January 28, 2025 • 35 mins
Unleash the Power of Google AI Studio: Transform Your Marketing Strategy in 2024
Attention business leaders: Are you ready to level up your marketing with cutting-edge AI? Join us for an exclusive live webinar with Jimmy Slagle, AI-powered marketing consultant and founder of generAI, as he takes us on a hands-on deep dive into Google AI Studio.
What you’ll learn:
- Why this tool is a must-have for business leaders in 2024.
- How to use Google AI Studio for video content insights like never before.
- Practical steps to implement it seamlessly into your marketing workflow.
Jimmy has done it all—from running a TikTok agency to founding an AI startup and helping brands unlock the power of AI tools. He’s here to show you practical applications of Google AI Studio and how to use it to transform your marketing strategy.
Link to Jimmy's Video Ad prompts: https://gener-ai.notion.site/Eva-Prompt-Database-16632402789480b984e0e1f736cf3907?pvs=4
Join our course, AI Business Transformation Course on February 17 at 12-2pm for 4weeks and get $100 off using LeveragingAI100 - https://multiplai.ai/ai-course/
About Leveraging AI
- The Ultimate AI Course for Business People: https://multiplai.ai/ai-course/
- YouTube Full Episodes: https://www.youtube.com/@Multiplai_AI/
- Connect with Isar Meitis: https://www.linkedin.com/in/isarmeitis/
- Join our Live Sessions, AI Hangouts and newsletter: https://services.multiplai.ai/events
If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Isar Meitis (00:00):
Hello and welcome to another live episode of the
(00:04):
leveraging AI podcast, thepodcast that shares practical
ethical ways to leverage AI, toimprove efficiency, grow your
business and advance yourcareer.
This is Isar Matis, your host,and I have a really exciting
show for you today.
I will start with.
A really piece of exciting newsfor myself personally.
And for you as listeners to theshow, we just crossed 200, 000
(00:24):
downloads off this podcast.
Now, when I started this podcastabout a year and a half ago, I
never thought that number wouldbe possible, but here we are.
So first of all, thank.
All of you for being loyallisteners.
And if it's your first time,welcome aboard.
You're joining a very largegroup of people, that drove this
number.
So I'm personally very excitedthat we're here to me.
The number is less exciting thanthe importance of what's we're
(00:46):
actually doing, right?
It's, it's very obvious to.
Everybody out there today,including recent recent research
and surveys by the worldeconomic forum that surveyed
more than a few thousands ofCEOs, that the number one thing
that is stopping AI success incompanies right now is AI skills
and knowledge.
That's the number one reasonthat AI is not growing and not
(01:06):
helping businesses grow as fastas it can.
And listening to this podcast isjust one way to do this.
For free.
And while you're walking yourdog, doing your dishes, driving
your car, mowing the yard,whatever it is that you're
doing, that you're doing this.
So as I mentioned, first of all,thank you so much.
I couldn't be more humbled andthankful for all of you
listening to the show anddriving you to these crazy
(01:27):
numbers, but it wouldn't bepossible without us providing
real value.
And the value today happens tobe talking about a specific AI
tool.
So there are many AI tools.
out there, right?
There's people use ChachiPT andGemini and Claude and Midjourney
and Lama and the, you know,every time there's like a new
contender, a newcomer likeDeepSeek now with DeepSeek R1,
(01:50):
which are amazing.
And there's always new tools,but there's always these tools
that are kind of like hiding inclear sight.
And yet most people are ignoringthem.
Despite the fact they provideincredible value and they are
free.
And so one of these tools isGoogle AI studio.
And our guest today is an AIstudio ninja.
(02:13):
So Jimmy Slagol has been in theAI world for Longer than most of
us.
So he actually started in playwith AI and working with
companies, helping themimplement AI before Chachapiti
came out.
So a year before Chachapiti cameout, which is more than most of
us can say, and he's beenfocused on that universe for The
(02:33):
last three years.
And that's what he's been doing.
And one of his favorite toolsthat again, most people don't
know about, or at least don'tuse as frequently as he does is
Google AI studio.
And he's going to share with ustoday on some tricks that you
can do in AI studio that areextremely powerful.
So I'm personally very excited.
I love Google AI studio.
I use it a lot.
So I'm sure we will geek about afew of these functionalities,
(02:55):
just So I'm very excited to haveJimmy as a guest of the show.
Jimmy, welcome to Leveraging AI.
Jimmy Slagle (03:46):
Thank you.
I appreciate it, man.
Yeah, I'm excited for this.
And yeah, to date myself, on, onwhen I started to use AI.
So I'm from Fargo, North Dakota,which is like a, you know,
normally it's usually a littlebehind the curve in terms of
technology.
but I started to make a lot ofcontent around AI in that like
early 2022 era.
And, one of like the local news,kind of journalist companies
(04:08):
reached out and asked if I'd bewilling to like talk about, you
know, my five favorite AI tools.
And, and it was so early thatlike ChatGBT wasn't even one of
those on the list because ithadn't come out yet.
so, you know, I'm sure that thatlist that I wrote is probably
completely irrelevant today.
But, but yeah, it's just, justfunny that, that, there was the
list out there that wouldn'tinclude ChatGBT of the best
(04:29):
tools.
Isar Meitis (04:31):
Yeah.
Yeah.
It was the days that everybodywas still doing like the chat
writers, right?
It was, I'm trying to thinkwhich was the one that got the
first, and then they changed thename like three times after
that, that was like the big blogwriting AI tool.
Yeah.
Yeah.
Jimmy Slagle (04:44):
I know like Jasper AI.
Jasper.
Isar Meitis (04:46):
Jasper.
Yeah.
But, but, but they had adifferent name back then.
Jimmy Slagle (04:49):
Yep.
Yeah, it was Jarvis.
Cause then I think exactly,
Isar Meitis (04:53):
there we go.
So yeah, that, that was probablyone of the items on the list.
I'm sure it
Jimmy Slagle (04:57):
was, I'm sure it was, but, but yeah.
Isar Meitis (05:01):
So yeah, for all of you who are joining us live,
whether you are with us live onLinkedIn or live on zoom.
So first of all, thank you.
And thank you for joining us forthis live session.
If you have any questions, Anyquestions, just post them in the
chat.
If you have anything else to addor comment, please do that.
Please start by introducingyourself.
Just say on LinkedIn, where areyou from?
What are you doing?
(05:21):
What can tools you look haveused a Google AI studio before.
So we kind of know where you arein your journey.
And before we get started, onelast, piece of.
Information or, or update, weare launching another cohort of
the AI business transformationcourse.
So I've been teaching thiscourse personally since April of
2023.
(05:42):
So not as early as Jimmy hasbeen in AI, but.
Very short into the AI game andhundreds or maybe even thousands
of people have taken this courseand are transforming their
businesses based on theknowledge that we provide.
It's four sessions, two hourseach.
So in a month you can be in acompletely different state of
knowledge and understanding bothin means of hands on skills on
(06:03):
multiple use cases, as well ason clear, roadmap and blueprint
on how to actually implement AIsuccessfully in your business.
If that's something you'reinterested in doing, there's
going to be a link in the shownotes.
We're going to drop the linkright now in the live session.
If you're here with us live, andif you're listening off this
podcast, you can use promo codeleveraging AI 100.
To get a hundred dollars off theprice of the course, the course
(06:25):
is filling up pretty quickly.
We just literally open it andthere's already like 10 people
signed up and we don't allowmore than 20 people at every
single cohort.
So if you're interested ingetting some proper AI training
for 2025 and you should comejoin us and act quickly.
But with that, Jimmy, let's talka little bit about Google AI
studio.
And I want you to start bytelling people Like, what is it?
(06:46):
Because it's not a chatbot,right?
It's like, it's something alittle different.
So let's explain to people whatis Google AI Studio, and then we
can dive into specific usecases.
Jimmy Slagle (06:54):
Cool.
Yeah.
Are you, are you shit?
cool.
If I share my screen and justshare.
Yeah,
Isar Meitis (06:58):
absolutely.
Definitely share your screen.
For those of you who arelistening to the podcast while
you're driving, don't stop andtry to look at the screen, but
if you want, this will beavailable on our YouTube channel
as well, but we will explaineverything that's on the screen.
So if you're just listening,you're not going to miss
anything important.
Jimmy Slagle (07:13):
Awesome.
Yeah.
So what's interesting aboutGoogle AI Studio is this is
technically for developers.
It was built for developers.
So people that wanted to kind ofuse this to then go and create
their own applications that'spowered by Google, this is the
interface that they use.
And so that's what makes it alittle different than the
Geminis of the world, the ChatGPTs of the world, or the Clods
(07:33):
of the world.
So this looks a little more,intimidating than.
what you're used to with, withchat GPT.
But what I love about Google AIstudio, and, and what we're
really going to dive into todayis it is the only, large
language model, at least that,that I'm aware of that's able to
analyze an actual video file.
And so what I mean by that is Ican upload an actual video.
(07:56):
and so if I take this and thisis like an ad, for example, I
can use that as the, data sourceinstead of it just being like an
image or a PDF or anything alongthose lines, it will take a
frame per second of this videoso you can now essentially use
chat GPT for any sort of video,marketing or use cases that you
(08:19):
have, which again is prettyendless.
So.
For example, I can say righthere, like, describe this ad in
extreme visual detail, and itwill be able to essentially do
that analysis, and, and give youexactly what is showing up in
this ad, which there's tons ofuse cases that we can dive into.
(08:39):
This is a really funny if youhaven't heard of the perfect
gene.
They're very like tongue incheek marketing.
but, but, but yeah, so it's,it's, it's a really just
fascinating tool and that's whatfirst got me interested to be
like, oh, wow, I never knew thatthere was a, there was a, large
language model that, That coulddo something like this.
So from the beginning, that's,that's kind of what, what makes
it a little different from, fromthose models.
(08:59):
we can, we can later get intosome of the other cool stuff
that Google is coming out, with,with like the real time, but for
now we'll focus just on thevideo analysis.
Isar Meitis (09:07):
Okay.
So before we dive into more ofthe video analysis and how it
can be used, like what ouractual business use case,
there's a question from Gwyn,and I hope I'm pronouncing it
correctly.
and the question is what is thepurpose of Google AI studio?
There are many tools out there.
So just want to Level set as weget into this.
So I will say you, you mentionedone aspect that it was initially
(09:28):
built for developers, but thereality is many of the tools
that are in it right now arevery much common person
friendly.
You don't need to write anycode.
You don't need to do anything.
And when you open Google AIstudio and to get to it, it's
just, AI studio.
google.
com.
If I remember correctly.
And if you go there, there'slike a menu on the side.
Of different tools that you canpick from.
(09:49):
So it's actually a lot more thanjust one tool.
You, there's multiple thingsthat you can do with it.
the first one is more or lesslike a chat bot.
And the biggest differencebetween that and just using
Gemini is a, it has someadditional functionality than
you do before B, you have morecontrol on how the model
actually works.
So you can control thetemperature, which controls how,
(10:11):
kind of like.
aligned with reality it isversus how will it be creative.
That's the best way I candescribe it.
And you can control some otherfunctionality and you're
actually running it against theAPI, meaning you need to give
them your credit card and you'rerunning against tokens, which
has pros and cons.
One of the biggest Rose is theyalways release their newest
(10:32):
models on Google AI studio firstbefore it's available on Gemini.
So right now, if you're runningGemini, you can get 1.
5 pro.
And if you're running here, youcan get Gemini two and Gemini
two flash are available to run.
The biggest benefit of Geminitwo versus 1.
5 is it has an even Biggercontext window, and you can
(10:54):
upload out to 2 million tokens.
And for those of you who don'tknow what that means, that's
about 1.
5 million words, which is by farthe largest contest window you
have in any tool out there.
So if you have a huge amount ofdata that you want to run
through a tool and ask questionsabout it, the next best one is
200, 000 tokens on, Claude, andthis is 2 million, that's 10 X.
(11:17):
More data that you can upload,and they're claiming 97.
7 percent accuracy whenanswering questions about the
data that you upload 97.
7 percent accuracy is almost asgood as my as my wife, and it's
way better than the averageperson, and so it's a great way
if you have a huge amount ofdata.
That's one reason to go to thistool.
(11:38):
What other stuff I'm, I'minterested to hear from you,
Jimmy, if you have other stuffthat you want to say about this
and how this is different thanjust going to any other tool.
Jimmy Slagle (11:47):
Yeah, no, I mean, I think, I think you hit it,
head on, especially the contextwindow.
Like I, I, I don't think we canunderstate how important that
is.
For example, like one of thecool things, cause I think we
all know, The prompts that yougive these models really are
what make them effective.
And so for like an AI creativestrategist, I mean, we, we have
(12:07):
taken almost everything that wehave learned.
And so, for those of you thatare tuning in, not visually,
this is if you paste it into,into 75 page, document, just
teaching this, this model,everything that it needs to know
about creative strategy.
So.
What the hooks are, what thebest ad concepts are, and just
everything else to make surethat it knows, exactly the
(12:28):
language and, best use cases andpractices in this space.
and so that's what makes this,this a, a just very, very
powerful, model.
Isar Meitis (12:37):
So again, to explain to people, this is a 75
page prompt.
If you load this to any othertool, you already take in either
all or a big chunk off thecontext window that is usable in
a single chat.
And here it's a drop in thebucket because there's 2 million
tokens and you've just taken,let's say 50, 000.
(12:58):
It's still.
50, 000 out of 2 million versus50, 000 out of 125, 000.
If you're running a chat, CPT asan example.
And so it's, it's a, it's ahuge, huge benefit.
And as I mentioned, the otherthing is if you're doing
research on variety of data andyou're loading multiple data
from multiple sources, and I'veuploaded, you know, Excel files
with over a million rows intothis, which will not open in
(13:21):
most tools that, you know, andyou can still ask questions
about it and get accurateinformation.
So it's very, very powerful fromthat perspective.
Let's jump back to the videoanalysis, right?
So there are tools out therethat can give me a summary of a
video, but that's verydifferent, right?
So let's explain how is thatdifferent?
And then if you give examples ofwhat use cases it can be used
for, that will be very helpful.
Okay.
Jimmy Slagle (13:42):
Yeah, for sure.
So right now, what youtechnically can do from a video
is just get the transcript.
it, the, they're outside ofGoogle AI studio.
there's no way to be able toactually get the visuals that
are happening within that video.
and so when you can getsummaries from videos, normally
that's just looking YouTube oranywhere else and just telling
you, Hey, here's everythingthat, that happens.
(14:03):
But when you get into the worldof marketing, whether that's
creating organic content.
or creating ads, you need an AImodel that's able to actually
see what's happening in thevideo to be able to be your,
your co creator, if you will.
you know, I like to, I like togive people the example all the
time.
If I showed you, let's just sayit's, you know, a video on
TikTok and you had to close youreyes and just hear what's
(14:27):
happening and, and, you know,essentially get the script of
it.
You can only give so muchfeedback on on how to make that
video better, but if you justopen your eyes and you can see
everything and hear everything,it makes it, you know, 10 times
more powerful.
and so, so that's where thistool really comes in is for
marketers that if you're, you'recreating any sort of video
content, you now just have this,this This model that deeply
(14:50):
understands, what's happeningand can give feedback, to how to
make that, that content better.
And obviously we've been pretty,pretty high level, but, you
know, I'm, I'm more than happyto dive into a lot of different
examples of like the specificbusiness use cases that, that
you can use to fully, get themost out of this tool.
Isar Meitis (15:08):
Yeah.
So again, the bottom line is itis the, the only tool that I
know of.
And again, you do that canactually see the video, the
visuals of what's happening.
So if you are either trying toanalyze why some video is
working and becoming viral, ifyou have, 10 of them.
So take 10 videos that wentviral on whatever platform.
It doesn't matter whether it'sLinkedIn or TikTok or Instagram
(15:29):
or wherever and say, okay,analyze why these went viral and
it can give you cues on what'shappening beyond what was said
in the transcript of actuallythe structure of the video, the
visuals of the video, the colorsof the video, the text in it,
like all of these things.
This is the only tool period.
That can do this right now.
So let's really look into a fewuse cases, how you've used this,
(15:51):
in, in an effective way forbusiness solutions.
Jimmy Slagle (15:55):
Yeah.
So right away, like, let's justsay that, you know, like I like
to make content on social media,and a big part of making
effective content is, ishonestly coming down to, you
know, essentially trying to getpeople to watch a video for as
long as possible.
The algorithm is going to rewardthat.
So, you can take a video thatyou created.
and like a very simple prompt,and I mean, in this case,
there's no system instructions.
(16:15):
If I were to like truly usethis, I'd probably add a lot
more to help it know, butnonetheless, like a simple
example is analyze this videoand identify the main viewer
drop off points.
Give me specific timestamps andsuggest three changes to improve
retention.
So again, it will analyze thatvideo.
It looks at okay, maybe theseare the areas where people are
going to drop off and then givesyou suggestions on how to make
that video or that potentialdrop off point, more engaging.
(16:38):
And so I think this is a greatway to just, again, have a
second, set of eyes.
Look at the video that'scompletely unbiased to tell you
This is where it's probablygoing to start to lose, lose
that attention.
So, this is one of the use casesthat I love another one.
So this was a very similar, likeagain, describe the hook of this
video in extreme detail, is itstrong enough to stop the
(17:00):
scroll, you know, beyond justretention, like the hook, the
first three seconds has to be,effective.
And so this is going to diveinto the hook evaluation.
It will then look at thestrength of the hook, and, and,
you know, it will tell you if itthinks it does have a good
chance of, getting people tostop scrolling and, and actually
watch, another really cool one,that was just mentioned is you
can upload multiple video files.
So it's not just one.
(17:20):
So let's just say you have, youhave four or five ads, and, and
you want to compare themessaging and, and ask like,
which one is most clear andcompelling.
for your audience and why.
So it will then again, analyzeall the different videos before
that you uploaded.
And I think, you know, sometimesit does bug out like, like this
is, this is very much so, amodel where if you give it too
much data, there will be somelike internal errors that can
(17:43):
happen.
So I like to kind of keep it atmaybe a couple minutes, for
video content.
But in theory, I think you'reable to upload videos that are
an hour long.
I haven't tried to do somethingthat long.
I did not know
Isar Meitis (17:52):
that and I've never tried something that long
either.
Jimmy Slagle (17:55):
Yeah, yeah, yeah.
So, so that, that might, thatmight be a little tough for it,
but short form, you know,vertical videos, ads, any of
that stuff.
you know, I usually like to dolike five to six, to just ask
about this, this analysis.
And it will give you the, giveyou the rundown of, let's
analyze the ads.
It will go through and do that.
And then we'll give you therecommendation.
So it thinks the fourth ad isthe clearest.
Uses direct, concise language,bullet points, and, and yeah,
(18:17):
the, the negative openingapproach creates intrigue.
So it's like good at being ableto analyze these patterns of
what, what is, effectivemessaging.
The other thing too, evenoutside of, you know, any trend
analysis, is, is you can justuse it for quality control.
Let's say you're making videosand you use an editor or you use
a designer or whatever.
You can have this be theproofreader.
(18:37):
so instead of you having to goand make sure all the captions
are, are correct, and there's nogrammar errors or anything, you
just throw it in here, ask it toproofread, the ad for any typos,
grammar errors, inconsistenciesin tone, any of that, and you
can have it just provide thosespecific corrections.
So, four quick examples.
Don't worry, there, there's manymore that we can go down, but
just to, just to get people's,creative juices flowing.
Isar Meitis (18:58):
No, this is absolutely brilliant.
I want to connect some of thedots that we said, because I
think there's a, there's abroader flow that I'm sure
you're doing based on your 75page, prompt.
So let's say you don't know howto create ads at all.
And you created a few videos andyou posted them wherever you
want to post them.
Doesn't matter where, and youget crickets and you're like,
well, I'm, I'm trying to dothis.
The best that I can.
I'm trying to see what otherpeople I'm saying.
(19:20):
I took a course on YouTube.
Like I followed some people'sguidance, nothing is happening.
How can I make this better?
Well, now you have a videoproduction consultant for almost
free.
And I'm going to touch on thenumbers in a second, because
there's a question on LinkedIn,but almost free.
And you can take multiple videosthat are successful, that you're
(19:41):
seeing that are getting a lot ofengagement, upload them together
and give it a prom saying, Hey,you're an expert in analyzing
video content on this and thatplatform.
I want you to compare these, tocompare these all together and
figure out why they're working,what could be the reasons and
create a best practices prompton what should be the best way
to create a video that can beviral and go blah, blah, blah,
blah, blah.
(20:01):
And then it will create it foryou.
And then you can use that as afilter for your next production,
right?
So now you have best practicesthat was created based on actual
success in your niche, in yourplatform, to your target
audience.
So everything is aligned withwhat you need to create on, on
your own.
And then you can upload yourvideo and say, okay, now analyze
(20:23):
it based on this.
And it's going to tell you whatyou probably need to change, as
far as, okay, you need astronger hook.
You need a faster transitionfrom this to that.
You need more, variation on thescreen because people get bored
after 1.
5 seconds because we createdmadness in this day era that
people like, Oh, it's the samething for the five seconds.
Like it's so boring.
and, and, you know, so patternvariation and stuff like that,
(20:45):
but it's going to tell you you.
All these things and it's goingto help you create better and
better videos that can bealigned to the other videos that
are more successful Aspects youcan't even think about, right?
So it will give you a lot ofstuff.
so really, really great stuff.
There was a question, onLinkedIn from Stan on, is there
a way to estimate the cost ofrunning the queries?
(21:07):
So, yeah, so, all these toolsagain, run this particular tool
runs on tokens.
These are tokens from, dependingon which model you pick, but in
general, they're very cheap.
So as an example, Gemini 1.
5 flash costs, 0.
0.
3.
75.
So three and a half, 3.
75 cents for every milliontokens of input.
(21:28):
So whatever you're writing inthe videos you're uploading and
15 cents for every milliontokens coming out.
So that gives you an idea.
obviously.
Two will be a little moreexpensive, but probably not by a
lot.
So it's completely negligiblebased on the value that you're
going to get from that.
So you can probably run a wholeweek off this thing for 5,
(21:50):
right.
or, or probably less.
And so it's, it's really, reallycheap.
And yet.
It's different than using Geminibecause it is pay per usage
versus pay per month kind ofthing.
any other things you want tomention specifically about the
video capability or you want tomove and talk a little bit about
the streamlined real timecapabilities?
Jimmy Slagle (22:10):
Yeah, the only other thing I'd say is, like, I
put together a database of a lotof prompts that, you can use.
This is more, maybe more tiedtowards advertising, but,
nonetheless, like, you couldeasily replicate a lot of these
prompts for, organic socialgoing viral and all that.
So, I don't know what the bestway would be to, to share this
with, with the people that, werehere, but.
I I'm more than happy to givethis out.
(22:32):
So, so people have a good kindof prompt library that they can,
that they can start with anduse, to just,
Isar Meitis (22:38):
yeah, that is awesome.
Just put it in the chat.
I'll move it to the other chatas well.
And, because I'll have the linkin the chat.
I will post it in the show notesas well.
So those of you listening afterthe fact, first of all, what the
hell are you doing listeningafter the fact?
We do this every Thursday atnoon.
You can come in, you can askyour questions and you can get
the information sooner.
but if you are not available tojoin us on Thursday noon every
(22:58):
week, then it will be in theshow notes and you can get to
this thing.
So thank you.
Thank you for sharing that.
Jimmy Slagle (23:04):
I would say, too, that the one other thing that I
think is worth showing about,Google AI Studio is, it, because
it's the developer platform, youcan start to do some pretty cool
things with, within theautomation and workflows
element.
So, in this case, let's just saythat you wanted to get a summary
of the ads that, the Ridge, ifyou haven't heard of them, it's
like a slim fitting wallet,Yeah, something like that.
(23:25):
and, and so you could create anautomation.
This is a tool called gum loop.
It's a no code automationworkflow builder.
And just like quickly, like allyou have to do, if you build one
of these out is upload a URL.
you would have, an agent scrapeto find the, the Facebook, ad
library account.
it can then, within Facebook,you can select video.
(23:45):
And it can then analyze.
In this case, I just said, likethe 10, like 10 ads to analyze.
and then you can create a prompton what exactly you want to look
for within those video files fora company like that.
and then you can.
Get a whole breakdown of thatwithout having to do a single
thing, into a Google doc ornotion or wherever you want this
(24:06):
to be.
And this is a, you know, 22 pagedocument that breaks down all of
the, the, you know, 10 ads thatthis company, had used.
So there's, there's going to bea lot of cool use cases that
will make it easier for people,to be able to do cool things and
kind of plug and play intodifferent, tools or data points.
But, this is at least one otherthing.
Start to see like, Oh, this isgoing to move really quick from
you just needing to have a oneshot prompt to, full fledged,
(24:29):
like AI workflows.
Isar Meitis (24:32):
And, and again, just not to confuse people that
was not Google AI studio.
That's a different tool, butit's just showing some of the
capabilities that AI is becomingmore and more capable as far as
stringing things together inmultiple steps, in a tool that
can bring together differentthings, including videos and
this and scraping and.
And the analysis and so on, andstuff like that.
(24:53):
there's a question, fromDanielle is the feature of any
Gemini account or only paidaccounts.
So this is so Google AI studiois not connected to you to
Gemini or your accountLiterally, all you have to do is
sign up and give them yourcredit card so you can use
tokens on the API And whetheryou have a pro account or
whatever other account Geminimakes absolutely no difference
(25:14):
You Can not even ever use Geminiand you can use, and you can
still use the Google AI studiofeatures.
So next capability of Google AIStudio, right?
Let's jump into that.
Jimmy Slagle (25:25):
Yeah.
Yeah.
So this was, this was released,a little over a month ago.
and, and there's now threedifferent things that you can
do.
The first one is talk withGemini.
Just to
Isar Meitis (25:33):
go back, we said in the beginning, there's a menu on
the left.
There's multiple things tochoose from.
The first thing we played withis just a prompt.
The second thing is calledstream real time.
And this is what we're talkingabout right now.
Jimmy Slagle (25:44):
Yep.
Yep.
So there's three different,capabilities within this stream
real time.
the first one is talk to Gemini.
So that's similar to like chatGBT is advanced voice mode,
where you can just have aconversation with an AI agent.
The next one is you can actuallyshow your camera, which we'll
see how this works on zoom, butnonetheless, if it, if it
doesn't work too well, but see,you can see that it's seeing
everything in the chat that'sgoing on live and it's listening
(26:08):
to what I'm saying.
So if I.
Take a second.
It should start to respond and,answer, you know, kind of what
I've been talking about.
Normally it gives the, the, thevoice overview too.
Normally it
Isar Meitis (26:18):
speaks and it actually generated it, but it's
not saying it, which is kind ofweird.
Jimmy Slagle (26:22):
Yeah.
But nonetheless, if you're noton a zoom call, it will actually
talk to you as it's, as you'reshowing it, whatever you want in
the camera.
and if you have your phone, likethis is super powerful because
then if you're out at thegrocery store or.
If you're, you know, on a walkand you see a cool plant or
whatever you want, like you cannow just be like, Oh, Hey, like,
what is this plant?
Or, Or if you're shopping
Isar Meitis (26:43):
or if you're shopping and you want to know
how much something costs onlineversus in the store that you're
in or stuff like that.
It, it knows all of that becauseit's Google, like it's got
access to all that extrainformation, but it can see what
your camera sees.
It can translate signs.
If you're in a foreign countryand you don't know how to read
the street signs or the menu atthe restaurant, you can
literally show it the menu andask it, and he will tell you
(27:03):
what things are and so on.
It's absolutely magical, thatyou can do that.
Actually share your camera.
Jimmy Slagle (27:09):
Yeah.
And even like, you know,business use cases, if you're in
an office and you, you like towhiteboard, you could show,
Gemini, the whiteboard thatyou're working on or, you know,
something, that's, that's morelike physically present than,
than, you know, just ascreenshot or, or anything along
those lines.
So, yeah, this is a really,we're, we're very early in, in
the real time AI, but.
I think that this is going to bethe big, big wave in 2025 of
(27:33):
apps that you'll start to seethat are coming out, utilizing
this technology.
Isar Meitis (27:37):
So there's a question.
Is that similar to Chachapiti?
Yes.
So if you have your, theChachapiti app on the phone, you
can do the same thing withChachapiti on the phone.
You can, you still cannot dothis with ChachuPT on the
computer for whatever reason,like even if you have the paid
version and the desktop app, I'ma, I'm a Mac user, so I don't
know, like on the PC, they stilldon't have that functionality,
(27:59):
even though I have it on myphone.
So there's, in general, there'sa huge variance between
different.
ins and outs of chat GPT, whichI don't understand, like stuff
that you can do on the regularchat, you cannot do on custom
GPTs and stuff you can do inthis version, you cannot do in
that version, like there's allthese different variations.
So one of the benefits of doingthis on Gemini is that you can
do it on your computer and notjust on your phone.
(28:21):
But yes, in functionality, it'svery similar to the
functionality that the chat GPTapp provides on the mobile app.
I want to talk about actuallythe third one.
So the third, you said thatthere are three options.
Option number one is just avoice option.
Number two is show Gemini,basically open the camera and
show you what you're seeing.
And option number three is shareyour screen.
(28:42):
And I, first of all, I'mwondering if you have any use
cases that you use this for, andthen I can share something that
I literally tried this week,which was really interesting.
Jimmy Slagle (28:49):
Yeah, there, there's a couple different
things that I've been using thisfor.
So the first one is as I'mbuilding out some of these.
different, like AI workflows,you can start to ask it like,
Oh, how can I like, like, youknow, looking through these
different nodes is kind of whatthey, they're called, like,
which one of these would be bestif I wanted to do X.
And so in this case, if I wantedto, you know, scrape Reddit to
(29:11):
find the top threads about, Whatpeople are talking about about a
I, it can, it can look ateverything that's happening on
my screen and recommend, Oh, youknow, you should go to the
Reddit analysis.
and, and then connect that tothis node.
And so it's kind of just likeyour helper as you're
navigating, any of these morecomplex websites.
So like, that's one that I'vereally been using it for.
(29:32):
even like, feedback on videos,like I'll sometimes show, like
pull up my screen so then Idon't have to download it and
export it, but it can just bereal time of like, Hey, what do
you think about this video?
I'll play it.
It then watches it and can giveme feedback in real time instead
of again, those one or twosteps.
So Those are two quick examplesof, of what I've been using it
for, and again, as you can seehere, like it's, it's seeing
(29:54):
what is happening on my screen,again, normally it would be
talking, and giving feedback,but, but yeah, so that's,
that's, that's one of the, the,a couple of these cases that
I've been seeing with this.
Isar Meitis (30:04):
Yeah.
So very similar to your firstuse case, I was working on a
really complex automation onmake, this weekend for one of my
clients and I got stuck onsomething and I couldn't figure
it out.
I was banging my head againstthe screen, the wall, the desk,
like everything.
And I watched like six YouTubevideos and I just couldn't
figure it out.
Like I couldn't find why it'snot doing what I think it needs
to do.
(30:24):
And they're like, well, wait aminute.
And then I opened this and Ishowed it to desktop and I'm
like, okay, here's what I'mtrying to do, here's where it's
failing, helping troubleshootthis.
And then it, it knows make, andit knows what I'm doing and it
understands parameters and Iliterally had a consultant for
free helping me solve a makeproblem that I couldn't solve.
(30:45):
And it ended up being somethingreally, really stupid, but.
By the way, it wasn't able tohelp me solve it, but it helped
me find the problem.
And then I solved it in fiveminutes, but I just couldn't
find what the hell was wrong,before.
And so think about the future ofany software.
Of any computer game, of anyprogramming problem, of anything
(31:08):
we're going to have.
First of all, on the generaltools, they will know all the
most common tools that we use,right?
So it will know Make, it willknow Microsoft Outlook, it will
know how to set up things inAdobe, like it will know all
that stuff because it alreadydoes.
So you can literally show itwhat you're doing and it can
give you tips and guide youthrough the steps, whether
you're a total beginner or anadvanced users.
(31:30):
Advanced users are getting stuckon something very specific,
which is incredible.
The other thing, companies willstart training it on their own
tools.
That will be a button.
So instead of the chat with usright now, but that is between
you and me, we're worthless 95percent of the time, you will
click a button and there's goingto be an AI.
That's going to look on thescreen with you on what you're
(31:51):
actually doing in your setupwith your problem.
And we'll guide you through howto do this.
And this is.
Not the future.
This is right now.
It's just a matter of time untilcompanies actually add that
functionality to the, to thetools.
But if any software company thatdoesn't do it is a complete
idiot, because it literally is acomplete game.
It talks to you as if it'ssomebody standing behind your
desk.
Shoulder and showing you what todo.
(32:13):
So if you haven't used this,even by the way, people are
using it for gaming right now.
So I showed it to my kids andthey started like, Oh my God, I
can use this for this.
And I can use it for that.
So they, they now have aconsultant playing computer
games as they're playingcomputer games.
It's just amazing.
I will say one thing, whiledoing my process, it got stuck
three times in the process.
And, and then he doesn'tremember because it.
It's now a new session.
(32:33):
So I had to start from thebeginning.
So that was a little annoying,but like you said, this is very,
very early on in this world.
And the next thing, by the way,To me, a little scary, but
there's no going back.
You know, the Pandora box isalready open is all the
wearables, right?
All these glasses that knows howto connect to this.
And then they see everything yousee, whether you're outside or
(32:53):
your computer screen or in a baror wherever you are, and it can
assist you and help you andguide you and provide you
information about.
Everything that's happening andthat's, again, on from a privacy
perspective, I'm not very happyabout it, but from a reality
perspective, there's no goingback like a few years from now,
almost everybody will wear somekind of a wearable that will be
connected to an AI into yourBluetooth headphones and we'll
(33:16):
be able to provide you anyinformation you want about
anything that it sees, which iseverything that you see.
Jimmy Slagle (33:22):
Yeah, exactly.
I mean, again, it's like this islike when chat GBT was released.
And immediately, you're going tohave tons of people that are
going to figure out specific usecases for this tool.
And that's what I think 2025 isgoing to be people are going to
be exploring it in wearables inin all the different vertical
like I'm sure we'll have one forCanva that's just been trained
on, you know, Everything that'spossible within Canva and
(33:42):
specifically to like gooddesign, video editing, like all
those, we're going to start tosee where all you have to do is
share your screen and you'regoing to have an expert because
again, you can give this custominstructions as well.
so you're going to start to seea lot of these, these kind of,
wrappers, if you will, of this,technology come out.
Isar Meitis (33:59):
Jimmy, this was a fantastic session.
Like we really touched on stuff.
I think most people, even peoplewho are like advanced users
don't have a clue that exists.
So I'm very grateful for youdoing this and taking us down
that path.
If people want to follow you,learn from you, work with you,
what are the best ways to dothat?
Jimmy Slagle (34:17):
Yeah, honestly, LinkedIn is probably where I'm
most active.
So just Jimmy Sligo, you can,you can connect with me and
start a conversation there.
I do like to make content on TikTOK as well.
fortunately it's, it's, it'sback.
And so it's back.
Yes.
It's back for like three
Isar Meitis (34:31):
seconds.
And then, you know, the newpresident said, well, whatever.
And yeah,
Jimmy Slagle (34:35):
exactly, exactly.
So, yeah, connect with me onsocials again.
Like, like really what I do isthose AI workflows, that I
quickly we help brands create alot of those and, and, and
figure out just how to do thingsbetter, faster, and more
efficient.
So, always down to chat how to,how to make, your company be
able to operate, especially themarketing team in the most
efficient way possible.
Isar Meitis (34:57):
This
Jimmy Slagle (34:57):
is awesome.
Isar Meitis (34:58):
Thank you so much.
Everybody who joined us again,thank you for being here.
I know you have other stuff todo on Thursday afternoons.
but if you, again, if you'relooking for more structured
training, the course is comingup on February 17th, you can
find it on my LinkedIn or mywebsite multiplied on AI, or
just find me and ask me if youhave any questions.
and that's it.
(35:18):
Thanks again for everybody forjoining live.
Thank you so much, Jimmy.
This was absolutely fantastic.
Jimmy Slagle (35:23):
Yes, you're welcome.
Thank you again for having meon.
Hello and welcome to another live episode of the
(00:04):
leveraging AI podcast, thepodcast that shares practical
ethical ways to leverage AI, toimprove efficiency, grow your
business and advance yourcareer.
This is Isar Matis, your host,and I have a really exciting
show for you today.
I will start with.
A really piece of exciting newsfor myself personally.
And for you as listeners to theshow, we just crossed 200, 000
(00:24):
downloads off this podcast.
Now, when I started this podcastabout a year and a half ago, I
never thought that number wouldbe possible, but here we are.
So first of all, thank.
All of you for being loyallisteners.
And if it's your first time,welcome aboard.
You're joining a very largegroup of people, that drove this
number.
So I'm personally very excitedthat we're here to me.
The number is less exciting thanthe importance of what's we're
(00:46):
actually doing, right?
It's, it's very obvious to.
Everybody out there today,including recent recent research
and surveys by the worldeconomic forum that surveyed
more than a few thousands ofCEOs, that the number one thing
that is stopping AI success incompanies right now is AI skills
and knowledge.
That's the number one reasonthat AI is not growing and not
(01:06):
helping businesses grow as fastas it can.
And listening to this podcast isjust one way to do this.
For free.
And while you're walking yourdog, doing your dishes, driving
your car, mowing the yard,whatever it is that you're
doing, that you're doing this.
So as I mentioned, first of all,thank you so much.
I couldn't be more humbled andthankful for all of you
listening to the show anddriving you to these crazy
(01:27):
numbers, but it wouldn't bepossible without us providing
real value.
And the value today happens tobe talking about a specific AI
tool.
So there are many AI tools.
out there, right?
There's people use ChachiPT andGemini and Claude and Midjourney
and Lama and the, you know,every time there's like a new
contender, a newcomer likeDeepSeek now with DeepSeek R1,
(01:50):
which are amazing.
And there's always new tools,but there's always these tools
that are kind of like hiding inclear sight.
And yet most people are ignoringthem.
Despite the fact they provideincredible value and they are
free.
And so one of these tools isGoogle AI studio.
And our guest today is an AIstudio ninja.
(02:13):
So Jimmy Slagol has been in theAI world for Longer than most of
us.
So he actually started in playwith AI and working with
companies, helping themimplement AI before Chachapiti
came out.
So a year before Chachapiti cameout, which is more than most of
us can say, and he's beenfocused on that universe for The
(02:33):
last three years.
And that's what he's been doing.
And one of his favorite toolsthat again, most people don't
know about, or at least don'tuse as frequently as he does is
Google AI studio.
And he's going to share with ustoday on some tricks that you
can do in AI studio that areextremely powerful.
So I'm personally very excited.
I love Google AI studio.
I use it a lot.
So I'm sure we will geek about afew of these functionalities,
(02:55):
just So I'm very excited to haveJimmy as a guest of the show.
Jimmy, welcome to Leveraging AI.
Jimmy Slagle (03:46):
Thank you.
I appreciate it, man.
Yeah, I'm excited for this.
And yeah, to date myself, on, onwhen I started to use AI.
So I'm from Fargo, North Dakota,which is like a, you know,
normally it's usually a littlebehind the curve in terms of
technology.
but I started to make a lot ofcontent around AI in that like
early 2022 era.
And, one of like the local news,kind of journalist companies
(04:08):
reached out and asked if I'd bewilling to like talk about, you
know, my five favorite AI tools.
And, and it was so early thatlike ChatGBT wasn't even one of
those on the list because ithadn't come out yet.
so, you know, I'm sure that thatlist that I wrote is probably
completely irrelevant today.
But, but yeah, it's just, justfunny that, that, there was the
list out there that wouldn'tinclude ChatGBT of the best
(04:29):
tools.
Isar Meitis (04:31):
Yeah.
Yeah.
It was the days that everybodywas still doing like the chat
writers, right?
It was, I'm trying to thinkwhich was the one that got the
first, and then they changed thename like three times after
that, that was like the big blogwriting AI tool.
Yeah.
Yeah.
Jimmy Slagle (04:44):
I know like Jasper AI.
Jasper.
Isar Meitis (04:46):
Jasper.
Yeah.
But, but, but they had adifferent name back then.
Jimmy Slagle (04:49):
Yep.
Yeah, it was Jarvis.
Cause then I think exactly,
Isar Meitis (04:53):
there we go.
So yeah, that, that was probablyone of the items on the list.
I'm sure it
Jimmy Slagle (04:57):
was, I'm sure it was, but, but yeah.
Isar Meitis (05:01):
So yeah, for all of you who are joining us live,
whether you are with us live onLinkedIn or live on zoom.
So first of all, thank you.
And thank you for joining us forthis live session.
If you have any questions, Anyquestions, just post them in the
chat.
If you have anything else to addor comment, please do that.
Please start by introducingyourself.
Just say on LinkedIn, where areyou from?
What are you doing?
(05:21):
What can tools you look haveused a Google AI studio before.
So we kind of know where you arein your journey.
And before we get started, onelast, piece of.
Information or, or update, weare launching another cohort of
the AI business transformationcourse.
So I've been teaching thiscourse personally since April of
2023.
(05:42):
So not as early as Jimmy hasbeen in AI, but.
Very short into the AI game andhundreds or maybe even thousands
of people have taken this courseand are transforming their
businesses based on theknowledge that we provide.
It's four sessions, two hourseach.
So in a month you can be in acompletely different state of
knowledge and understanding bothin means of hands on skills on
(06:03):
multiple use cases, as well ason clear, roadmap and blueprint
on how to actually implement AIsuccessfully in your business.
If that's something you'reinterested in doing, there's
going to be a link in the shownotes.
We're going to drop the linkright now in the live session.
If you're here with us live, andif you're listening off this
podcast, you can use promo codeleveraging AI 100.
To get a hundred dollars off theprice of the course, the course
(06:25):
is filling up pretty quickly.
We just literally open it andthere's already like 10 people
signed up and we don't allowmore than 20 people at every
single cohort.
So if you're interested ingetting some proper AI training
for 2025 and you should comejoin us and act quickly.
But with that, Jimmy, let's talka little bit about Google AI
studio.
And I want you to start bytelling people Like, what is it?
(06:46):
Because it's not a chatbot,right?
It's like, it's something alittle different.
So let's explain to people whatis Google AI Studio, and then we
can dive into specific usecases.
Jimmy Slagle (06:54):
Cool.
Yeah.
Are you, are you shit?
cool.
If I share my screen and justshare.
Yeah,
Isar Meitis (06:58):
absolutely.
Definitely share your screen.
For those of you who arelistening to the podcast while
you're driving, don't stop andtry to look at the screen, but
if you want, this will beavailable on our YouTube channel
as well, but we will explaineverything that's on the screen.
So if you're just listening,you're not going to miss
anything important.
Jimmy Slagle (07:13):
Awesome.
Yeah.
So what's interesting aboutGoogle AI Studio is this is
technically for developers.
It was built for developers.
So people that wanted to kind ofuse this to then go and create
their own applications that'spowered by Google, this is the
interface that they use.
And so that's what makes it alittle different than the
Geminis of the world, the ChatGPTs of the world, or the Clods
(07:33):
of the world.
So this looks a little more,intimidating than.
what you're used to with, withchat GPT.
But what I love about Google AIstudio, and, and what we're
really going to dive into todayis it is the only, large
language model, at least that,that I'm aware of that's able to
analyze an actual video file.
And so what I mean by that is Ican upload an actual video.
(07:56):
and so if I take this and thisis like an ad, for example, I
can use that as the, data sourceinstead of it just being like an
image or a PDF or anything alongthose lines, it will take a
frame per second of this videoso you can now essentially use
chat GPT for any sort of video,marketing or use cases that you
(08:19):
have, which again is prettyendless.
So.
For example, I can say righthere, like, describe this ad in
extreme visual detail, and itwill be able to essentially do
that analysis, and, and give youexactly what is showing up in
this ad, which there's tons ofuse cases that we can dive into.
(08:39):
This is a really funny if youhaven't heard of the perfect
gene.
They're very like tongue incheek marketing.
but, but, but yeah, so it's,it's, it's a really just
fascinating tool and that's whatfirst got me interested to be
like, oh, wow, I never knew thatthere was a, there was a, large
language model that, That coulddo something like this.
So from the beginning, that's,that's kind of what, what makes
it a little different from, fromthose models.
(08:59):
we can, we can later get intosome of the other cool stuff
that Google is coming out, with,with like the real time, but for
now we'll focus just on thevideo analysis.
Isar Meitis (09:07):
Okay.
So before we dive into more ofthe video analysis and how it
can be used, like what ouractual business use case,
there's a question from Gwyn,and I hope I'm pronouncing it
correctly.
and the question is what is thepurpose of Google AI studio?
There are many tools out there.
So just want to Level set as weget into this.
So I will say you, you mentionedone aspect that it was initially
(09:28):
built for developers, but thereality is many of the tools
that are in it right now arevery much common person
friendly.
You don't need to write anycode.
You don't need to do anything.
And when you open Google AIstudio and to get to it, it's
just, AI studio.
google.
com.
If I remember correctly.
And if you go there, there'slike a menu on the side.
Of different tools that you canpick from.
(09:49):
So it's actually a lot more thanjust one tool.
You, there's multiple thingsthat you can do with it.
the first one is more or lesslike a chat bot.
And the biggest differencebetween that and just using
Gemini is a, it has someadditional functionality than
you do before B, you have morecontrol on how the model
actually works.
So you can control thetemperature, which controls how,
(10:11):
kind of like.
aligned with reality it isversus how will it be creative.
That's the best way I candescribe it.
And you can control some otherfunctionality and you're
actually running it against theAPI, meaning you need to give
them your credit card and you'rerunning against tokens, which
has pros and cons.
One of the biggest Rose is theyalways release their newest
(10:32):
models on Google AI studio firstbefore it's available on Gemini.
So right now, if you're runningGemini, you can get 1.
5 pro.
And if you're running here, youcan get Gemini two and Gemini
two flash are available to run.
The biggest benefit of Geminitwo versus 1.
5 is it has an even Biggercontext window, and you can
(10:54):
upload out to 2 million tokens.
And for those of you who don'tknow what that means, that's
about 1.
5 million words, which is by farthe largest contest window you
have in any tool out there.
So if you have a huge amount ofdata that you want to run
through a tool and ask questionsabout it, the next best one is
200, 000 tokens on, Claude, andthis is 2 million, that's 10 X.
(11:17):
More data that you can upload,and they're claiming 97.
7 percent accuracy whenanswering questions about the
data that you upload 97.
7 percent accuracy is almost asgood as my as my wife, and it's
way better than the averageperson, and so it's a great way
if you have a huge amount ofdata.
That's one reason to go to thistool.
(11:38):
What other stuff I'm, I'minterested to hear from you,
Jimmy, if you have other stuffthat you want to say about this
and how this is different thanjust going to any other tool.
Jimmy Slagle (11:47):
Yeah, no, I mean, I think, I think you hit it,
head on, especially the contextwindow.
Like I, I, I don't think we canunderstate how important that
is.
For example, like one of thecool things, cause I think we
all know, The prompts that yougive these models really are
what make them effective.
And so for like an AI creativestrategist, I mean, we, we have
(12:07):
taken almost everything that wehave learned.
And so, for those of you thatare tuning in, not visually,
this is if you paste it into,into 75 page, document, just
teaching this, this model,everything that it needs to know
about creative strategy.
So.
What the hooks are, what thebest ad concepts are, and just
everything else to make surethat it knows, exactly the
(12:28):
language and, best use cases andpractices in this space.
and so that's what makes this,this a, a just very, very
powerful, model.
Isar Meitis (12:37):
So again, to explain to people, this is a 75
page prompt.
If you load this to any othertool, you already take in either
all or a big chunk off thecontext window that is usable in
a single chat.
And here it's a drop in thebucket because there's 2 million
tokens and you've just taken,let's say 50, 000.
(12:58):
It's still.
50, 000 out of 2 million versus50, 000 out of 125, 000.
If you're running a chat, CPT asan example.
And so it's, it's a, it's ahuge, huge benefit.
And as I mentioned, the otherthing is if you're doing
research on variety of data andyou're loading multiple data
from multiple sources, and I'veuploaded, you know, Excel files
with over a million rows intothis, which will not open in
(13:21):
most tools that, you know, andyou can still ask questions
about it and get accurateinformation.
So it's very, very powerful fromthat perspective.
Let's jump back to the videoanalysis, right?
So there are tools out therethat can give me a summary of a
video, but that's verydifferent, right?
So let's explain how is thatdifferent?
And then if you give examples ofwhat use cases it can be used
for, that will be very helpful.
Okay.
Jimmy Slagle (13:42):
Yeah, for sure.
So right now, what youtechnically can do from a video
is just get the transcript.
it, the, they're outside ofGoogle AI studio.
there's no way to be able toactually get the visuals that
are happening within that video.
and so when you can getsummaries from videos, normally
that's just looking YouTube oranywhere else and just telling
you, Hey, here's everythingthat, that happens.
(14:03):
But when you get into the worldof marketing, whether that's
creating organic content.
or creating ads, you need an AImodel that's able to actually
see what's happening in thevideo to be able to be your,
your co creator, if you will.
you know, I like to, I like togive people the example all the
time.
If I showed you, let's just sayit's, you know, a video on
TikTok and you had to close youreyes and just hear what's
(14:27):
happening and, and, you know,essentially get the script of
it.
You can only give so muchfeedback on on how to make that
video better, but if you justopen your eyes and you can see
everything and hear everything,it makes it, you know, 10 times
more powerful.
and so, so that's where thistool really comes in is for
marketers that if you're, you'recreating any sort of video
content, you now just have this,this This model that deeply
(14:50):
understands, what's happeningand can give feedback, to how to
make that, that content better.
And obviously we've been pretty,pretty high level, but, you
know, I'm, I'm more than happyto dive into a lot of different
examples of like the specificbusiness use cases that, that
you can use to fully, get themost out of this tool.
Isar Meitis (15:08):
Yeah.
So again, the bottom line is itis the, the only tool that I
know of.
And again, you do that canactually see the video, the
visuals of what's happening.
So if you are either trying toanalyze why some video is
working and becoming viral, ifyou have, 10 of them.
So take 10 videos that wentviral on whatever platform.
It doesn't matter whether it'sLinkedIn or TikTok or Instagram
(15:29):
or wherever and say, okay,analyze why these went viral and
it can give you cues on what'shappening beyond what was said
in the transcript of actuallythe structure of the video, the
visuals of the video, the colorsof the video, the text in it,
like all of these things.
This is the only tool period.
That can do this right now.
So let's really look into a fewuse cases, how you've used this,
(15:51):
in, in an effective way forbusiness solutions.
Jimmy Slagle (15:55):
Yeah.
So right away, like, let's justsay that, you know, like I like
to make content on social media,and a big part of making
effective content is, ishonestly coming down to, you
know, essentially trying to getpeople to watch a video for as
long as possible.
The algorithm is going to rewardthat.
So, you can take a video thatyou created.
and like a very simple prompt,and I mean, in this case,
there's no system instructions.
(16:15):
If I were to like truly usethis, I'd probably add a lot
more to help it know, butnonetheless, like a simple
example is analyze this videoand identify the main viewer
drop off points.
Give me specific timestamps andsuggest three changes to improve
retention.
So again, it will analyze thatvideo.
It looks at okay, maybe theseare the areas where people are
going to drop off and then givesyou suggestions on how to make
that video or that potentialdrop off point, more engaging.
(16:38):
And so I think this is a greatway to just, again, have a
second, set of eyes.
Look at the video that'scompletely unbiased to tell you
This is where it's probablygoing to start to lose, lose
that attention.
So, this is one of the use casesthat I love another one.
So this was a very similar, likeagain, describe the hook of this
video in extreme detail, is itstrong enough to stop the
(17:00):
scroll, you know, beyond justretention, like the hook, the
first three seconds has to be,effective.
And so this is going to diveinto the hook evaluation.
It will then look at thestrength of the hook, and, and,
you know, it will tell you if itthinks it does have a good
chance of, getting people tostop scrolling and, and actually
watch, another really cool one,that was just mentioned is you
can upload multiple video files.
So it's not just one.
(17:20):
So let's just say you have, youhave four or five ads, and, and
you want to compare themessaging and, and ask like,
which one is most clear andcompelling.
for your audience and why.
So it will then again, analyzeall the different videos before
that you uploaded.
And I think, you know, sometimesit does bug out like, like this
is, this is very much so, amodel where if you give it too
much data, there will be somelike internal errors that can
(17:43):
happen.
So I like to kind of keep it atmaybe a couple minutes, for
video content.
But in theory, I think you'reable to upload videos that are
an hour long.
I haven't tried to do somethingthat long.
I did not know
Isar Meitis (17:52):
that and I've never tried something that long
either.
Jimmy Slagle (17:55):
Yeah, yeah, yeah.
So, so that, that might, thatmight be a little tough for it,
but short form, you know,vertical videos, ads, any of
that stuff.
you know, I usually like to dolike five to six, to just ask
about this, this analysis.
And it will give you the, giveyou the rundown of, let's
analyze the ads.
It will go through and do that.
And then we'll give you therecommendation.
So it thinks the fourth ad isthe clearest.
Uses direct, concise language,bullet points, and, and yeah,
(18:17):
the, the negative openingapproach creates intrigue.
So it's like good at being ableto analyze these patterns of
what, what is, effectivemessaging.
The other thing too, evenoutside of, you know, any trend
analysis, is, is you can justuse it for quality control.
Let's say you're making videosand you use an editor or you use
a designer or whatever.
You can have this be theproofreader.
(18:37):
so instead of you having to goand make sure all the captions
are, are correct, and there's nogrammar errors or anything, you
just throw it in here, ask it toproofread, the ad for any typos,
grammar errors, inconsistenciesin tone, any of that, and you
can have it just provide thosespecific corrections.
So, four quick examples.
Don't worry, there, there's manymore that we can go down, but
just to, just to get people's,creative juices flowing.
Isar Meitis (18:58):
No, this is absolutely brilliant.
I want to connect some of thedots that we said, because I
think there's a, there's abroader flow that I'm sure
you're doing based on your 75page, prompt.
So let's say you don't know howto create ads at all.
And you created a few videos andyou posted them wherever you
want to post them.
Doesn't matter where, and youget crickets and you're like,
well, I'm, I'm trying to dothis.
The best that I can.
I'm trying to see what otherpeople I'm saying.
(19:20):
I took a course on YouTube.
Like I followed some people'sguidance, nothing is happening.
How can I make this better?
Well, now you have a videoproduction consultant for almost
free.
And I'm going to touch on thenumbers in a second, because
there's a question on LinkedIn,but almost free.
And you can take multiple videosthat are successful, that you're
(19:41):
seeing that are getting a lot ofengagement, upload them together
and give it a prom saying, Hey,you're an expert in analyzing
video content on this and thatplatform.
I want you to compare these, tocompare these all together and
figure out why they're working,what could be the reasons and
create a best practices prompton what should be the best way
to create a video that can beviral and go blah, blah, blah,
blah, blah.
(20:01):
And then it will create it foryou.
And then you can use that as afilter for your next production,
right?
So now you have best practicesthat was created based on actual
success in your niche, in yourplatform, to your target
audience.
So everything is aligned withwhat you need to create on, on
your own.
And then you can upload yourvideo and say, okay, now analyze
(20:23):
it based on this.
And it's going to tell you whatyou probably need to change, as
far as, okay, you need astronger hook.
You need a faster transitionfrom this to that.
You need more, variation on thescreen because people get bored
after 1.
5 seconds because we createdmadness in this day era that
people like, Oh, it's the samething for the five seconds.
Like it's so boring.
and, and, you know, so patternvariation and stuff like that,
(20:45):
but it's going to tell you you.
All these things and it's goingto help you create better and
better videos that can bealigned to the other videos that
are more successful Aspects youcan't even think about, right?
So it will give you a lot ofstuff.
so really, really great stuff.
There was a question, onLinkedIn from Stan on, is there
a way to estimate the cost ofrunning the queries?
(21:07):
So, yeah, so, all these toolsagain, run this particular tool
runs on tokens.
These are tokens from, dependingon which model you pick, but in
general, they're very cheap.
So as an example, Gemini 1.
5 flash costs, 0.
0.
3.
75.
So three and a half, 3.
75 cents for every milliontokens of input.
(21:28):
So whatever you're writing inthe videos you're uploading and
15 cents for every milliontokens coming out.
So that gives you an idea.
obviously.
Two will be a little moreexpensive, but probably not by a
lot.
So it's completely negligiblebased on the value that you're
going to get from that.
So you can probably run a wholeweek off this thing for 5,
(21:50):
right.
or, or probably less.
And so it's, it's really, reallycheap.
And yet.
It's different than using Geminibecause it is pay per usage
versus pay per month kind ofthing.
any other things you want tomention specifically about the
video capability or you want tomove and talk a little bit about
the streamlined real timecapabilities?
Jimmy Slagle (22:10):
Yeah, the only other thing I'd say is, like, I
put together a database of a lotof prompts that, you can use.
This is more, maybe more tiedtowards advertising, but,
nonetheless, like, you couldeasily replicate a lot of these
prompts for, organic socialgoing viral and all that.
So, I don't know what the bestway would be to, to share this
with, with the people that, werehere, but.
I I'm more than happy to givethis out.
(22:32):
So, so people have a good kindof prompt library that they can,
that they can start with anduse, to just,
Isar Meitis (22:38):
yeah, that is awesome.
Just put it in the chat.
I'll move it to the other chatas well.
And, because I'll have the linkin the chat.
I will post it in the show notesas well.
So those of you listening afterthe fact, first of all, what the
hell are you doing listeningafter the fact?
We do this every Thursday atnoon.
You can come in, you can askyour questions and you can get
the information sooner.
but if you are not available tojoin us on Thursday noon every
(22:58):
week, then it will be in theshow notes and you can get to
this thing.
So thank you.
Thank you for sharing that.
Jimmy Slagle (23:04):
I would say, too, that the one other thing that I
think is worth showing about,Google AI Studio is, it, because
it's the developer platform, youcan start to do some pretty cool
things with, within theautomation and workflows
element.
So, in this case, let's just saythat you wanted to get a summary
of the ads that, the Ridge, ifyou haven't heard of them, it's
like a slim fitting wallet,Yeah, something like that.
(23:25):
and, and so you could create anautomation.
This is a tool called gum loop.
It's a no code automationworkflow builder.
And just like quickly, like allyou have to do, if you build one
of these out is upload a URL.
you would have, an agent scrapeto find the, the Facebook, ad
library account.
it can then, within Facebook,you can select video.
(23:45):
And it can then analyze.
In this case, I just said, likethe 10, like 10 ads to analyze.
and then you can create a prompton what exactly you want to look
for within those video files fora company like that.
and then you can.
Get a whole breakdown of thatwithout having to do a single
thing, into a Google doc ornotion or wherever you want this
(24:06):
to be.
And this is a, you know, 22 pagedocument that breaks down all of
the, the, you know, 10 ads thatthis company, had used.
So there's, there's going to bea lot of cool use cases that
will make it easier for people,to be able to do cool things and
kind of plug and play intodifferent, tools or data points.
But, this is at least one otherthing.
Start to see like, Oh, this isgoing to move really quick from
you just needing to have a oneshot prompt to, full fledged,
(24:29):
like AI workflows.
Isar Meitis (24:32):
And, and again, just not to confuse people that
was not Google AI studio.
That's a different tool, butit's just showing some of the
capabilities that AI is becomingmore and more capable as far as
stringing things together inmultiple steps, in a tool that
can bring together differentthings, including videos and
this and scraping and.
And the analysis and so on, andstuff like that.
(24:53):
there's a question, fromDanielle is the feature of any
Gemini account or only paidaccounts.
So this is so Google AI studiois not connected to you to
Gemini or your accountLiterally, all you have to do is
sign up and give them yourcredit card so you can use
tokens on the API And whetheryou have a pro account or
whatever other account Geminimakes absolutely no difference
(25:14):
You Can not even ever use Geminiand you can use, and you can
still use the Google AI studiofeatures.
So next capability of Google AIStudio, right?
Let's jump into that.
Jimmy Slagle (25:25):
Yeah.
Yeah.
So this was, this was released,a little over a month ago.
and, and there's now threedifferent things that you can
do.
The first one is talk withGemini.
Just to
Isar Meitis (25:33):
go back, we said in the beginning, there's a menu on
the left.
There's multiple things tochoose from.
The first thing we played withis just a prompt.
The second thing is calledstream real time.
And this is what we're talkingabout right now.
Jimmy Slagle (25:44):
Yep.
Yep.
So there's three different,capabilities within this stream
real time.
the first one is talk to Gemini.
So that's similar to like chatGBT is advanced voice mode,
where you can just have aconversation with an AI agent.
The next one is you can actuallyshow your camera, which we'll
see how this works on zoom, butnonetheless, if it, if it
doesn't work too well, but see,you can see that it's seeing
everything in the chat that'sgoing on live and it's listening
(26:08):
to what I'm saying.
So if I.
Take a second.
It should start to respond and,answer, you know, kind of what
I've been talking about.
Normally it gives the, the, thevoice overview too.
Normally it
Isar Meitis (26:18):
speaks and it actually generated it, but it's
not saying it, which is kind ofweird.
Jimmy Slagle (26:22):
Yeah.
But nonetheless, if you're noton a zoom call, it will actually
talk to you as it's, as you'reshowing it, whatever you want in
the camera.
and if you have your phone, likethis is super powerful because
then if you're out at thegrocery store or.
If you're, you know, on a walkand you see a cool plant or
whatever you want, like you cannow just be like, Oh, Hey, like,
what is this plant?
Or, Or if you're shopping
Isar Meitis (26:43):
or if you're shopping and you want to know
how much something costs onlineversus in the store that you're
in or stuff like that.
It, it knows all of that becauseit's Google, like it's got
access to all that extrainformation, but it can see what
your camera sees.
It can translate signs.
If you're in a foreign countryand you don't know how to read
the street signs or the menu atthe restaurant, you can
literally show it the menu andask it, and he will tell you
(27:03):
what things are and so on.
It's absolutely magical, thatyou can do that.
Actually share your camera.
Jimmy Slagle (27:09):
Yeah.
And even like, you know,business use cases, if you're in
an office and you, you like towhiteboard, you could show,
Gemini, the whiteboard thatyou're working on or, you know,
something, that's, that's morelike physically present than,
than, you know, just ascreenshot or, or anything along
those lines.
So, yeah, this is a really,we're, we're very early in, in
the real time AI, but.
I think that this is going to bethe big, big wave in 2025 of
(27:33):
apps that you'll start to seethat are coming out, utilizing
this technology.
Isar Meitis (27:37):
So there's a question.
Is that similar to Chachapiti?
Yes.
So if you have your, theChachapiti app on the phone, you
can do the same thing withChachapiti on the phone.
You can, you still cannot dothis with ChachuPT on the
computer for whatever reason,like even if you have the paid
version and the desktop app, I'ma, I'm a Mac user, so I don't
know, like on the PC, they stilldon't have that functionality,
(27:59):
even though I have it on myphone.
So there's, in general, there'sa huge variance between
different.
ins and outs of chat GPT, whichI don't understand, like stuff
that you can do on the regularchat, you cannot do on custom
GPTs and stuff you can do inthis version, you cannot do in
that version, like there's allthese different variations.
So one of the benefits of doingthis on Gemini is that you can
do it on your computer and notjust on your phone.
(28:21):
But yes, in functionality, it'svery similar to the
functionality that the chat GPTapp provides on the mobile app.
I want to talk about actuallythe third one.
So the third, you said thatthere are three options.
Option number one is just avoice option.
Number two is show Gemini,basically open the camera and
show you what you're seeing.
And option number three is shareyour screen.
(28:42):
And I, first of all, I'mwondering if you have any use
cases that you use this for, andthen I can share something that
I literally tried this week,which was really interesting.
Jimmy Slagle (28:49):
Yeah, there, there's a couple different
things that I've been using thisfor.
So the first one is as I'mbuilding out some of these.
different, like AI workflows,you can start to ask it like,
Oh, how can I like, like, youknow, looking through these
different nodes is kind of whatthey, they're called, like,
which one of these would be bestif I wanted to do X.
And so in this case, if I wantedto, you know, scrape Reddit to
(29:11):
find the top threads about, Whatpeople are talking about about a
I, it can, it can look ateverything that's happening on
my screen and recommend, Oh, youknow, you should go to the
Reddit analysis.
and, and then connect that tothis node.
And so it's kind of just likeyour helper as you're
navigating, any of these morecomplex websites.
So like, that's one that I'vereally been using it for.
(29:32):
even like, feedback on videos,like I'll sometimes show, like
pull up my screen so then Idon't have to download it and
export it, but it can just bereal time of like, Hey, what do
you think about this video?
I'll play it.
It then watches it and can giveme feedback in real time instead
of again, those one or twosteps.
So Those are two quick examplesof, of what I've been using it
for, and again, as you can seehere, like it's, it's seeing
(29:54):
what is happening on my screen,again, normally it would be
talking, and giving feedback,but, but yeah, so that's,
that's, that's one of the, the,a couple of these cases that
I've been seeing with this.
Isar Meitis (30:04):
Yeah.
So very similar to your firstuse case, I was working on a
really complex automation onmake, this weekend for one of my
clients and I got stuck onsomething and I couldn't figure
it out.
I was banging my head againstthe screen, the wall, the desk,
like everything.
And I watched like six YouTubevideos and I just couldn't
figure it out.
Like I couldn't find why it'snot doing what I think it needs
to do.
(30:24):
And they're like, well, wait aminute.
And then I opened this and Ishowed it to desktop and I'm
like, okay, here's what I'mtrying to do, here's where it's
failing, helping troubleshootthis.
And then it, it knows make, andit knows what I'm doing and it
understands parameters and Iliterally had a consultant for
free helping me solve a makeproblem that I couldn't solve.
(30:45):
And it ended up being somethingreally, really stupid, but.
By the way, it wasn't able tohelp me solve it, but it helped
me find the problem.
And then I solved it in fiveminutes, but I just couldn't
find what the hell was wrong,before.
And so think about the future ofany software.
Of any computer game, of anyprogramming problem, of anything
(31:08):
we're going to have.
First of all, on the generaltools, they will know all the
most common tools that we use,right?
So it will know Make, it willknow Microsoft Outlook, it will
know how to set up things inAdobe, like it will know all
that stuff because it alreadydoes.
So you can literally show itwhat you're doing and it can
give you tips and guide youthrough the steps, whether
you're a total beginner or anadvanced users.
(31:30):
Advanced users are getting stuckon something very specific,
which is incredible.
The other thing, companies willstart training it on their own
tools.
That will be a button.
So instead of the chat with usright now, but that is between
you and me, we're worthless 95percent of the time, you will
click a button and there's goingto be an AI.
That's going to look on thescreen with you on what you're
(31:51):
actually doing in your setupwith your problem.
And we'll guide you through howto do this.
And this is.
Not the future.
This is right now.
It's just a matter of time untilcompanies actually add that
functionality to the, to thetools.
But if any software company thatdoesn't do it is a complete
idiot, because it literally is acomplete game.
It talks to you as if it'ssomebody standing behind your
desk.
Shoulder and showing you what todo.
(32:13):
So if you haven't used this,even by the way, people are
using it for gaming right now.
So I showed it to my kids andthey started like, Oh my God, I
can use this for this.
And I can use it for that.
So they, they now have aconsultant playing computer
games as they're playingcomputer games.
It's just amazing.
I will say one thing, whiledoing my process, it got stuck
three times in the process.
And, and then he doesn'tremember because it.
It's now a new session.
(32:33):
So I had to start from thebeginning.
So that was a little annoying,but like you said, this is very,
very early on in this world.
And the next thing, by the way,To me, a little scary, but
there's no going back.
You know, the Pandora box isalready open is all the
wearables, right?
All these glasses that knows howto connect to this.
And then they see everything yousee, whether you're outside or
(32:53):
your computer screen or in a baror wherever you are, and it can
assist you and help you andguide you and provide you
information about.
Everything that's happening andthat's, again, on from a privacy
perspective, I'm not very happyabout it, but from a reality
perspective, there's no goingback like a few years from now,
almost everybody will wear somekind of a wearable that will be
connected to an AI into yourBluetooth headphones and we'll
(33:16):
be able to provide you anyinformation you want about
anything that it sees, which iseverything that you see.
Jimmy Slagle (33:22):
Yeah, exactly.
I mean, again, it's like this islike when chat GBT was released.
And immediately, you're going tohave tons of people that are
going to figure out specific usecases for this tool.
And that's what I think 2025 isgoing to be people are going to
be exploring it in wearables inin all the different vertical
like I'm sure we'll have one forCanva that's just been trained
on, you know, Everything that'spossible within Canva and
(33:42):
specifically to like gooddesign, video editing, like all
those, we're going to start tosee where all you have to do is
share your screen and you'regoing to have an expert because
again, you can give this custominstructions as well.
so you're going to start to seea lot of these, these kind of,
wrappers, if you will, of this,technology come out.
Isar Meitis (33:59):
Jimmy, this was a fantastic session.
Like we really touched on stuff.
I think most people, even peoplewho are like advanced users
don't have a clue that exists.
So I'm very grateful for youdoing this and taking us down
that path.
If people want to follow you,learn from you, work with you,
what are the best ways to dothat?
Jimmy Slagle (34:17):
Yeah, honestly, LinkedIn is probably where I'm
most active.
So just Jimmy Sligo, you can,you can connect with me and
start a conversation there.
I do like to make content on TikTOK as well.
fortunately it's, it's, it'sback.
And so it's back.
Yes.
It's back for like three
Isar Meitis (34:31):
seconds.
And then, you know, the newpresident said, well, whatever.
And yeah,
Jimmy Slagle (34:35):
exactly, exactly.
So, yeah, connect with me onsocials again.
Like, like really what I do isthose AI workflows, that I
quickly we help brands create alot of those and, and, and
figure out just how to do thingsbetter, faster, and more
efficient.
So, always down to chat how to,how to make, your company be
able to operate, especially themarketing team in the most
efficient way possible.
Isar Meitis (34:57):
This
Jimmy Slagle (34:57):
is awesome.
Isar Meitis (34:58):
Thank you so much.
Everybody who joined us again,thank you for being here.
I know you have other stuff todo on Thursday afternoons.
but if you, again, if you'relooking for more structured
training, the course is comingup on February 17th, you can
find it on my LinkedIn or mywebsite multiplied on AI, or
just find me and ask me if youhave any questions.
and that's it.
(35:18):
Thanks again for everybody forjoining live.
Thank you so much, Jimmy.
This was absolutely fantastic.
Jimmy Slagle (35:23):
Yes, you're welcome.
Thank you again for having meon.
Popular Podcasts
40s and Free Agents: NFL Draft Season
Daniel Jeremiah of Move the Sticks and Gregg Rosenthal of NFL Daily join forces to break down every team's needs this offseason.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Bobby Bones Show
Listen to 'The Bobby Bones Show' by downloading the daily full replay.