking about the

labyrinth or something. And I've chosen

to use a watercolor bunnies as my style

and my theme for maybe my blog or

something. So that's what that looks

like. And then we take all of that

information. Then we add it to a Google

sheet called my little Instagram sheet

right over here. And this is now just

basically an entry into a database that

a later workflow can use to

automatically publish on my behalf.

We'll do the same thing for both

LinkedIn and for Facebook. And the way

that I've separated this is I have

multiple tabs here, one for LinkedIn,

one for Facebook. And these posts are

sort of custom curated to that specific

platform. You know how sometimes

LinkedIn posts are longer than Instagram

posts which tend to be short. That's

sort of the idea. Now after that what we

have is we have a schedule trigger node

in the second half of this workflow.

What this does is every day let's check

it at 7 a.m. This goes through the

previous sheet that I'm showing you over

here. And it actually looks to see hey

which row does not have a posted on

value? Probably this one here. So we

haven't actually posted this yet. It

then goes and it publishes this to the

specific platform of interest. So if

it's Instagram, it'll use the Instagram

graph API or the Metagraph API to post

on Instagram. If it's LinkedIn, it'll do

an HTTP request and then publish

directly to LinkedIn. Then it's

Facebook, it'll publish to Facebook. I

don't want to have to go through and

then delete them all from my Instagram,

LinkedIn, and Facebook. So I'm not

actually going to run this one, but in

that way, what we get is we get a fully

contained system that allows us to

generate a list of, let's say, 10 blog

posts and then just drip them out one at

a time. The next system I built in

make.com. And this is one of the

simplest and most straightforward

systems you could sell to basically any

business on planet Earth that operates

using invoices. I've shown this a few

times just because I think it really is

important for people to realize the

systems that make money do not have to

be these big, complex, scarylooking

agents that you guys see all over

YouTube right now. In reality, usually

it's the very simple systems like this

that you guys can actually generate

substantial ROIs on cuz they look a lot

less complicated because you're

typically able to explain them and

interpret them for the business owner.

And because in this case this does

something very useful. Let me explain.

So over here I have a Google sheet. This

Google sheet is filled with just a big

list of invoices that I've sent with

different days. Okay. So this was an

invoice we sent to TechCore Solutions.

This is another one we sent to Global

Dynamics Corp. This is one we sent to

Quantum Analytics. Another one we sent

to Blue Sky. Hopefully you guys could

tell, but all of these are examples. I

just had AI generate me a bunch. The

value here though is you can actually

have this hook up directly to basically

any payment processor out there. Stripe,

QuickBooks, Zero, whatever the heck you

guys are using, you guys can actually

just use that as the data source instead

of the Google sheet. Okay. Now, as we

see, there's a date sent column over

here on the lefth hand side. What this

automatic invoice collection system does

is it basically allows us to follow up

on invoices that haven't been paid yet.

And I can't overstate just how much

money is usually tied up in unpaid

invoices for the average B2B agency or

manufacturing company or or so on and so

forth. So, this is a really big deal. It

allows us to make a lot of money for the

company very quickly. Okay, so the way

that this works is we start by searching

through the rows in that database. This

is what you'd replace with, let's say,

something like Stripe or Zero or

whatnot. And specifically, we're

checking to see if the status field is

equal to overdue. So basically, back in

the Google sheet, there's this little

status field. And if the status field is

equal to overdue, then we return it.

Okay? So there's overdue, overdue,

overdue. For the purpose of this demo,

why don't I just say that they're all

overdue. So what happens is we'll return

these. Okay? And now because they're all

overdue, we're going to get a big list

of results. See 20 here. Then what we do

is we grab the date. Okay, we basically

see how many days has it been since

we've sent that invoice. So if it's been

7 days, what we do is we go up this top

route and then we create this email. If

it's been 14 days, we go down this

second route and we create another

email. If it's been 21 days, we create

another one. If it's been 28, 35, 42.

And the value here is all of these email

modules are basically just writing a

follow-up in slightly different words.

So all you have to do as a business is

one time write your sequence of

follow-up emails, have some source that

runs every day like I showed you a

minute ago and then you just send people

follow-up emails every day assuming or

rather every week or so, whatever sort

of cadence you want um with this system.

So if what I do here is let me just

limit this to I don't know the first

five so I don't send a ton of emails. If

I run this, this will pull up all of the

sheets. They'll then grab all of the

dates. Now, what it's going to do is

it's going to look for ones that are

seven, I guess, 14, 21, and 28. I'm

realizing now that I don't think we

actually have any that are those exact

dates. So, let me just grab all of them

and see if there's one. All right, so it

looks like we now have a couple here. It

looks like it identified one invoice

that has not been followed up with in 21

days. Then, it identified another one

that has been outstanding for 28 days.

Okay, so I don't know exactly which ones

these would be. I just quickly updated

the date field here so I could do this

demo. But the end result is we send an

email to this person, Rachel, saying,

"Apologies. I know I followed up about

this a few times now, but just wondering

about that invoice I sent. Let me know

if there's anything you need on it,

please." So, you can imagine, you know,

after 28 days of still waiting for

somebody to pay your invoice.

Communication like this is fair and

reasonable. If it's been 21 days, maybe

William checking in on my invoice from a

couple weeks ago. All good. This is a

very simple and easy Lego block that you

could slot into any business to more or

less immediately add value. This next

system is a cyclic content generator

that returns as output a Google doc with

an extraordinarily wellressearched and

in-depth article. This was one of the

very first systems that I've ever

published on YouTube. I built that back

in make.com I think about a year and a

half ago when I was just getting up and

running on this platform and it had

great results and a lot of people have

used this since. But there's still a ton

of value with the system. So, let me run

you through what it looks like. In order

to run this, what we do is we click this

test workflow button. This little form

will pop up and it'll ask us for two

things, a keyword and an email. The

keyword is what you want to create a

blog post about. So in my case, maybe

I'm running a blog that does travel

stuff and I want to optimize for the

keyword Scotland cabins. Okay? And I'm

just going to send myself this email as

a result. You click submit and

immediately after you click submit, the

very first thing that happens is we run

an open AI search that goes out on the

internet. I'm communicating with this

via API that goes out on the internet

and then it actually finds blog posts

that other people have written about

this keyword. Then it builds outlines

using their content. Okay, this is

really cool. It's like a parasite sort

of style system. So 10 of Scotland's

best designed luxury cabins, one in

Edinburg and so on and so on and so

forth. Okay, it actually goes and it

finds a bunch of them. After that, what

we do is we find specific citations from

those blog posts. Then we ask for

high-ranking articles about Scotland

cabins and the outlines in markdown

format. So we're doing we're basically

building this structured outline. After

that, we extract and we format these

outlines, okay, using another OpenAI

call. From there, what we do is we

actually progressively generate better

and better outlines. So now I'm feeding

this into another OpenAI module which

generates a high quality comprehensive

outline for a topic given a crappier

outline over here. And then what we do

is we separate that into sections. So

now instead of it just being one big

outline, what we actually have is we

have an item containing an array. Inside

of that array is the intro, the next

step, the next step, the next step. And

this is really the key and the reason

why this is a cyclic generator. We split

these out into separate items and we

actually have AI write us a section for

every item. So we're actually passing in

a single heading into AI and having it

write an additional or rather a new

section. Um so we're piecing together a

whole article just uh you know heading

by heading by heading. Now obviously

because we're writing this in this case

10 times this section takes

substantially longer than most other

sections. But while this is running let

me explain the rest to you. Here is a

limit node. This limit node is simply a

little preventative bug fixing measure

that I put in because sometimes I had so

many headings in the system that it was

difficult for me to keep track of it.

What we do after is we aggregate all of

these sections together and then we

actually summarize the below article

section in three sentences returning the

output in JSON. Now the reason why we're

doing this is kind of nifty and what I

want to do is I want to take this

summary and I actually want to use it to

generate an image for that section. Once

I have the image to that section, what

I'm doing is I'm actually going to feed

in the previous section plus the image

to construct a new blog section with

both an image and the text itself. So,

we're being kind of nifty here and we're

I don't know if you want to call this

hacky or if you want to call this

ingenious. I prefer ingenious, but

basically what we're doing is we're

constructing the article section by

section. We're generating an image for

every section. The current lowest

hanging fruit here, which sometimes

occurs, is the OpenAI image module will

time out or rate limit if you don't

currently have a high enough tier. So,

if you guys are a tier one and you find

that this happens from time to time,

just spend a little bit of money and

upgrade to tier two. I think you need

like $30 or $50 on the card. If you use

AI as much as um I do, this shouldn't

necessarily be a big deal. You'll use

that $50 reasonably quickly. So, after

that, we feed in all of the objects into

the OpenAI generate an image module.

After the images are done generating, we

get their outputs here. We'll actually

merge the outputs of the previous node

and that image node, and then we'll get

a picture alongside a section. So, let

me actually show you what that might

look like. Looks something like this. In

my case, I'm just using pen handdrawn

illustrations, which, you know, end up

being okay. I wouldn't say they're the

best in the world, but they're also not

the worst. Then we do some data

processing nodes. We aggregate those 10

items to one item. Then we convert them

all into HTML before creating an HTML

text file. And then this section right

over here, this is just something you

have to do if you want to get like a

Google doc um to uh be generated nicely

with good formatting. If I go over here

to where it says share link and email,

what you end up with is you end up with

an email notification basically inviting

you to share the link. So, if I go over

here to my Gmail, we'll see over here

that an item has been shared right over

here. The ultimate guide to luxury and

unique cabins in

Scotland. So, I'm just going to open

this with a Google doc cuz that's HTML

back there. Once we open it with a

Google Doc, you'll get something that

looks like this. Now, the images aren't

entirely perfectly sized or whatnot. If

you wanted to go that extra step, you

would have to make an HTTP call to the

Google Docs API. This is reasonably

simple to do. I just, you know, I didn't

want to spend God knows how long

fiddling over the various API

configuration parameters. Also, I

personally like I'm not a big fan of

just publishing stuff like this

immediately. I do like to have a human

in the loop when I publish blog content

because I personally find I think

anybody here that's done any sort of

publishing will find if you just

sprinkle a few minutes of somebody's

time going over an article fixing even

minor mistakes. You improve the end

output like the quality of that article

multiple orders of magnitude versus if

it's just entirely automated. So, this

is somebody that's run a $92,000 a month

content writing company that used AI to

do this sort of stuff. I still stand by

it, but you know, having slight little

formatting issues or whatnot when people

read the article is not the end of the

world. But, your end result is something

like Scotland has been has long been a

favorite destination for travelers

seeking self-catering holidays, offering

visitors the flexibility and comfort of

a home away from home experience set

against stunning natural landscapes. In

recent years, cabins and self-catering

rentals have notably increased in

popularity with approximately 31% of

visitors choosing cabins as their

preferred accommodation. That's pretty

cool and interesting. Thank you very

much, AI. As you guys can tell, you guys

could actually generate some pretty high

quality content with this, assuming that

you guys get the the image prompt down.

And there are a variety of other image

generators that you guys could use for

this. There's nothing wrong with

publishing this directly to, let's say,

WordPress if you guys wanted to, or some

other blog service, web flow or whatnot.

All you would have to do is just replace

that last set of modules with instead

of, in my case, generating a Google doc,

just publish directly to WordPress with

whatever logic is required in order to

do that. Okay. And then the last system

here is a simple email categorization

system I built in make.com and I've

showed this a couple of times. This is

one of the simplest systems to get your

foot in the door at, you know, a small

to mid-size business because it handles

an issue that most founders have been in

personally. And that's where you just

get a ton of emails. Your email ends up

being the bottleneck of your company and

your whole business. And so what this

system does is this basically just

watches for a new email that comes in,

forwards it over to a web hook, and then

it categorizes it using artificial

intelligence. And I'll show you this in

real time in a moment. This AI is

responsible for giving it one of four

labels. Sponsorship requests, people

selling me stuff, invoices, and

receipts, and worthwhile. This is just

for my own business. Obviously, you

would categorize this however you want

for the client that you're working with.

Then I give it some additional

instructions. And then all it does is it

moves the email to the right category,

marks it as red. Yeah, that's uh more or

less it. So, let me show you what this

actually looks like. If I run this once,

we're now currently waiting for that web

hook to come in, right? I'm just going

to open my second email here. My first

is going to email my second. Say, "Hi,

I'd like to sell you something. Hello,

this is me selling you something. Please

reply, please."

not too far off from uh some of the

messages I actually get. Okay, so I just

sent myself an email. Uh that email is

coming. Gmail just takes a second

because it has to like hold on to it for

5 seconds to give you the ability to

undo it. Okay, we then returned the

email here with a bunch of text

underneath. Then finally, we categorize

that email. What you'll see is the end

result is the category is people selling

me stuff. So the email categorization

system did a good job. It is not

worthwhile. So then what we do? We just

move them into that label. Now you have

to move using this module's parlance but

in Gmail it's called add a label. So now

if I go back here let me give this a

quick little refresh. Go down to people

selling me stuff. You'll see that this

has now been marked as people selling me

stuff. And then as you can see like it's

it's it's gray which shows that I've now

looked at it or whatever. Same thing

here. We no longer see in the main inbox

and my life is a lot easier as a result

of simple systems like this. So are all

of the business owners I've sold them

to. The cool thing with this system is

you can actually just show the client

how to update the filter and then you

can have them play around with the

filter. It's actually pretty simple. You

record a simple loom. you say, "Hey, you

just jump right over here into this." Or

what you could do is you could pull it

from, let's say, a Google sheet or a

Google doc or something and have them

edit it there. Variety of different

means with which you can, but this also

just gives them a level of control over

the end result in the system and makes

them feel like they're more of a part of

it. Awesome. You guys could sell any of

these systems for $1,500 a pop or more.

I've seen people sell a couple of these

systems for more than 10K, and I myself

have sold my content generator system

for at least that amount on a couple of

occasions. There's a lot of value in

simple linear left to right flows like

this, as I'm sure you guys could tell.

Not everything has to be fancy chat bots

or AI agents. You guys can print money

so long as you solve a customer problem

and not just like pitch them any

solution. So my hope is now you guys

have five more ways to solve customer

problems here. And I walked you through

it from start to finish. So you guys

should know everything that you need to

about all the systems. You guys will

find templates and blueprints to

literally every one of these systems

inside of Maker School. It's my zero to1

automation community with daily

accountability to walk you through

everything you need to know in order to

get your very first customer in this

niche. I will literally give you a

dayby-day road map. So, however long you

stay in my program, you will have a list

of tasks to do on day one, a list of

tasks to do on day two, and so on and so

on and so forth. We have over 2,000

people in the community as of the time

of this recording, and I increase the

price every 100 members. So, if you've

been on the fence, let this be the sign

and let this be the permission you need

in order to take that next step and dive

into the lovely world of AI and

automation. Aside from that, really

appreciate everybody watching my videos

till the end. If you're still here,

you're a real one. Like, comment,

subscribe. Do whatever you can to help

me bump up to the top of the AGO. I'll

catch you on the next video. Thanks so

much.