June 6, 2025 • 36 mins
Check out my newsletter at TKOPOD.com and join my new community at TKOwners.com
HoldCo Bros are back! In this episode, Nik and I talked about some of the most practical AI use cases we’ve seen in a while. I shared a new idea I spun up just 10 minutes before we hit record, literally bought the domain and had Replit building it mid-episode. We talked about how to turn AI-generated pitch decks into real income, how to use Manus and Perplexity Labs to shortcut market research and make money with it, and how quizzes powered by LLMs can reshape hiring and lead gen. This one's packed with ideas for anyone trying to actually use AI to build and grow something fast.
Learn more about Nik here: http://linktr.ee/cofoundersnik
Share your ideas with us:
- Nik@cofounders.com
- Chris@cofounders.com
Timestamps below. Enjoy!
---
Watch this on YouTube instead here: tkopod.co/p-yt
Ask me a question on or off the show here: http://tkopod.co/p-ask
Learn more about me: http://tkopod.co/p-cjk
Learn about my company: http://tkopod.co/p-cof
Follow me on Twitter here: http://tkopod.co/p-x
Free weekly business ideas newsletter: http://tkopod.co/p-nl
Share this podcast: http://tkopod.co/p-all
Scrape small business data: http://tkopod.co/p-os
---
00:00 The Underrated Potential of AI
02:58 Innovative AI Applications and Ideas
05:48 Leveraging AI for Personalization and Hiring
09:14 Pitch Decks and AI: A New Frontier
12:08 Using AI for Market Research and Lead Generation
15:00 Prompt Engineering: Enhancing AI Outputs
17:52 Perplexity Labs: Revolutionizing Research
21:07 B2B Applications of AI-Driven Reports
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
I feel like this is the most underrated things in AI right
now. No, let's talk about it.
So there's a guy on Twitter, pitch deck guy.
He goes and he scrapes SEC filings when people launch a new
fund or anything and he will do cold outreach to them and say,
hey, congratulations, I'd love to make a pitch deck for you.
Here's My Portfolio. What does he charge 3500 bucks?
Nick Manus has an API that couldbe fully automated as of today.
You're just like giving value tothem right away.
(00:22):
It's genius. Oh my gosh, that's a great idea.
How easy would it be? At first I was like, this is
just deep research, No? Here's what Perplexity Labs
does. How do we meet money on this?
OK so I am chock full of ideas today, but I have one that's
literally 10 minutes old. Like I just bought the domain
name 10 minutes ago and Replit is building it literally as I
(00:43):
speak because I just can't wait.I'm not going to lie, Nick, you
kind of inspired this idea. Oh, am I going to be offended by
this? No, no, no, you're not.
You're not. So recently on your podcast
Economics, you've been asking people everyday, how are they're
using AI, right? And so I've just been thinking
of that question, How are you using AI?
(01:04):
How are you using AI? And it's like, I feel like
that's going to become a standard question over the next
decade, right? Was like, what do you do for
work? What do you do for work?
What like that question will never get old, but instead it's
going to be not let's going to replace it, but it's it's going
to be that type of question where it's like, oh, how are you
using AI, right? Because that is the hardest part
with all of this is just thinking of how to use it.
(01:24):
Like, Oh my gosh, this new feature is amazing.
It's going to change the world. How do I use it?
Or more importantly, where in mybusiness or where in my life do
I apply this? Because I just don't know.
And it's too much. You don't even know where to
start. You don't even know which
question to ask to get started. Like, OK, which of these tools
do I even use to get started? I I totally agree.
Yeah. OK.
So I went to Namecheap and I looked up how do you use ai.com
(01:47):
and it was not taken. No, it was not taken.
So I bought it and then I'm going to.
Show him. It's a good name.
I'm going to share my screen andI I went to Google trends.
I did this after I bought it to be fair, and I'm sharing the
Google Trends chart of the phrase.
How do you use AI? It peaked like at the local peak
was February, which is interesting out I think
(02:09):
operator, maybe operator came out in February.
That's the. Operator, yeah, that sounds
about right. But it's still climbing.
Like if you're just listening, it's a it's a good looking
chart. OK.
And so I started thinking, and this is literally 10 minutes
old, so this is probably a bad idea at this point, but you'll
help me make it a good one, right?
Yeah, right. And I'm not thinking of a
billion dollar idea right now. I'm just thinking of a rapper,
(02:30):
some sort of a website, a gimmicky website, maybe a
newsletter, something. This is what I have repple
building me right now. Can I read you the prompt?
Yes. OK, I want to build a website
where people can merely submit ideas to others of how they are
using AI. This is the key.
But you can only see everyone else's use cases once you submit
your own use case. And then you can upvote others
(02:53):
ideas of how to use AI after yousubmit your own.
So it's like you have to give tothe system before you get back
from the system. You got to give.
You got to give. Did you give and then they can
subscribe to be notified of all the top upvoted ways.
People are using AI every week. So upvote ability, you can't see
(03:13):
anything on the website until and when you go to the website,
it's just an open cursor box, just like you see on replet or
ChatGPT or anything. And you can type in ABCD and get
something. It doesn't matter, right?
And then once you type it in, you can see a bunch of use
cases. You can upvote, yadda, yadda,
yadda. And then as like a newsletter
play or an e-mail capture play, you could say, hey, we'll take
the top five most upvoted ideas and e-mail them to you every
(03:36):
week. What do we think?
Oh, I love it. I think it's an amazing idea.
It's kind of how I feel about AI, generally speaking, is it's
an idea generator. I'm the guy who's in a group.
When we're trying to figure something out, I'll throw out
the first bad idea because we kind of need that first bad idea
at least just. 2nd through 13th.Maybe all of them.
Maybe every idea is a bad idea. Maybe every idea that comes up
(03:59):
that's bad in that conversation is they're all from you, for
instance. But when you're brainstorming,
you need somebody to throw out something first and then you can
make a decision. It's it's kind of like when
you're at home and your wife's like, hey, honey, do you like
the blue color for this wall or the yellow color for this wall?
No matter what you say, she's going to go with the opposite.
It just finally gives her an opportunity to be like, yeah, I
(04:20):
don't like that. I'm going to go the other way.
Right. That's kind of how that's how I
view AI as I get like the first iteration from it.
And I'm like, Nah, I don't like that.
And then I start working with it.
So I, yeah, I think there's a ton of demand from people just
to hear how other people are using AI just for an idea that
they can kind of steal and, and apply to their life.
Or like if you want her to spendtime with you, you say.
(04:41):
Hey, do you want? Do you want to spend time with
Raul or myself tonight? Your private tennis coach?
And you always suggest Raul, because then she'll hang out
with you, right? Yeah, I, you know, oddly, oddly
enough, she she just always tends to pick Raul.
It's really weird. No matter.
There's really no correlation towhat I say.
(05:02):
Yeah, and. Sometimes she doesn't even wait
for my answer. It's like she knew what I was
thinking. Here's why I like this idea.
Again, you and I don't talk before we do these things.
One of my ideas today is I love personality quizzes.
I know everybody does because you're like, I want to know
(05:22):
which House of Griffin or you know, which House of Hogwarts I
would be in. Am I Gryffindor?
Am I Slytherin? And you do this personality quiz
like my son. He wants to know which Marvel
character he is. Like we all want to know what
our personalities are. And I've actually, I've created
a few of these. I'm I'm going to share my
screen. So here's how this works.
I make you free videos. I actually know what I'm talking
about. I have no greasy sales pitch at
the end. And if you implement what I talk
(05:42):
about, you'll make a lot more money and have a better life.
And all I ask for in return is that you hit the subscribe
button and maybe even the notification bell just like
that. Thank you.
Hang with me for a second, I promise that this is going to
make sense. I interviewed a bunch of people
for my podcast and I was just like trying to figure out what
what do they do with AI? And so from all of their
answers, I was able to kind of understand which LLMS people use
(06:03):
for which use cases. I went to Chat TBT and I was
like, help me create this quiz based off of all of this data.
And so it helped me create this quiz.
I went then to Lovable and here it is, right?
This is the entrepreneur's guideto LLMS.
Which AI model is right for yourbusiness?
And there's Chat TBT, there's Claw, there's Perplexity, and
there's Grok. And the quiz helps you identify
which AI model aligns with your specific business goals, your
(06:24):
ideal AI implementation strategy, and which AI tools
deliver the highest ROI for yoursituation.
You know, you start the quiz andthen it just starts asking you
questions. What's your comfort level with
tech, etcetera, etcetera. So I'm not going to go through
the whole thing. It's a great idea, but to your.
Point the thing that AI is really good at right now, these
LLMS, is taking a ton of data, filtering it, and kind of
regurgitating it in a way that is concise.
(06:47):
It's not yet good at creating novel ideas, but it's really
good at taking a lot of information and synthesizing it
and giving it to you. So along the lines of what
you're talking about, I think that there are tons of
opportunities to use AI to give other people ideas of how to use
AI. This is one example.
This will give you an idea of like, should I go with Grok?
(07:08):
Should I go with Chatty Petit? Should I go with Google?
You know which one of those? You go through this quiz, you
answer some questions and it just spits out a general, hey,
you're not very tech savvy, justgo chat TBT.
It's the best user interface, right?
But I think there's a million sort of niche implementation or
offerings that you can create around this.
How do you know yourself better?Like I'll just go through this
(07:28):
really quickly. I'm just going to click on some
of these blah, blah, blah, and I'll show you at the very end,
this is what it looks like. And again, it's using the LLMS.
Of course, I got to have a gating mechanism.
You got to put in your e-mail ifyou want to see your score,
boom, here's my profile. I'm the client whisperer.
Yeah, my primary model is Claude.
(07:49):
And then it gives me all the reasons for that, etcetera.
So your idea, which is like a community of sharing idea, and
in order to get into it, you've got to give, I think it's
freaking brilliant because it just automatically gets people
to engage. This is another type of idea.
Playing on that, how do you get AI to reach into you and your
insights and give you a recommendation of what might or
(08:10):
might not be useful to you? And obviously this is one way
I've kind of applied it. That is amazing.
How long did this take you realistically?
So this is the first time I did it.
So it took me a couple of hours.If I were to do it now, probably
an hour. Because what you have to do is I
have to take all the data, synthesize it, put it in a
format where then I can go to Chachi PT and say, hey, create
the quiz for me. And then Chachi PT creates the
(08:31):
quiz. And then I go to lovable and I
go back and forth, blah, blah, blah.
And then it it creates this. It's honestly doesn't take that
much intellectual power, just kind of takes some time to go
through the steps. I'm sure there's a way I could
even automate it. I just haven't done it.
This is beautiful. So can I keep building on this?
Yeah, please. OK, so along this idea, I think
(08:52):
there's a huge opportunity for someone to step into the space.
You've heard of Dribble again, there's another company well
found. I was doing some research Contra
like these are niche job boards effectively where they're only
targeting, you know, one specific skill set like
dribbles, very specific designers.
Yeah, well found, I believe is very specific.
It's like Angel list job listings or something like that.
(09:15):
So these are like VC Angel listing start-ups, but they
found a niche. They're not in everything to
everybody. I think by coupling this like
quiz idea with a job board, you could match job seekers and job
offers employers with each other.
That's personality based and notnecessarily skill or background
(09:37):
based. Because many times you get into
a position, it's like, oh, they just didn't match, they mesh.
They weren't a cultural fit. And most companies is my belief,
if they're a cultural fit, you can kind of train them on the
things that you want them to do.So you could apply this think to
any niche globally. You could apply this to any
niche. You could build a company that's
just matching people who are digital designers for painting
companies. Maybe maybe that's too specific,
(09:59):
but I think pretty easily you could build a wrapper, you could
build a software platform, etcetera.
That's like high level VC land tech opportunity.
But I think for smaller companies, so for people like
you and I, we could create our own internal quizzes.
You're hiring people for your content team.
I'm hiring people for my contentteam.
Like a Myers Briggs for the culture that we want to see in
(10:19):
our companies. Exactly.
And we have so much data about us, right?
We can say like, hey Chachi Petit, create a quiz that
somebody can take that will meshwith me.
A quiz that will prove if someone is good working for
someone. That's very insecure.
You know what I'm saying? Yeah, like for that would be for
(10:39):
you, not. That would be for me.
That would be for me. Yeah, yeah, yeah.
It's asymmetric, right? Like I don't have to pay for the
for Myers Briggs access. I don't have to have them log
into a specific website or get credentials for it.
It's like, hey, I created this wrapper, take this quiz.
I want to see if we're a mesh and at least it would give me
some insights into where their strengths and weaknesses are.
(11:01):
So that would be like an internal thing that pretty much
any company pretty easily could create a route that.
Would do that dude. Dude what are you going to pay a
recruiter to place someone like we're talking 30% of first year
salary. 10 to 30%, I mean, just depends on the role of a
recruiter, yeah. $150,000 a yearsalary, $10,000 to place
someone. Let's say you set up an
(11:22):
automation to message everyone on LinkedIn that is in HR in
hiring. And you say, listen, I match
people with HR departments basedon their personality type or
their culture fit. But this is very unique.
I want to know exactly what yourcompany's culture fit is.
Would you mind simply taking this quick quiz and I'm going to
(11:42):
tell you exactly what your scoreis.
And then worst case scenario, I send you some qualified
applicants and then that hiring decision maker, they feel this
and it might take 1020 minutes. It's like a type form.
You don't know when it ends. You've got some cost files that
you're just answering, answering, answering.
And it's building a like an archetype of your company,
exactly the type of person that you like to hireright.
(12:04):
And then it spits out a result. It's like a new numerical result
or something. And then on the other side of
that equation, you, you send a similar quiz that matches it up
with job seekers. And then you can send
notifications to the hiring managers of saying, Hey, hey,
let's say it's a scale of one to100.
Hey, you're a perfect score of aperfect candidate is an 87.
(12:25):
Look at these 80 sevens that we just found that just filled out
the same form, right? Look at these 85 to 80 eights
that we just found. And then you could match them,
place them. Dude, OK, I'm going to go on the
other side of it. This could be a massive
differentiator for any company that they build their own quiz.
It's like, hey, and I don't know, you're better at this than
I am. I don't know how they would get
people to take it. But let's say they build their
(12:46):
own quiz. It's a lead magnet.
People start taking it and then they're just they're just
building in order to get the quiz results, you've got to put
your e-mail in. So they're just building this
list where anytime they have an open position coming up, they
can go to the list and be like, Oh yeah, Shonda 857, she was a
nine in these skill sets. Let's reach out to her.
So let's think of a company thathas a.
Really good. They could white label it from
(13:08):
us, right? Like they would use it
internally, Yeah. Or, let's use you in as an
example, how many followers do you have across platforms now?
3 million. 3,000,000 let's say that you decided next week that
you're like, hey guys, I built this cool quiz with AI.
It only took me 20 minutes. Check it out.
And it was specifically for thislike it was asking questions
(13:28):
about potential employees. For every video that you post,
how many views do you get? 800,000.
OK, 800,000 to to 1,000,000 and let's just say 1% of those
people, so 8000 people to 10,000people actually go and fill out
the survey. You've got 10,000 people who
filled out the survey with an e-mail address that you can go
at any time now to query if you're looking for a position,
(13:50):
if you're like, oh, I need an accountant, Oh, I need a
whatever social media manager. I'm going to go look at this
list right now and see how many of them actually matched with my
company's culture score. It's a no brainer.
Think about that for HR department, even like big
companies HR departments, they could easily build this within
an HR department. It's not hard and.
You don't even have to include questions about their skills.
(14:12):
They could only be questions about their personality, you
know, totally or their their preferences.
Do you ask any random questions in your interviews?
Yeah. What are some of them?
Like historically, throughout your life, what have you been
like, abnormally obsessed with? One question that I always ask
is which one of these scenarios stresses you out more?
(14:33):
Being dropped into the middle ofa foreign country with no plan,
no itinerary, no idea where you are or going to a foreign
country, and every minute of every day that you're there is
scheduled out to a tee and you have no flexibility to do
anything outside of what's scheduled for you.
(14:54):
Now, which one of those would give you more stress?
Scheduled one. Yeah, it's the second one,
right? Like, so usually if I'm
interviewing a marketer, to be honest, if they're like, oh, the
first one would stress me out more.
I'm like, yeah, you're not goingto be a good cold caller.
Like if you don't do good ambiguity, I'm not.
And I know like there are a lot of other factors there, but
that's one of the questions thatI ask.
So to me that the whole quiz would be those types of
(15:15):
questions. It's not going to be like.
How would you rate yourself as ateam player for for?
Because everybody's going to answer that question the way
that they think the employer wants to hear it.
So video is cool, but you know what's better?
Long form audio via podcast and my newsletter, TKO pod.com.
Go there to subscribe for free to my newsletter.
It's one e-mail a week, very tactical.
And then go to my audio podcast,3 episodes a week.
(15:38):
Stuff like this, you're going tolove it all free, no sleazy
sales bitch. TKO pod.com because if you ask
them, do you have a high tolerance for ambiguity?
Absolutely. I do.
But you don't really know if that's true, if that's what
you're really getting at, right?Like.
Right, another just quick interview question.
I love is kind of it's very similar to yours.
I think I stole this from my first million, but it's like,
(15:58):
what was the last project you spent all night working on or
lost sleepover? Like I just want to hear what
they nerd out about. Like what are you obsessed with?
That's a really interesting question.
OK. Can I go next?
Yeah. All right.
So have you played around at allwith Manus's new feature of
creating pitch decks? No, I've read about it but I
(16:19):
have not played with it. I feel like this is the most
underrated things in AI right now.
No one's talking about it. Why are we talking about this?
I'm going to share my screen. While you're pulling that up, so
I I've messed with GPT 4 O rightand the PDFs and the slide decks
that they can build are incredible, but you have to
build them sequentially and you have to build them one at a
time. It's not just going to spit out
10 slides for your slide deck. So what you're saying is Manus
(16:42):
actually spits out the whole deck?
Holy crap. I'm going to share my screen.
All right, so this is Manus, This is my prompt.
This is the whole thing. Make a PowerPoint deck for me,
showing the growth of the RV park industry.
Like no details, nothing. OK.
And it worked, It worked, It worked.
It worked for 12 minutes and 31 seconds and it spit out this 12
(17:05):
slides. Holy crap where did it get that?
Hold on, go back to the first picture dude.
I could be wrong. Have you used this picture
before in an RV deck? It looks really familiar.
It looks like one that I have used, but I don't think I've
used this exact one. OK, but I mean look at this.
Oh my. North American market growth.
RV ownership. This looks like the deck that I
(17:26):
use for the same purpose. Now, does Manus, do you have
this deck in Manus? Like is it pulling from stuff
you've already created or is this?
No. Totally.
Honest about RV park stuff? Yeah.
I didn't teach it anything. Isn't that crazy?
So when I use ChatGPT it spits it out as APDF.
It's uneditable. Is this editable?
(17:47):
It's spit it out as APPT as a PowerPoint.
Really. Yeah.
Holy crap man. So side quest here, how much are
you using Manus right now? Not that much.
I just started playing around with it when this new feature
came out. I was using it a month or two
ago, but I I just moved on. What were you using it for a
month or two ago? It's like agentic like features,
(18:09):
it's kind of acts like operator,but it wasn't as good in my
experience. And so when I didn't like that
output, I just I went back to church PT.
What are people using Mantas forright now?
Mantas. Yeah, that's what I said.
I think they're using it for this a lot.
I don't know. It's a Chinese owned company and
(18:32):
so kind of TBD on what's happening with our data here,
but we'll see. I think it'll work out, but
we'll see. How big is the pitch deck
industry? It's multi billions the
corporate world is built on. Pitch sticks.
Pitch sticks. Yeah, it's, it's the modern idea
conveyor that we use. So there's a guy on Twitter, I
(18:56):
think he's the pitch deck guy and his whole business, this is
it. He goes and he scrapes like the
filings, SEC filings when peoplelaunch a new fund or anything,
you know, people that are about to need a really nice pitch deck
and he will do cold outreach to them in the right time, the
right moment and say, hey, I sawyou launched a fund for buying
single family homes. Congratulations, I'd love to
(19:17):
make a pitch deck for you. Here's My Portfolio.
What does he charge? 3500 bucks?
Love it. And they look really good.
They're really good. So Nick Manus has an API who's
stopping you or anyone from building an agent or an app or
anything. It could be just an automation
with with Gumloop or Lindy or Zapier that scrapes all those
(19:40):
SEC filings. The funds that are launched
permissionlessly goes to Manus through the API and says here's
all the information for this fund.
I want you to make a really goodlooking pitch deck according to
these pitch deck. I wanted to have this design and
send it back to me. Boom, you get it.
Then you go to Apollo or any other feature that finds
(20:00):
people's contact information andyou cold e-mail the pitch deck
to them with watermarks that says, hey, congratulations on
the fund, I made you a pitch deck.
Would you like to buy this? Here's the first five slides.
Could that could be fully automated as of today?
OK, so there's the pitch deck guy and then there's Matthew
Gore. I had him on my podcast.
He's the pitch deck cowboy. He has like that's his business.
(20:21):
Has he built pitch decks for people?
Funny story really quickly, I was on my way to Bogus Basin in
Boise, ID. I'm buying a freaking burrito at
this random place. Order it.
I'm picking it up and this guy next to me goes, are you on
Twitter? And I was like, why yes, I am.
He's like, are you Co founder's Nick?
I'm like, yeah, yeah. He's like, I'm the pitch deck
guy. I'm like, wait, what?
(20:42):
He's like, yeah, yeah, yeah, yeah, I'm the pitch deck guy.
I live in Dallas, but I'm actually from Boise.
So just like random. Yeah, randomly on my way up the
hill, I met him, which was. Which is crazy.
Yeah, totally random. I think that the ability to
convey a story is like so valuable that most people think
of pitch decks as like, oh, a major presentation or I'm trying
(21:04):
to raise money, or it's like something super serious.
I think just the ability of pitch decks to convey any idea
is so valuable that, yes, there's the SEC filings, but you
could permissionlessly offer this as a marketing agency as a
way to get clients. You're like, hey, Ethan
Driscoll, This would be a perfect, like, cold outreach for
him just to permissionlessly have Manus take all of his
(21:26):
ideas, put them into a slide deck by scraping whatever the
list is that he's created off the Internet and just generate
business that way. Yeah, I, I love your idea.
I also think there's a bunch of different applications you could
use for for cold outreach. 2 Twomore ideas You heard of
slideshare.net? No.
They get it's either 5 or 10 million visits a month.
(21:46):
What the fridge? It's just a website for
uploading your slides, your decks to like to be hosted, to
be viewed, right. So if you go to slideshare.net,
you can see like Netflix's famous 125 slide presentation on
culture there. And then you can also see, you
know, Econ one O 1 from Alabama Community College, some student
(22:06):
group. They uploaded there so they
could present it in class and like it has 3 views.
Hundreds, if not thousands of slides are uploaded to
slideshare.net every single day.This company sold to LinkedIn in
2012 for 112 million and then they sold from LinkedIn to
scribed for I don't know how much it wasn't disclosed.
So this is a big business. It's a nine figure business.
(22:28):
Why not like do the same thing with that, like create a
permission listing that anytime a new slide is uploaded to
SlideShare, you revamp it, you redesign it.
So instead of making one from scratch for the SEC filings,
you're just redesigning them, sending them back to the
original creator and saying, hey, you may like this better,
you may not. If you do buy it for 20 bucks,
200 bucks, whatever. And automate.
(22:52):
And automate it or, you know, teachers paying teachers,
teacherspayteachers.com. It's a place for teachers to
sell their that layered lesson plans.
You can go on there. It's such a good name.
Hey, teachers paying teachers.com, What do they do?
Well, they pay teachers. You could go on there and like
DM, mass DM all the teachers that are selling lesson plans
and say, hey, why don't you sella a pitch deck alongside the
(23:15):
Lesson plan. Like a Lesson plan is great but
you'll probably sell even more if you have a pitch deck
version. I'll convert that for you for 20
bucks. What are we doing or.
Anything with a Lesson plan, People who teach church, like
they'll say we have a a Lesson plan for Sunday.
What do we typically do? We're like, maybe we have some
on our iPad, whatever. We could just easily upload an
outline and be like, hey, createa PowerPoint presentation based
(23:36):
off of this. What are we talking about right
now? Dude?
What? Are we doing?
What are we doing? Why are we?
Talking right now, we need to begrinding building.
Oh my gosh. Maria close.
Clear. The rest of my day, I don't know
how, Maria. Is.
Can I build off this for a second?
(23:57):
Please. All right.
So it's not easy, but what you and I are talking and we're
like, oh, dude, why wouldn't we do this?
All we have to do go to Chachi Butiti, enter this prompt and
then it, it automates here and all of a sudden we've got a, a
product. I still think for many people,
the prompt is like the hardest thing for them to get past.
Like to think creatively enough in order to get the prompt to
(24:17):
give you the output that you want is incredibly hard.
And I found something this week that I want to share with you.
It's going to be stupid simple and you're going to make fun of
me, but I think has one of the best pieces of advice I've ever
gotten when it comes to prompting.
Are you ready for it? Yep.
All right, here we go. All right, so you've probably
seen this, this was a Reddit post and this guy's talking
(24:38):
about how he uses the LLM to generate prompts for him.
So in the example that you just gave where you're sitting there
and you're like, I want Manas tocreate me a slide deck based on
this information, but like, how do I even give it the specific
information that I want in orderfor it to create the slide deck?
He's essentially teaching you how to create the prompt by
(24:58):
using the LLMS to create the prompt.
So the first thing that he does is he tells the LLM generate A
detailed prompt engineering guide.
The audience is, let's say graphic designers.
So you can fill in the role there, and the role might look
like book authors, software developers, customer support,
whatever. And then he pastes in five
examples of how I want my promptto work.
(25:18):
So what he's doing there is he'sgiving it context to draw.
He's helping the LLM understand what exactly he's looking for by
providing it with five examples of products that he really
likes. You got to give it more context.
Got to give it more context. All right, and now this is
probably the best advice I've ever read.
This next one I then instructed to quote, generate a prompt that
(25:40):
could have generated the examples outputs and include a
better set of examples. Dude, that blew my mind when I
read it. So, so he's sitting there and
he's like, hey, this is who you are, you're a graphic designer.
Here are 5 outputs that I really, really like.
Now I want you to to reverse engineer it and think of what
the prompt would be that would create the outputs that you're
(26:02):
currently seeing. So it's getting the LLM to think
about, OK, what are all the different elements within the
examples that are given to me that I need to get to in order
for this prompt to generate something similar to what I'm
seeing. What do you think of that?
It's genius. Like I want to hang up right now
and test this. Have you read this before?
(26:24):
I feel like this is. Something that you've seen?
OK. It blew my mind because it's
like so simple and yet it's going to be so.
I've already started using it and the outputs that I'm getting
are way better than I've ever used, than I've ever gotten
before. It's pretty.
How long does it take though to put all this together?
Which part like the five examples?
(26:45):
Yeah. Obviously it depends on what
you're trying to build. I've been using it for content
and like podcast ideation. So from that perspective it's,
it's been fairly easy because I,I kind of know exactly what I'm
looking for and I can go and find the outputs.
If I was starting a business from scratch, like we were just
talking about the slide deck generator, I would probably have
(27:06):
to go and find just five slide decks that I think look really
good. I don't know where I'll go and
look. So then the key is to say, all
right, look at these 5 slide decks.
Give me a prompt that would produce something that looks
like these five. Yeah.
And again, that's what we were originally talking about.
It may not even give you a good output, but it'll at least start
you down the road of messing around with this.
(27:29):
So you're going to see the the next steps actually build on
this. So then he goes to a new chat
and he says generate A detailed prompt evaluation guide.
The audience is graphic designer.
So what it's telling me that LLMis basically help me audit a
(27:50):
prompt and give me a guide that I can use to audit that prompt.
He then takes that what it's created and he pastes it into
the original prompt and then he tells it generate 3 improved
alternative prompts based on this based on the evaluation of
the prompt and then he picks thebest 1.
So effectively what he's doing is he's like getting the LLMS to
massage it to a point where he likes it and it makes sense.
(28:11):
Now, there were some comments onthis that I thought were
hilarious. One guy, he was like, I honestly
thought of that guy who you loveon Instagram who's always making
fun of the business gurus. And the, the quote was like, and
then what I do is I write a prompt that asks it to evaluate
the prompting that I wrote in order to evaluate the prompt
that evaluates prompts. It was like, wait, what the
freak? So he does get pretty meta here,
(28:34):
but you're using the LLMS to leverage your time and and get
you to a place where you can actually create these prompts
really quickly. I'm so excited to use this.
This is perfect. Oh, it's pretty cool, right?
Yeah, I've heard about asking the LLM what to ask it, like,
all right, all right, here's what I want.
Here's what I'm looking for. I want to make a a delicious
(28:54):
tomato salad. What should be the prompt I give
you to give me the best output possible?
That's like level 1 and this is like level 7 of that.
There are two things that I can remember that have been like,
oh, that's genius and always have stuck with me when it comes
to AI. One of them you gave me, which
was remember this when you're working on a thread and it tells
you something or it spits something out and you're like,
(29:15):
Hey, remember this makes a mental note.
The other one is this, it's like, OK, I would like this
graphic design or I like this adcopy.
What's a prompt that you would have written in order to get
this output? Like that to me is genius
because you are reverse engineering.
And that's that's effectively what we're trying to do anyways
when we see stuff that we're trying to copy.
Oh. That's gold.
(29:36):
It's gold, Jerry. All right, what do you know
about Perplexity Labs? I know a lot but assume I know
nothing. OK, cool, cool, cool.
Yeah, yeah, yeah, yeah, 'cause Iknow a lot, but I know our
audience doesn't. Yep, Yep.
So I didn't know much about it until earlier.
I I've been playing with it quite a bit.
You do have to have the $20.00 amonth perplexity plan.
(29:57):
But at first I was like, oh, this is just deep research.
Like Perplexity already had deepresearch.
It was fine. This is just more deep research.
No, no, no. So here's how perplexity is
different from deep research. So let's say I want to know that
like the market landscape of allthe keto apps out there, I want
(30:18):
to build a keto app and I want to know how profitable are the
ones that already exist? What's their cost per
acquisition? If they're using Facebook ads,
how many downloads do the top five keto apps get?
I want to know all this stuff. I'm going to go to deep research
either on Perplexity or on ChatGPT and give it that prompt.
And it's about 10 minutes later,it's going to give me thousands
of words with like a really thorough output.
(30:41):
Here's what Perplexity Labs does.
I just prompted this and my prompt was build an interactive
chart showing the correlation between sourdough baking and RV
traveling in a post COVID world.I don't know why, I just that's
all I can think of. So it gives you like multimedia,
(31:01):
it gives you the information you're looking for in different
ways. It will use charts, it will use
PDFs that you can download whereappropriate.
It's kind of the difference between having an intern go do a
little research for you and sending you a memo, and then
having a six figure employee that's like a researcher going
out and spending weeks aggregating all this data,
(31:23):
knowing when to use a chart, when to use text, when to use
pictures, when to use APDF. That's kind of the difference
between perplexity, so. What we're looking at, if you
can't see this right now, is there's an interactive chart
what's on the X&Y axis. Sourdough market size in
millions and then time period. Yeah, and then market size and
RV shipments in thousands. So RV shipments in thousands is
(31:45):
over here. Oh, so I'm assuming that the red
is RV shipments? Yes.
So that's seasonal basically. Yep.
Interesting. Goes up in the summer.
And so you asked if they were correlated.
Does it say it's correlated? Yep, it's -.004 weak.
Negative correlation suggests inverse relationship during
certain periods. Interesting dude.
(32:07):
It gives you the charts. You can visualize it, but then
it actually gives you a write up.
Yeah, like just as and here's a non interactive chart showing
the same thing. And then the proprietary charts,
these are not like they wouldn'tscrape the web and found some
Yahoo who made this and post it.It made these, yeah, it, I think
it took, just took raw data fromthe web and made these itself.
(32:29):
And look, here's a, here's a PDFor a CSVI can download.
So sometimes it gives you a CSVPDF, whatever you want,
sources on everything. So how do we use this?
How do we make money on this? Got to make money on this.
Got to figure out how to make money on this.
I mean, to me, it just goes backto this allows you to leverage
(32:52):
your time even more. As companies get bigger and they
have more resources, they can throw more resources at smaller
problems. So this would be an example of a
of a smaller problem. It's like in the past, I'm
growing a company, I need to do all this research.
But you, Chris Koerner, as you're doing all the market
research, you're kind of like internalizing all of this stuff,
(33:12):
right? You're looking at different
websites, you're reading different blog posts, whatever.
You're reading different charts.This gets you all the
information that you would have otherwise gone out and just
surfed the web for quickly in a format that you can digest.
So to me it would be a business built around servicing decision
makers, people who who have highleverage positions that are
(33:33):
making decisions all day. And that's kind of where you get
to as you grow your businesses. Dude, like look at this prompt
result asking about the keto appmarket.
It's all right here. All the iOS ratings, the annual
price, how many weekly downloads, weekly revenue,
active users like and it has 37 different sources for this data.
Like this is crazy. This would be like if if I got
(33:55):
in this ideation frenzy late at night as I often do and wanted
to. I really wanted to launch this
keto app. I would stay up all night, like
just doing this research that itdid for me in 11 minutes, OK?
Here's a lead magnet I think would be really cool.
Digital agencies. They could be doing anything.
They could be doing automation, they could be doing AI
implementation within a company,they could be digital marketing,
(34:18):
graphic design, anything. If they use this and they
generated a report for every company that's on their list
because every, every one of these businesses has like an
ideal customer. And let's just say your ideal
customer is importers of pastel fabrics.
I don't even know. You generate that report one
time and you e-mail it to them as a lead magnet to get your
(34:40):
foot in the door. So then it spurs the next
conversation because you're giving them something of real
value where it's like, hey, nextyear, these are the projected, I
guess I don't even know tastes that are going to be in the
pastel market. I know you make these decisions
every single year around this time.
Hopefully this helps you. The automation guys at
automationla-bs.com, whatever itis, they're going to read that,
(35:01):
they're going to see that you're, you can track whether or
not they open it and then you can easily follow up with them
next day, next week, whenever. But it's a really cool lead
magnet that like that's really valuable.
We talked about this before IbisWorld where I work, they're
charging $2000 reports. That's a $2000 report
potentially right there that youcan just use as a lead magnet
now as a lead magnet. Think about that I.
(35:22):
Know like if you're a consultant, like how big of
retainers could you get by leading with something like this
dude? Any business that wants a
customer name a business that's B to.
BH FAC Supply. Cool HVAC supply?
I'm an HVAC supply company and I'm looking for HVAC business
owners. Why don't you just send an
industry report for their specific market to every single
(35:45):
HVAC owner? Yeah, it pulls in real estate
data. Yeah, it pulls in, yeah.
Yeah. Here's where the growth trends
are of your market overall. Here's the age of properties in
your market. Here's the, I don't know,
typical home value. Here's the supply, whatever it
gives, like the whole entire thing.
Thought you might like this for your market.
Hope you enjoy it. Let me know if you have any
questions. You're just like giving value to
(36:07):
them right away. And who are they going to call
the next time? I mean, not necessarily
everybody's going to call, but just in terms of a lead magnet
and building trust with people, that would be phenomenal in B to
B sales, Any B to B sale. That's a freaking cool idea.
That's a freaking great idea. All right.
What do you think? Please share it with a friend
and we'll see you next time on the Kerner office.
I feel like this is the most underrated things in AI right
now. No, let's talk about it.
So there's a guy on Twitter, pitch deck guy.
He goes and he scrapes SEC filings when people launch a new
fund or anything and he will do cold outreach to them and say,
hey, congratulations, I'd love to make a pitch deck for you.
Here's My Portfolio. What does he charge 3500 bucks?
Nick Manus has an API that couldbe fully automated as of today.
You're just like giving value tothem right away.
(00:22):
It's genius. Oh my gosh, that's a great idea.
How easy would it be? At first I was like, this is
just deep research, No? Here's what Perplexity Labs
does. How do we meet money on this?
OK so I am chock full of ideas today, but I have one that's
literally 10 minutes old. Like I just bought the domain
name 10 minutes ago and Replit is building it literally as I
(00:43):
speak because I just can't wait.I'm not going to lie, Nick, you
kind of inspired this idea. Oh, am I going to be offended by
this? No, no, no, you're not.
You're not. So recently on your podcast
Economics, you've been asking people everyday, how are they're
using AI, right? And so I've just been thinking
of that question, How are you using AI?
(01:04):
How are you using AI? And it's like, I feel like
that's going to become a standard question over the next
decade, right? Was like, what do you do for
work? What do you do for work?
What like that question will never get old, but instead it's
going to be not let's going to replace it, but it's it's going
to be that type of question where it's like, oh, how are you
using AI, right? Because that is the hardest part
with all of this is just thinking of how to use it.
(01:24):
Like, Oh my gosh, this new feature is amazing.
It's going to change the world. How do I use it?
Or more importantly, where in mybusiness or where in my life do
I apply this? Because I just don't know.
And it's too much. You don't even know where to
start. You don't even know which
question to ask to get started. Like, OK, which of these tools
do I even use to get started? I I totally agree.
Yeah. OK.
So I went to Namecheap and I looked up how do you use ai.com
(01:47):
and it was not taken. No, it was not taken.
So I bought it and then I'm going to.
Show him. It's a good name.
I'm going to share my screen andI I went to Google trends.
I did this after I bought it to be fair, and I'm sharing the
Google Trends chart of the phrase.
How do you use AI? It peaked like at the local peak
was February, which is interesting out I think
(02:09):
operator, maybe operator came out in February.
That's the. Operator, yeah, that sounds
about right. But it's still climbing.
Like if you're just listening, it's a it's a good looking
chart. OK.
And so I started thinking, and this is literally 10 minutes
old, so this is probably a bad idea at this point, but you'll
help me make it a good one, right?
Yeah, right. And I'm not thinking of a
billion dollar idea right now. I'm just thinking of a rapper,
(02:30):
some sort of a website, a gimmicky website, maybe a
newsletter, something. This is what I have repple
building me right now. Can I read you the prompt?
Yes. OK, I want to build a website
where people can merely submit ideas to others of how they are
using AI. This is the key.
But you can only see everyone else's use cases once you submit
your own use case. And then you can upvote others
(02:53):
ideas of how to use AI after yousubmit your own.
So it's like you have to give tothe system before you get back
from the system. You got to give.
You got to give. Did you give and then they can
subscribe to be notified of all the top upvoted ways.
People are using AI every week. So upvote ability, you can't see
(03:13):
anything on the website until and when you go to the website,
it's just an open cursor box, just like you see on replet or
ChatGPT or anything. And you can type in ABCD and get
something. It doesn't matter, right?
And then once you type it in, you can see a bunch of use
cases. You can upvote, yadda, yadda,
yadda. And then as like a newsletter
play or an e-mail capture play, you could say, hey, we'll take
the top five most upvoted ideas and e-mail them to you every
(03:36):
week. What do we think?
Oh, I love it. I think it's an amazing idea.
It's kind of how I feel about AI, generally speaking, is it's
an idea generator. I'm the guy who's in a group.
When we're trying to figure something out, I'll throw out
the first bad idea because we kind of need that first bad idea
at least just. 2nd through 13th.Maybe all of them.
Maybe every idea is a bad idea. Maybe every idea that comes up
(03:59):
that's bad in that conversation is they're all from you, for
instance. But when you're brainstorming,
you need somebody to throw out something first and then you can
make a decision. It's it's kind of like when
you're at home and your wife's like, hey, honey, do you like
the blue color for this wall or the yellow color for this wall?
No matter what you say, she's going to go with the opposite.
It just finally gives her an opportunity to be like, yeah, I
(04:20):
don't like that. I'm going to go the other way.
Right. That's kind of how that's how I
view AI as I get like the first iteration from it.
And I'm like, Nah, I don't like that.
And then I start working with it.
So I, yeah, I think there's a ton of demand from people just
to hear how other people are using AI just for an idea that
they can kind of steal and, and apply to their life.
Or like if you want her to spendtime with you, you say.
(04:41):
Hey, do you want? Do you want to spend time with
Raul or myself tonight? Your private tennis coach?
And you always suggest Raul, because then she'll hang out
with you, right? Yeah, I, you know, oddly, oddly
enough, she she just always tends to pick Raul.
It's really weird. No matter.
There's really no correlation towhat I say.
(05:02):
Yeah, and. Sometimes she doesn't even wait
for my answer. It's like she knew what I was
thinking. Here's why I like this idea.
Again, you and I don't talk before we do these things.
One of my ideas today is I love personality quizzes.
I know everybody does because you're like, I want to know
(05:22):
which House of Griffin or you know, which House of Hogwarts I
would be in. Am I Gryffindor?
Am I Slytherin? And you do this personality quiz
like my son. He wants to know which Marvel
character he is. Like we all want to know what
our personalities are. And I've actually, I've created
a few of these. I'm I'm going to share my
screen. So here's how this works.
I make you free videos. I actually know what I'm talking
about. I have no greasy sales pitch at
the end. And if you implement what I talk
(05:42):
about, you'll make a lot more money and have a better life.
And all I ask for in return is that you hit the subscribe
button and maybe even the notification bell just like
that. Thank you.
Hang with me for a second, I promise that this is going to
make sense. I interviewed a bunch of people
for my podcast and I was just like trying to figure out what
what do they do with AI? And so from all of their
answers, I was able to kind of understand which LLMS people use
(06:03):
for which use cases. I went to Chat TBT and I was
like, help me create this quiz based off of all of this data.
And so it helped me create this quiz.
I went then to Lovable and here it is, right?
This is the entrepreneur's guideto LLMS.
Which AI model is right for yourbusiness?
And there's Chat TBT, there's Claw, there's Perplexity, and
there's Grok. And the quiz helps you identify
which AI model aligns with your specific business goals, your
(06:24):
ideal AI implementation strategy, and which AI tools
deliver the highest ROI for yoursituation.
You know, you start the quiz andthen it just starts asking you
questions. What's your comfort level with
tech, etcetera, etcetera. So I'm not going to go through
the whole thing. It's a great idea, but to your.
Point the thing that AI is really good at right now, these
LLMS, is taking a ton of data, filtering it, and kind of
regurgitating it in a way that is concise.
(06:47):
It's not yet good at creating novel ideas, but it's really
good at taking a lot of information and synthesizing it
and giving it to you. So along the lines of what
you're talking about, I think that there are tons of
opportunities to use AI to give other people ideas of how to use
AI. This is one example.
This will give you an idea of like, should I go with Grok?
(07:08):
Should I go with Chatty Petit? Should I go with Google?
You know which one of those? You go through this quiz, you
answer some questions and it just spits out a general, hey,
you're not very tech savvy, justgo chat TBT.
It's the best user interface, right?
But I think there's a million sort of niche implementation or
offerings that you can create around this.
How do you know yourself better?Like I'll just go through this
(07:28):
really quickly. I'm just going to click on some
of these blah, blah, blah, and I'll show you at the very end,
this is what it looks like. And again, it's using the LLMS.
Of course, I got to have a gating mechanism.
You got to put in your e-mail ifyou want to see your score,
boom, here's my profile. I'm the client whisperer.
Yeah, my primary model is Claude.
(07:49):
And then it gives me all the reasons for that, etcetera.
So your idea, which is like a community of sharing idea, and
in order to get into it, you've got to give, I think it's
freaking brilliant because it just automatically gets people
to engage. This is another type of idea.
Playing on that, how do you get AI to reach into you and your
insights and give you a recommendation of what might or
(08:10):
might not be useful to you? And obviously this is one way
I've kind of applied it. That is amazing.
How long did this take you realistically?
So this is the first time I did it.
So it took me a couple of hours.If I were to do it now, probably
an hour. Because what you have to do is I
have to take all the data, synthesize it, put it in a
format where then I can go to Chachi PT and say, hey, create
the quiz for me. And then Chachi PT creates the
(08:31):
quiz. And then I go to lovable and I
go back and forth, blah, blah, blah.
And then it it creates this. It's honestly doesn't take that
much intellectual power, just kind of takes some time to go
through the steps. I'm sure there's a way I could
even automate it. I just haven't done it.
This is beautiful. So can I keep building on this?
Yeah, please. OK, so along this idea, I think
(08:52):
there's a huge opportunity for someone to step into the space.
You've heard of Dribble again, there's another company well
found. I was doing some research Contra
like these are niche job boards effectively where they're only
targeting, you know, one specific skill set like
dribbles, very specific designers.
Yeah, well found, I believe is very specific.
It's like Angel list job listings or something like that.
(09:15):
So these are like VC Angel listing start-ups, but they
found a niche. They're not in everything to
everybody. I think by coupling this like
quiz idea with a job board, you could match job seekers and job
offers employers with each other.
That's personality based and notnecessarily skill or background
(09:37):
based. Because many times you get into
a position, it's like, oh, they just didn't match, they mesh.
They weren't a cultural fit. And most companies is my belief,
if they're a cultural fit, you can kind of train them on the
things that you want them to do.So you could apply this think to
any niche globally. You could apply this to any
niche. You could build a company that's
just matching people who are digital designers for painting
companies. Maybe maybe that's too specific,
(09:59):
but I think pretty easily you could build a wrapper, you could
build a software platform, etcetera.
That's like high level VC land tech opportunity.
But I think for smaller companies, so for people like
you and I, we could create our own internal quizzes.
You're hiring people for your content team.
I'm hiring people for my contentteam.
Like a Myers Briggs for the culture that we want to see in
(10:19):
our companies. Exactly.
And we have so much data about us, right?
We can say like, hey Chachi Petit, create a quiz that
somebody can take that will meshwith me.
A quiz that will prove if someone is good working for
someone. That's very insecure.
You know what I'm saying? Yeah, like for that would be for
(10:39):
you, not. That would be for me.
That would be for me. Yeah, yeah, yeah.
It's asymmetric, right? Like I don't have to pay for the
for Myers Briggs access. I don't have to have them log
into a specific website or get credentials for it.
It's like, hey, I created this wrapper, take this quiz.
I want to see if we're a mesh and at least it would give me
some insights into where their strengths and weaknesses are.
(11:01):
So that would be like an internal thing that pretty much
any company pretty easily could create a route that.
Would do that dude. Dude what are you going to pay a
recruiter to place someone like we're talking 30% of first year
salary. 10 to 30%, I mean, just depends on the role of a
recruiter, yeah. $150,000 a yearsalary, $10,000 to place
someone. Let's say you set up an
(11:22):
automation to message everyone on LinkedIn that is in HR in
hiring. And you say, listen, I match
people with HR departments basedon their personality type or
their culture fit. But this is very unique.
I want to know exactly what yourcompany's culture fit is.
Would you mind simply taking this quick quiz and I'm going to
(11:42):
tell you exactly what your scoreis.
And then worst case scenario, I send you some qualified
applicants and then that hiring decision maker, they feel this
and it might take 1020 minutes. It's like a type form.
You don't know when it ends. You've got some cost files that
you're just answering, answering, answering.
And it's building a like an archetype of your company,
exactly the type of person that you like to hireright.
(12:04):
And then it spits out a result. It's like a new numerical result
or something. And then on the other side of
that equation, you, you send a similar quiz that matches it up
with job seekers. And then you can send
notifications to the hiring managers of saying, Hey, hey,
let's say it's a scale of one to100.
Hey, you're a perfect score of aperfect candidate is an 87.
(12:25):
Look at these 80 sevens that we just found that just filled out
the same form, right? Look at these 85 to 80 eights
that we just found. And then you could match them,
place them. Dude, OK, I'm going to go on the
other side of it. This could be a massive
differentiator for any company that they build their own quiz.
It's like, hey, and I don't know, you're better at this than
I am. I don't know how they would get
people to take it. But let's say they build their
(12:46):
own quiz. It's a lead magnet.
People start taking it and then they're just they're just
building in order to get the quiz results, you've got to put
your e-mail in. So they're just building this
list where anytime they have an open position coming up, they
can go to the list and be like, Oh yeah, Shonda 857, she was a
nine in these skill sets. Let's reach out to her.
So let's think of a company thathas a.
Really good. They could white label it from
(13:08):
us, right? Like they would use it
internally, Yeah. Or, let's use you in as an
example, how many followers do you have across platforms now?
3 million. 3,000,000 let's say that you decided next week that
you're like, hey guys, I built this cool quiz with AI.
It only took me 20 minutes. Check it out.
And it was specifically for thislike it was asking questions
(13:28):
about potential employees. For every video that you post,
how many views do you get? 800,000.
OK, 800,000 to to 1,000,000 and let's just say 1% of those
people, so 8000 people to 10,000people actually go and fill out
the survey. You've got 10,000 people who
filled out the survey with an e-mail address that you can go
at any time now to query if you're looking for a position,
(13:50):
if you're like, oh, I need an accountant, Oh, I need a
whatever social media manager. I'm going to go look at this
list right now and see how many of them actually matched with my
company's culture score. It's a no brainer.
Think about that for HR department, even like big
companies HR departments, they could easily build this within
an HR department. It's not hard and.
You don't even have to include questions about their skills.
(14:12):
They could only be questions about their personality, you
know, totally or their their preferences.
Do you ask any random questions in your interviews?
Yeah. What are some of them?
Like historically, throughout your life, what have you been
like, abnormally obsessed with? One question that I always ask
is which one of these scenarios stresses you out more?
(14:33):
Being dropped into the middle ofa foreign country with no plan,
no itinerary, no idea where you are or going to a foreign
country, and every minute of every day that you're there is
scheduled out to a tee and you have no flexibility to do
anything outside of what's scheduled for you.
(14:54):
Now, which one of those would give you more stress?
Scheduled one. Yeah, it's the second one,
right? Like, so usually if I'm
interviewing a marketer, to be honest, if they're like, oh, the
first one would stress me out more.
I'm like, yeah, you're not goingto be a good cold caller.
Like if you don't do good ambiguity, I'm not.
And I know like there are a lot of other factors there, but
that's one of the questions thatI ask.
So to me that the whole quiz would be those types of
(15:15):
questions. It's not going to be like.
How would you rate yourself as ateam player for for?
Because everybody's going to answer that question the way
that they think the employer wants to hear it.
So video is cool, but you know what's better?
Long form audio via podcast and my newsletter, TKO pod.com.
Go there to subscribe for free to my newsletter.
It's one e-mail a week, very tactical.
And then go to my audio podcast,3 episodes a week.
(15:38):
Stuff like this, you're going tolove it all free, no sleazy
sales bitch. TKO pod.com because if you ask
them, do you have a high tolerance for ambiguity?
Absolutely. I do.
But you don't really know if that's true, if that's what
you're really getting at, right?Like.
Right, another just quick interview question.
I love is kind of it's very similar to yours.
I think I stole this from my first million, but it's like,
(15:58):
what was the last project you spent all night working on or
lost sleepover? Like I just want to hear what
they nerd out about. Like what are you obsessed with?
That's a really interesting question.
OK. Can I go next?
Yeah. All right.
So have you played around at allwith Manus's new feature of
creating pitch decks? No, I've read about it but I
(16:19):
have not played with it. I feel like this is the most
underrated things in AI right now.
No one's talking about it. Why are we talking about this?
I'm going to share my screen. While you're pulling that up, so
I I've messed with GPT 4 O rightand the PDFs and the slide decks
that they can build are incredible, but you have to
build them sequentially and you have to build them one at a
time. It's not just going to spit out
10 slides for your slide deck. So what you're saying is Manus
(16:42):
actually spits out the whole deck?
Holy crap. I'm going to share my screen.
All right, so this is Manus, This is my prompt.
This is the whole thing. Make a PowerPoint deck for me,
showing the growth of the RV park industry.
Like no details, nothing. OK.
And it worked, It worked, It worked.
It worked for 12 minutes and 31 seconds and it spit out this 12
(17:05):
slides. Holy crap where did it get that?
Hold on, go back to the first picture dude.
I could be wrong. Have you used this picture
before in an RV deck? It looks really familiar.
It looks like one that I have used, but I don't think I've
used this exact one. OK, but I mean look at this.
Oh my. North American market growth.
RV ownership. This looks like the deck that I
(17:26):
use for the same purpose. Now, does Manus, do you have
this deck in Manus? Like is it pulling from stuff
you've already created or is this?
No. Totally.
Honest about RV park stuff? Yeah.
I didn't teach it anything. Isn't that crazy?
So when I use ChatGPT it spits it out as APDF.
It's uneditable. Is this editable?
(17:47):
It's spit it out as APPT as a PowerPoint.
Really. Yeah.
Holy crap man. So side quest here, how much are
you using Manus right now? Not that much.
I just started playing around with it when this new feature
came out. I was using it a month or two
ago, but I I just moved on. What were you using it for a
month or two ago? It's like agentic like features,
(18:09):
it's kind of acts like operator,but it wasn't as good in my
experience. And so when I didn't like that
output, I just I went back to church PT.
What are people using Mantas forright now?
Mantas. Yeah, that's what I said.
I think they're using it for this a lot.
I don't know. It's a Chinese owned company and
(18:32):
so kind of TBD on what's happening with our data here,
but we'll see. I think it'll work out, but
we'll see. How big is the pitch deck
industry? It's multi billions the
corporate world is built on. Pitch sticks.
Pitch sticks. Yeah, it's, it's the modern idea
conveyor that we use. So there's a guy on Twitter, I
(18:56):
think he's the pitch deck guy and his whole business, this is
it. He goes and he scrapes like the
filings, SEC filings when peoplelaunch a new fund or anything,
you know, people that are about to need a really nice pitch deck
and he will do cold outreach to them in the right time, the
right moment and say, hey, I sawyou launched a fund for buying
single family homes. Congratulations, I'd love to
(19:17):
make a pitch deck for you. Here's My Portfolio.
What does he charge? 3500 bucks?
Love it. And they look really good.
They're really good. So Nick Manus has an API who's
stopping you or anyone from building an agent or an app or
anything. It could be just an automation
with with Gumloop or Lindy or Zapier that scrapes all those
(19:40):
SEC filings. The funds that are launched
permissionlessly goes to Manus through the API and says here's
all the information for this fund.
I want you to make a really goodlooking pitch deck according to
these pitch deck. I wanted to have this design and
send it back to me. Boom, you get it.
Then you go to Apollo or any other feature that finds
(20:00):
people's contact information andyou cold e-mail the pitch deck
to them with watermarks that says, hey, congratulations on
the fund, I made you a pitch deck.
Would you like to buy this? Here's the first five slides.
Could that could be fully automated as of today?
OK, so there's the pitch deck guy and then there's Matthew
Gore. I had him on my podcast.
He's the pitch deck cowboy. He has like that's his business.
(20:21):
Has he built pitch decks for people?
Funny story really quickly, I was on my way to Bogus Basin in
Boise, ID. I'm buying a freaking burrito at
this random place. Order it.
I'm picking it up and this guy next to me goes, are you on
Twitter? And I was like, why yes, I am.
He's like, are you Co founder's Nick?
I'm like, yeah, yeah. He's like, I'm the pitch deck
guy. I'm like, wait, what?
(20:42):
He's like, yeah, yeah, yeah, yeah, I'm the pitch deck guy.
I live in Dallas, but I'm actually from Boise.
So just like random. Yeah, randomly on my way up the
hill, I met him, which was. Which is crazy.
Yeah, totally random. I think that the ability to
convey a story is like so valuable that most people think
of pitch decks as like, oh, a major presentation or I'm trying
(21:04):
to raise money, or it's like something super serious.
I think just the ability of pitch decks to convey any idea
is so valuable that, yes, there's the SEC filings, but you
could permissionlessly offer this as a marketing agency as a
way to get clients. You're like, hey, Ethan
Driscoll, This would be a perfect, like, cold outreach for
him just to permissionlessly have Manus take all of his
(21:26):
ideas, put them into a slide deck by scraping whatever the
list is that he's created off the Internet and just generate
business that way. Yeah, I, I love your idea.
I also think there's a bunch of different applications you could
use for for cold outreach. 2 Twomore ideas You heard of
slideshare.net? No.
They get it's either 5 or 10 million visits a month.
(21:46):
What the fridge? It's just a website for
uploading your slides, your decks to like to be hosted, to
be viewed, right. So if you go to slideshare.net,
you can see like Netflix's famous 125 slide presentation on
culture there. And then you can also see, you
know, Econ one O 1 from Alabama Community College, some student
(22:06):
group. They uploaded there so they
could present it in class and like it has 3 views.
Hundreds, if not thousands of slides are uploaded to
slideshare.net every single day.This company sold to LinkedIn in
2012 for 112 million and then they sold from LinkedIn to
scribed for I don't know how much it wasn't disclosed.
So this is a big business. It's a nine figure business.
(22:28):
Why not like do the same thing with that, like create a
permission listing that anytime a new slide is uploaded to
SlideShare, you revamp it, you redesign it.
So instead of making one from scratch for the SEC filings,
you're just redesigning them, sending them back to the
original creator and saying, hey, you may like this better,
you may not. If you do buy it for 20 bucks,
200 bucks, whatever. And automate.
(22:52):
And automate it or, you know, teachers paying teachers,
teacherspayteachers.com. It's a place for teachers to
sell their that layered lesson plans.
You can go on there. It's such a good name.
Hey, teachers paying teachers.com, What do they do?
Well, they pay teachers. You could go on there and like
DM, mass DM all the teachers that are selling lesson plans
and say, hey, why don't you sella a pitch deck alongside the
(23:15):
Lesson plan. Like a Lesson plan is great but
you'll probably sell even more if you have a pitch deck
version. I'll convert that for you for 20
bucks. What are we doing or.
Anything with a Lesson plan, People who teach church, like
they'll say we have a a Lesson plan for Sunday.
What do we typically do? We're like, maybe we have some
on our iPad, whatever. We could just easily upload an
outline and be like, hey, createa PowerPoint presentation based
(23:36):
off of this. What are we talking about right
now? Dude?
What? Are we doing?
What are we doing? Why are we?
Talking right now, we need to begrinding building.
Oh my gosh. Maria close.
Clear. The rest of my day, I don't know
how, Maria. Is.
Can I build off this for a second?
(23:57):
Please. All right.
So it's not easy, but what you and I are talking and we're
like, oh, dude, why wouldn't we do this?
All we have to do go to Chachi Butiti, enter this prompt and
then it, it automates here and all of a sudden we've got a, a
product. I still think for many people,
the prompt is like the hardest thing for them to get past.
Like to think creatively enough in order to get the prompt to
(24:17):
give you the output that you want is incredibly hard.
And I found something this week that I want to share with you.
It's going to be stupid simple and you're going to make fun of
me, but I think has one of the best pieces of advice I've ever
gotten when it comes to prompting.
Are you ready for it? Yep.
All right, here we go. All right, so you've probably
seen this, this was a Reddit post and this guy's talking
(24:38):
about how he uses the LLM to generate prompts for him.
So in the example that you just gave where you're sitting there
and you're like, I want Manas tocreate me a slide deck based on
this information, but like, how do I even give it the specific
information that I want in orderfor it to create the slide deck?
He's essentially teaching you how to create the prompt by
(24:58):
using the LLMS to create the prompt.
So the first thing that he does is he tells the LLM generate A
detailed prompt engineering guide.
The audience is, let's say graphic designers.
So you can fill in the role there, and the role might look
like book authors, software developers, customer support,
whatever. And then he pastes in five
examples of how I want my promptto work.
(25:18):
So what he's doing there is he'sgiving it context to draw.
He's helping the LLM understand what exactly he's looking for by
providing it with five examples of products that he really
likes. You got to give it more context.
Got to give it more context. All right, and now this is
probably the best advice I've ever read.
This next one I then instructed to quote, generate a prompt that
(25:40):
could have generated the examples outputs and include a
better set of examples. Dude, that blew my mind when I
read it. So, so he's sitting there and
he's like, hey, this is who you are, you're a graphic designer.
Here are 5 outputs that I really, really like.
Now I want you to to reverse engineer it and think of what
the prompt would be that would create the outputs that you're
(26:02):
currently seeing. So it's getting the LLM to think
about, OK, what are all the different elements within the
examples that are given to me that I need to get to in order
for this prompt to generate something similar to what I'm
seeing. What do you think of that?
It's genius. Like I want to hang up right now
and test this. Have you read this before?
(26:24):
I feel like this is. Something that you've seen?
OK. It blew my mind because it's
like so simple and yet it's going to be so.
I've already started using it and the outputs that I'm getting
are way better than I've ever used, than I've ever gotten
before. It's pretty.
How long does it take though to put all this together?
Which part like the five examples?
(26:45):
Yeah. Obviously it depends on what
you're trying to build. I've been using it for content
and like podcast ideation. So from that perspective it's,
it's been fairly easy because I,I kind of know exactly what I'm
looking for and I can go and find the outputs.
If I was starting a business from scratch, like we were just
talking about the slide deck generator, I would probably have
(27:06):
to go and find just five slide decks that I think look really
good. I don't know where I'll go and
look. So then the key is to say, all
right, look at these 5 slide decks.
Give me a prompt that would produce something that looks
like these five. Yeah.
And again, that's what we were originally talking about.
It may not even give you a good output, but it'll at least start
you down the road of messing around with this.
(27:29):
So you're going to see the the next steps actually build on
this. So then he goes to a new chat
and he says generate A detailed prompt evaluation guide.
The audience is graphic designer.
So what it's telling me that LLMis basically help me audit a
(27:50):
prompt and give me a guide that I can use to audit that prompt.
He then takes that what it's created and he pastes it into
the original prompt and then he tells it generate 3 improved
alternative prompts based on this based on the evaluation of
the prompt and then he picks thebest 1.
So effectively what he's doing is he's like getting the LLMS to
massage it to a point where he likes it and it makes sense.
(28:11):
Now, there were some comments onthis that I thought were
hilarious. One guy, he was like, I honestly
thought of that guy who you loveon Instagram who's always making
fun of the business gurus. And the, the quote was like, and
then what I do is I write a prompt that asks it to evaluate
the prompting that I wrote in order to evaluate the prompt
that evaluates prompts. It was like, wait, what the
freak? So he does get pretty meta here,
(28:34):
but you're using the LLMS to leverage your time and and get
you to a place where you can actually create these prompts
really quickly. I'm so excited to use this.
This is perfect. Oh, it's pretty cool, right?
Yeah, I've heard about asking the LLM what to ask it, like,
all right, all right, here's what I want.
Here's what I'm looking for. I want to make a a delicious
(28:54):
tomato salad. What should be the prompt I give
you to give me the best output possible?
That's like level 1 and this is like level 7 of that.
There are two things that I can remember that have been like,
oh, that's genius and always have stuck with me when it comes
to AI. One of them you gave me, which
was remember this when you're working on a thread and it tells
you something or it spits something out and you're like,
(29:15):
Hey, remember this makes a mental note.
The other one is this, it's like, OK, I would like this
graphic design or I like this adcopy.
What's a prompt that you would have written in order to get
this output? Like that to me is genius
because you are reverse engineering.
And that's that's effectively what we're trying to do anyways
when we see stuff that we're trying to copy.
Oh. That's gold.
(29:36):
It's gold, Jerry. All right, what do you know
about Perplexity Labs? I know a lot but assume I know
nothing. OK, cool, cool, cool.
Yeah, yeah, yeah, yeah, 'cause Iknow a lot, but I know our
audience doesn't. Yep, Yep.
So I didn't know much about it until earlier.
I I've been playing with it quite a bit.
You do have to have the $20.00 amonth perplexity plan.
(29:57):
But at first I was like, oh, this is just deep research.
Like Perplexity already had deepresearch.
It was fine. This is just more deep research.
No, no, no. So here's how perplexity is
different from deep research. So let's say I want to know that
like the market landscape of allthe keto apps out there, I want
(30:18):
to build a keto app and I want to know how profitable are the
ones that already exist? What's their cost per
acquisition? If they're using Facebook ads,
how many downloads do the top five keto apps get?
I want to know all this stuff. I'm going to go to deep research
either on Perplexity or on ChatGPT and give it that prompt.
And it's about 10 minutes later,it's going to give me thousands
of words with like a really thorough output.
(30:41):
Here's what Perplexity Labs does.
I just prompted this and my prompt was build an interactive
chart showing the correlation between sourdough baking and RV
traveling in a post COVID world.I don't know why, I just that's
all I can think of. So it gives you like multimedia,
(31:01):
it gives you the information you're looking for in different
ways. It will use charts, it will use
PDFs that you can download whereappropriate.
It's kind of the difference between having an intern go do a
little research for you and sending you a memo, and then
having a six figure employee that's like a researcher going
out and spending weeks aggregating all this data,
(31:23):
knowing when to use a chart, when to use text, when to use
pictures, when to use APDF. That's kind of the difference
between perplexity, so. What we're looking at, if you
can't see this right now, is there's an interactive chart
what's on the X&Y axis. Sourdough market size in
millions and then time period. Yeah, and then market size and
RV shipments in thousands. So RV shipments in thousands is
(31:45):
over here. Oh, so I'm assuming that the red
is RV shipments? Yes.
So that's seasonal basically. Yep.
Interesting. Goes up in the summer.
And so you asked if they were correlated.
Does it say it's correlated? Yep, it's -.004 weak.
Negative correlation suggests inverse relationship during
certain periods. Interesting dude.
(32:07):
It gives you the charts. You can visualize it, but then
it actually gives you a write up.
Yeah, like just as and here's a non interactive chart showing
the same thing. And then the proprietary charts,
these are not like they wouldn'tscrape the web and found some
Yahoo who made this and post it.It made these, yeah, it, I think
it took, just took raw data fromthe web and made these itself.
(32:29):
And look, here's a, here's a PDFor a CSVI can download.
So sometimes it gives you a CSVPDF, whatever you want,
sources on everything. So how do we use this?
How do we make money on this? Got to make money on this.
Got to figure out how to make money on this.
I mean, to me, it just goes backto this allows you to leverage
(32:52):
your time even more. As companies get bigger and they
have more resources, they can throw more resources at smaller
problems. So this would be an example of a
of a smaller problem. It's like in the past, I'm
growing a company, I need to do all this research.
But you, Chris Koerner, as you're doing all the market
research, you're kind of like internalizing all of this stuff,
(33:12):
right? You're looking at different
websites, you're reading different blog posts, whatever.
You're reading different charts.This gets you all the
information that you would have otherwise gone out and just
surfed the web for quickly in a format that you can digest.
So to me it would be a business built around servicing decision
makers, people who who have highleverage positions that are
(33:33):
making decisions all day. And that's kind of where you get
to as you grow your businesses. Dude, like look at this prompt
result asking about the keto appmarket.
It's all right here. All the iOS ratings, the annual
price, how many weekly downloads, weekly revenue,
active users like and it has 37 different sources for this data.
Like this is crazy. This would be like if if I got
(33:55):
in this ideation frenzy late at night as I often do and wanted
to. I really wanted to launch this
keto app. I would stay up all night, like
just doing this research that itdid for me in 11 minutes, OK?
Here's a lead magnet I think would be really cool.
Digital agencies. They could be doing anything.
They could be doing automation, they could be doing AI
implementation within a company,they could be digital marketing,
(34:18):
graphic design, anything. If they use this and they
generated a report for every company that's on their list
because every, every one of these businesses has like an
ideal customer. And let's just say your ideal
customer is importers of pastel fabrics.
I don't even know. You generate that report one
time and you e-mail it to them as a lead magnet to get your
(34:40):
foot in the door. So then it spurs the next
conversation because you're giving them something of real
value where it's like, hey, nextyear, these are the projected, I
guess I don't even know tastes that are going to be in the
pastel market. I know you make these decisions
every single year around this time.
Hopefully this helps you. The automation guys at
automationla-bs.com, whatever itis, they're going to read that,
(35:01):
they're going to see that you're, you can track whether or
not they open it and then you can easily follow up with them
next day, next week, whenever. But it's a really cool lead
magnet that like that's really valuable.
We talked about this before IbisWorld where I work, they're
charging $2000 reports. That's a $2000 report
potentially right there that youcan just use as a lead magnet
now as a lead magnet. Think about that I.
(35:22):
Know like if you're a consultant, like how big of
retainers could you get by leading with something like this
dude? Any business that wants a
customer name a business that's B to.
BH FAC Supply. Cool HVAC supply?
I'm an HVAC supply company and I'm looking for HVAC business
owners. Why don't you just send an
industry report for their specific market to every single
(35:45):
HVAC owner? Yeah, it pulls in real estate
data. Yeah, it pulls in, yeah.
Yeah. Here's where the growth trends
are of your market overall. Here's the age of properties in
your market. Here's the, I don't know,
typical home value. Here's the supply, whatever it
gives, like the whole entire thing.
Thought you might like this for your market.
Hope you enjoy it. Let me know if you have any
questions. You're just like giving value to
(36:07):
them right away. And who are they going to call
the next time? I mean, not necessarily
everybody's going to call, but just in terms of a lead magnet
and building trust with people, that would be phenomenal in B to
B sales, Any B to B sale. That's a freaking cool idea.
That's a freaking great idea. All right.
What do you think? Please share it with a friend
and we'll see you next time on the Kerner office.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.