00:00:00
now if you open up the YouTube search
00:00:02
panel and you type in AI sales
00:00:04
automation you're going to be met with a
00:00:06
ton of NA workflows that look incredibly
00:00:09
complex are those workflows cool i mean
00:00:12
yeah 100% now do they add tangible value
00:00:15
to your business often time not entirely
00:00:18
now for the past 5 years I've been
00:00:20
working and consulting with a bunch of
00:00:22
agencies helping them leverage these
00:00:24
systems tangibly into their internal
00:00:26
processes now in today's video I'm just
00:00:29
going to pull back the curtains and
00:00:30
share with you exactly some of the tools
00:00:31
that I've been building and how they are
00:00:33
driving results and allowing these
00:00:35
businesses to actually improve their
00:00:37
lead flow so let's get into it now
00:00:40
before I dive into the automation I
00:00:42
pretty much want to do some background
00:00:43
in terms of why we're doing this i know
00:00:45
it's very very easy to get lost in the
00:00:47
technicalities but I think it's so
00:00:48
important as business owners to actually
00:00:50
know why we're doing what we're doing
00:00:52
which will only help how we actually
00:00:53
build the alteration down the line in
00:00:55
the future and for you as I mentioned
00:00:57
this will be given i will give you this
00:00:58
template us doing this kind of research
00:01:00
now will kind of help you understand
00:01:02
what you should go and change in the
00:01:03
template that I give you so why does
00:01:05
everybody want these automations in the
00:01:07
first place well traditional B2B sales
00:01:09
or sales in general it's very very
00:01:10
manual right often time you spend a
00:01:13
bunch of time trying to do lead research
00:01:15
and find out who's a good lead to reach
00:01:16
out to now thanks to AI where it's
00:01:19
really really easy for us now to do
00:01:21
research at scale it's a lot lot easier
00:01:24
but this introduces another problem
00:01:26
everybody's becoming lazy incredibly
00:01:28
lazy and they just expect AI to read
00:01:30
their mind you know people just say um
00:01:33
look at this profile and tell me if
00:01:34
they're a good lead but what even is a
00:01:36
good lead right like you wouldn't just
00:01:38
tell a virtual assistant to go and get a
00:01:41
good lead for you you would break it
00:01:43
down into real like instruction just
00:01:46
what is a good lead to you right but
00:01:49
people don't do that with AI they just
00:01:50
give AI thinking AI is this all knowing
00:01:53
all powerful thing but you need to tell
00:01:55
it exactly what you're looking for in
00:01:57
order for it to give you what it is that
00:01:59
you want right and again if you automate
00:02:01
crap you're going to get crap ultimately
00:02:02
so and it's quite frankly why I tell
00:02:05
people to peel back first and do things
00:02:07
manually and when you do things manually
00:02:09
you can map out every single step that
00:02:11
it takes for you to do what you're
00:02:12
trying to do so maybe you know that like
00:02:15
first you always like check the site
00:02:19
right and then oh and then maybe if I
00:02:22
copy this over here and I duplicate it
00:02:25
then you know second you always check
00:02:27
their LinkedIn right and then maybe you
00:02:31
know third you always like I could say
00:02:35
check like the decision makers and the
00:02:39
past companies they worked at for
00:02:41
example whatever it is that you So when
00:02:43
you're able to map out like manually
00:02:45
what it takes for you to get whether
00:02:47
someone's a good fit or a bad fit then
00:02:49
you can pass that into the AI which I
00:02:51
which is why I think it's so important
00:02:52
to do these things manually to even know
00:02:54
what it is we're trying to automate in
00:02:56
the first place right now all of this
00:02:59
stuff has just led to the past year the
00:03:01
past 2 years everyone is just sending a
00:03:04
bunch of noise and noise has just been
00:03:06
thrown from one inbox to another inbox
00:03:09
like "Hey XY Z something very very
00:03:11
generic but there's clearly AI written
00:03:13
and it's not going to get you anywhere."
00:03:15
Which is why a lot of these automations
00:03:17
we're seeing they're not really adding
00:03:19
value because they're not specific
00:03:21
they're not um they haven't really taken
00:03:24
proven manual steps and automated it
00:03:26
they're just automations but hey in
00:03:29
order to automate something you need to
00:03:30
know how it works manually which is why
00:03:32
I think it's really really important
00:03:33
that you take the time to actually know
00:03:36
what it is that you want to automate
00:03:38
first now a powerful combination when
00:03:40
you're able to effectively use in lead
00:03:43
genen like signals for example as well
00:03:45
as being able to prompt the AI with a
00:03:47
lot of information you have a really
00:03:49
really powerful combination to build an
00:03:51
AI um sales co-pilot now with all that
00:03:54
being said let me kind of get into what
00:03:55
it is that we're building here this this
00:03:57
social signaling campaign leverages two
00:03:59
tool very very simple it leverages air
00:04:02
tableable and it leverages n right now
00:04:05
air Table if you don't know what Air
00:04:06
Table is Air Table is pretty much a
00:04:08
Google Sheets on Stairways it's a Google
00:04:11
Sheets is almost a database where you
00:04:13
can create a bunch of tables and you can
00:04:15
like create automations around the
00:04:17
tables and the records a record is just
00:04:19
a row in a table in a sheet and you can
00:04:22
just do so much very very easily and
00:04:24
effectively it's a really really good no
00:04:26
code platform so N sorry Air Table is
00:04:30
going to act as our like storing
00:04:32
mechanism right and we're going to use n
00:04:34
to trigger some automations and then
00:04:36
whenever we want to make changes or
00:04:38
updates or store information we're going
00:04:40
to call air table to store that
00:04:41
information i hope that makes sense so
00:04:44
let me go back into n and kind of
00:04:45
explain what's going on here again as
00:04:47
you can see this is it yes there's a few
00:04:50
steps but compared to these crazy flows
00:04:52
you see on YouTube it's nothing to that
00:04:55
cuz often time those flows they're not
00:04:57
necessary it's too many automated you're
00:05:00
letting AI do way too much when
00:05:02
realistically maybe you want an AI
00:05:04
workflow with predefined steps when an
00:05:06
AI agent just does whatever the hell it
00:05:08
wants so what is even happening here um
00:05:11
in this social tracking system what
00:05:13
we're doing is we are
00:05:15
tracking for leaders on LinkedIn that
00:05:18
talk about something similar to what we
00:05:20
sell in which we can we can use their
00:05:23
engagement that they're getting to
00:05:25
pretty much take their leads and sell to
00:05:28
them because we know that person is
00:05:30
somewhat interested in what we sell so
00:05:32
let me give you an example um let's say
00:05:34
I don't know Andrew Huberman is like a
00:05:36
health enthusiast right now if you're a
00:05:38
health coach it might be valuable for
00:05:40
you to look at all of his engages to see
00:05:43
if any of them fit your criteria maybe
00:05:46
anyone engaging on Andrew Huberman's
00:05:48
post that is a founder or a
00:05:50
director so a they're showing interest
00:05:53
in health content but b they kind of fit
00:05:55
your seniority level that's a good
00:05:57
person to get in front of because now
00:05:58
you know that your message isn't falling
00:06:00
on deaf ears because that person is
00:06:02
somewhat interested in the thing that
00:06:04
you sell cuz you're a health coach right
00:06:06
so a similar approach here so what I've
00:06:08
done is in air table I've got a table of
00:06:11
thought leaders these thought leaders
00:06:12
are just three people you might be
00:06:14
familiar with them this is very very
00:06:16
simple of people that talk about like
00:06:18
leveraging AI and sales so somewhat
00:06:21
similar to what it is that I do now on N
00:06:25
creating a schedule and every day for
00:06:27
example we're getting those four leaders
00:06:30
and we're finding any new post that they
00:06:32
have made right um emphasis on new post
00:06:35
we don't want to scrape any previous
00:06:36
post because we would have already
00:06:38
scraped them and it needs to be up to
00:06:40
date there's no point checking their
00:06:41
post from 6 months ago because anyone
00:06:44
that liked their post 6 months ago may
00:06:46
not still now be in the market so it's
00:06:48
really really important that we have
00:06:50
this automation that is actively doing
00:06:52
something every single day to pull fresh
00:06:54
insights because if if a CMO of a recre
00:06:59
of a I don't know of a creative agency
00:07:01
has interacted with two of these
00:07:03
people's post in the past two days that
00:07:06
is some real intent that they're
00:07:07
somewhat interested in how can we
00:07:10
leverage AI cells into us into our thing
00:07:12
right and of course you want to track
00:07:14
that and then get in front of them in an
00:07:16
effective way which which we'll get into
00:07:18
in a moment if you have time in this
00:07:19
video so what you can see here is in air
00:07:23
table I have a table for thought leaders
00:07:25
and I have a table for post scrapes so
00:07:28
what that means is I'm keeping a track
00:07:29
of all the post that we do scrape so
00:07:31
when I come back to nan we have a few
00:07:34
steps here where we're filtering out any
00:07:36
existing post right so this is getting
00:07:38
the existing post and then aggregating
00:07:40
it to turn it into um an array so if
00:07:43
you're familiar with n n works of items
00:07:46
and arrays we want it as an array
00:07:47
because then we can basically use a code
00:07:50
node to basically filter out any of the
00:07:53
people as I'm doing
00:07:54
here then we're fetching the engages of
00:07:57
all of those posts that we just
00:07:59
pulled then we're for every engager
00:08:02
we're getting their profile and then
00:08:04
we're then passing it into an open AI
00:08:06
module here notice we're not using an AI
00:08:08
agent cuz we don't need to here and an
00:08:11
AI agent is great if you're using tools
00:08:13
like a fire scrape their website or do
00:08:16
something else if you're just speaking
00:08:18
with a large language model often time
00:08:20
you do not need an agent and in fact I'm
00:08:23
going to open this up and I'm going to
00:08:24
show you what I'm doing here so this is
00:08:26
a pretty detailed prompt because um this
00:08:28
is how you get really good results but
00:08:30
pretty much what I'm doing is I'm
00:08:33
telling the agent very very detailed
00:08:35
that look you're a source assistant i'm
00:08:37
going to give you a bunch of
00:08:39
information pardon me about a company
00:08:42
about a profile and a company what I
00:08:44
want you to do okay here's some
00:08:46
information that I may give you i want
00:08:48
you to have a conversation with yourself
00:08:50
to try and figure out if that is the
00:08:52
right person to sell to so I'm telling
00:08:54
it what I sell i'm an AI consultant i
00:08:56
sell AI agents and automation services
00:08:58
to service- based businesses i tend to
00:09:01
sell to people over a headcount of 10
00:09:03
and I like to sell to first world
00:09:05
countries so I want you to basically
00:09:07
have a dialogue with yourself you know
00:09:09
like think about who is a potential
00:09:11
customer and then I'm doing something
00:09:13
really really cool here and in fact I'm
00:09:14
going to make a separate video for this
00:09:16
cuz it is quite complex but I'm pretty
00:09:18
much telling the AI to output in XML
00:09:21
tags where I want you to output the a so
00:09:25
like why did you come to that reasoning
00:09:27
and then a classification of like your
00:09:29
actual output the reason for this
00:09:31
thinking stage is what we're doing here
00:09:33
is we're practically turning GPT40 mini
00:09:36
into like 03 right or 01 we're turning
00:09:40
like a GPT model into like one of those
00:09:43
thinking models because we're actively
00:09:45
telling the AI to think through your
00:09:47
process and then at the end of that
00:09:49
output your response and this um really
00:09:52
really increases the output of what
00:09:53
you're able to get so I'm doing that and
00:09:56
then it's telling me if the person's a
00:09:58
fit if they are a fit then not add them
00:10:00
into a table so again um it's not
00:10:03
incredibly complex but it is highly
00:10:05
highly effective because then what
00:10:06
you're left with is you're going to be
00:10:08
left with a
00:10:09
um a table of prospects but have shown
00:10:13
in intent in posts that relate to you
00:10:16
they fit your criteria and also you can
00:10:19
track how many times they're engaging
00:10:21
and every single time this runs we're
00:10:23
able to aggregate what what's happening
00:10:25
here so it's really really simple but
00:10:26
really effective what I would then do is
00:10:29
at the end you can add in a step to add
00:10:31
them to like an automation platform like
00:10:33
hey reach or smart lead but one thing
00:10:35
that's really important to note here and
00:10:37
this is where most people go
00:10:39
well where am I
00:10:42
going so where most people go wrong is
00:10:46
that they'll use
00:10:49
AI to send a pretty
00:10:52
generic
00:10:55
AI each and you've just kind of undone
00:11:00
all that hard work you've used AI to do
00:11:02
really really good account and lead
00:11:04
research and now you want to use AI to
00:11:07
send a message like you
00:11:09
know like big fan of your content for
00:11:13
example and now this just screams AI or
00:11:16
especially when you add that little um
00:11:18
that little
00:11:20
dash this dash just screams AI but
00:11:23
anyway like there's no real need to go
00:11:26
crazy with AI to produce the message
00:11:28
because all the AI was the heavy lifting
00:11:32
when it came to fight came to doing the
00:11:34
research and now if you know that this
00:11:36
person has shown intent on um your post
00:11:40
on in your niche there's no real need to
00:11:44
get crazy or fancy here you can send a
00:11:47
very very simple message like "Hey I've
00:11:50
seen you're interested in leveraging AI
00:11:51
agents into your ops um me too i'm a big
00:11:55
fan of the person you know XY Z." I
00:11:59
thought it'd be great to exchange a note
00:12:00
right just a very very easy networky
00:12:02
kind of message instead of you trying to
00:12:04
get way too like I don't know
00:12:07
unnecessary with the AI outputting the
00:12:09
message i personally always advocate for
00:12:12
just using AI for the research and then
00:12:14
when it comes to the message your list
00:12:16
that you have built is the message
00:12:18
because you've built a list of people
00:12:20
that they fit your criteria whatever it
00:12:22
may be their service businesses in first
00:12:24
world countries and they're showing
00:12:27
interest in your kind of niche so
00:12:29
already that's a personalized message in
00:12:31
itself there's no need to let AI ruin
00:12:33
all the hard work you've done i really
00:12:36
hope that makes sense and I'll share
00:12:38
some more videos in the future where I
00:12:39
kind of show exactly how that would look
00:12:41
so yeah so here's the workflow and in
00:12:43
fact I will run it it will take some
00:12:44
time to be fair maybe I'll come back
00:12:45
when it's done but um again getting
00:12:48
we're getting we're getting the four
00:12:51
leaders we're fetching their posts we
00:12:54
are then fetching the
00:12:56
engagement we're then fetching the
00:12:58
profile and the company for each engager
00:13:00
on that post we are then using all of
00:13:03
that information to find out if the
00:13:05
person is a good fit for what we're
00:13:07
selling we are then sort of you know
00:13:09
aggregating it and then we are updating
00:13:12
the the person back into our air table
00:13:15
and it's going to take some time so
00:13:16
maybe I'll just pause it for now cuz the
00:13:18
video is is coming on quite long but
00:13:20
again guys I will share the JSON in the
00:13:23
description or just ask me and I'll send
00:13:24
it to you and I'll also share the air
00:13:26
table base as well um so yeah this is
00:13:29
how you can leverage social signals and
00:13:31
using AI effectively to pretty much
00:13:34
create a campaign that is always on
00:13:36
which allows you to get in front of the
00:13:38
right people at and as you can see
00:13:40
they're coming in now at the right time
00:13:42
so yeah all the best guys and um take
00:13:44
care speak soon a bat
