Better Sign Shop Podcast

Learn how to build a better sign and print shop from a few crusty sign guys who've made
more mistakes than they care to admit.

Conversations and advice on pricing, sales, marketing, workflow, growth, and more.

You're listening to the Better Sign Shop Podcast with your hosts, Peter Kourounis, Michael
Reilly, and Bryant Gillespie.

Hello.

Welcome back to the next episode of the Better Sign Shop podcast.

I'm your host, Brian Gillespie, joined by Mr.

Mike Riley, the burrito bunny, is that what that says?

Burrito bunny.

gonna be both someday I'm going to show up in my bunny costume and the burrito costume.

You do have a burrito costume now.

unrelated project, but I'm excited for that one too.

Well, Mike, the last AI episode was rather ominous.

We're going to try to stay off of that tone, but you know, if we end up there, we'll see.

In this episode, I think we'll just jump right into like what's happened since the last
episode.

AI is changing like super fast.

I know you've gone down a rabbit hole.

But I don't have a super structured outline.

Maybe we'll just talk about where we're at AI, February, we probably need to date this
one, right?

February 20th, February 19th, my daughter's birthday is tomorrow.

That's exciting.

Yeah, and then we'll just dive into some of the tools that we've been using because this
shit is gonna be outdated in six weeks again.

I was just going to say, refresh our memory for context.

How long ago was it that we recorded the other one?

was only not even two months ago.

Six, seven months ago, six, seven weeks ago.

Yeah.

Yeah.

I mean, in, that short amount of time, what

What has happened to you, my friend?

Me?

Well, I've, I've like just completely like fully gone, like tin, tin foil hat, like AI
rabbit hole.

I'm so in, I'm in so deep at this point that like, I can't, I don't know which way's up.

It's crazy.

It's such, it's such an insane technology.

Like when we did our last episode, you know, I would say I probably kind of was a little
bit more than a casual observer and using the tool every day, but just on the very surface

level.

And since then, I've really, you know, dug extremely deep into the tools and a lot of, you
know, prompt building stuff and really trying to understand how the technology functions

under the surface and, and what it does more.

And the more you learn about it, like the, the crazier it gets, like, it's like, it is
like a rabbit hole.

It's like, the farther you dig into it, trying to understand it and look for answers and
like,

the more you uncover and it just continues to add, you know, more to the rabbit hole.

It's never ending.

Like in the rate that it's changing and growing is, is, is mind boggling.

I, you know, I think, was it three weeks ago that deep seek was released and

Where it shows you the thinking behind it, which is super.

Yeah, instead of just saying thinking, it actually shows you its train of thought.

And it is so human like.

I mean, when I when I first saw I mean, it's kind of like, I'm used to it now.

But when I first saw it, it was it was actually unnerving.

You know, because it was actually was referring to itself in the in first person.

And in it was an actual such a human

train of thought like like stream of consciousness thinking.

And to have seen I think I think that was kind like the aha moment for me when I when I
actually saw what is going through its quote unquote brain behind the scenes.

I realized this is something that's way way above.

I don't know my pay grade.

I don't know.

But yeah, I mean, since we last talked about this deep seat came out.

Open AI is released.

three.

quad hasn't done anything.

Still, same old clod, but they're still beating everybody somehow.

Yeah, it's still a very good model.

You and I are very much like, it felt like over the last three to four weeks, maybe last
six weeks, since you started diving into this a little deeper, we're locked into this hold

my beer thing of like, hey, here's what I did today.

And then you're like, no, no, no, wait, hold my beer.

Have you seen this?

Right before we started recording for everybody listening, Brian and I were just kind of
catching up a little bit and talking about like, you know, my my insane work schedule.

And I'm usually up till two or three in the morning every night.

And I feel bad for Brian because I know when he sits down on his computer in the morning,
he's got just like this waterfall of messages from me on Slack, like, dude, look at this

crazy thing that I found on Reddit last night.

You know, at this point, it's like just like scrolled right past it.

But I mean, I feel like

I can remember the meme of the guy from It's Always Sunny in Philadelphia, know, with all
the, you know, pictures on the wall and the strings connecting.

I mean, that's kind of how it feels because when you start digging into it, it's, it is,
it's, it's such a revolutionary technology.

and so I don't even know what the, what the word for it, what the word for it is, but the
more, the more I understand about it, the more I realize it's not, it isn't a novelty.

And it's not just a tool right now we want to look at it as it's a tool, it's a tool we
can use to help us do our jobs more efficiently.

And that's true.

But under the surface, this is a lot more to it than that.

It's it's more than just a tool.

And I don't know, I'm not supposed to make an ominous.

I'm excited to see where that goes.

I mean, it's hard not to wind up there, I waffle back and forth between like, you first
experience some of these things, like these new models, and you're like, holy shit, this

is going to change everything.

Then you get a little used to it and you understand some of the quirks and you're like,
okay, now I can figure out which box to put this in.

But when we were talking about this shit a year and a half, two years ago though, it's...

I never thought it would be this far, this fast.

what's to say that it's not going to continue to just accelerate and get there at an even
quicker pace?

I think that, you know, I think it is.

if you, I wish I don't have a screenshot I could share, but maybe you can cut this into
the video.

There's a, I'm sure you've seen the, the, the growth curve of, you know, the intelligence
of the models and, and, and as it, you know, as it progresses, every model is released

closer and closer and that curve is essentially on an exponential, you know, vertical
incline right now.

I mean, it's at the point where it's

it's taken off like a rocket and I don't You know, is it ever going to slow down at that
point?

I mean, what's the I don't think anybody even understands how it works So how do we know
what the theoretical limit of what this thing's capabilities are?

You know

All right.

So we waxed it like a high level.

Maybe we'll walk through, which tools are you using on a daily basis now?

then, maybe you share one, I'll share one.

We can go back and forth and maybe play it that way to try and...

We could probably do this all day.

The listener probably hopes to get some value out of this other than just listening to us.

wax poetically about it.

what's your main driver as far as AI,

That's a that's a that's a loaded question.

I don't have one.

I use literally on a daily basis.

I use chat GPT and Claude and I have them both open on my browser always and they're never
closed at this point.

I mean, that's to the point where

I want to say I can't make a decision without them.

But I find myself sorry, I got myself into the habit of using them to bounce any idea I
have off of, you know, whether it's a business idea, or something I'm trying to design, or

what to have for dinner, you know, I find myself turning to them as maybe like a an
advisor almost, or

appear that I trust to, you know, give me objective opinions about something that I'm
doing.

But I find that, you know, Chagy Petey and Claude definitely give different responses.

So depending on the nature of the question, I'm asking it all, you know, kind of make a
choice between which one.

And then I'm also using mid journey a bit for image generation.

But, you know, to be honest with you,

Image generation is not, it's not where it needs to be or where I think we all want it to
be.

It can create some really beautiful, amazing images sometimes, but as a, you know, more of
a technical designer like me, beyond ideation, it's not really too helpful.

Like right now I can't generate output that I can actually use for anything, but it can
help me generate ideas.

but yeah, have you been on mentoring lately?

Have you checked it out?

I have not.

It's interesting, you had to be on Discord to do it.

They've really improved the UI, they actually have their own UI now that you use versus
having a good Discord, which is a lot better, and it's got a lot of settings and you can

tweak the randomness and how many hallucinations it allows and stuff like that.

And it's better than it was, and the image quality's good.

It's not that the images it creates aren't good, it just...

something is as niche and technical as signage design like I do it.

It can't come up with anything that's truly production ready, I guess, for lack of a
better term.

But it is great for ideation.

But I think that graphic designers at least specialized graphic designers, and logo
designers to I think I think branding designers, graphic designer in general, that we

don't have a lot to worry about anytime soon.

Like AI is not going to take our jobs because it just without

without a truly human subjective element to it.

I don't think it can ever fully reproduce what we we see.

Or emulating it right?

That makes sense.

But as far as like writing stuff, I mean, if you're writing copy for a website, business
plan, anything like that, like it's unbelievable at that.

And as time goes on, these models get better and better and better.

you know, I, I asked it to write a blog post the other day about using CorelDRAW to design
sections for aluminum extrusions for like sign cabinets and stuff, like sign composition.

I was asking it to, to help me with that.

It's just, you know, it's, it's mind boggling.

Like it spit out a three page blog post that was

so technically accurate.

It looked like I wrote it, you know.

And that's something that's not like common everyday information.

I couldn't Google that information and find it.

So at this point, where's it

It's buried in there somewhere in the machine.

I know right like it's it's the ghost of the machine right

Well, let's zoom in on one specific thing.

Are you going to chat GPT and Claude, or are you using this typing mine thing that you're

yes, typing mind is the online platform that I use.

It's kind of like a, how would you describe?

It's basically like looking at this and I'm afraid to share my screen because I'm going to
get robotic again because I got like 35 things going.

But maybe we just weave this into the edit.

Do me the trash, you're mine.

Yeah, you try to see if it works for you.

Like I've got like $1,500 or $2,500 worth of cameras set up.

like I've got a fancy MacBook that's maxed out from three or four years ago, but it still
just bogs down.

It just doesn't like anything.

Probably all the cables and connections going.

But basically, typing mind is what I would call it just like a GPT wrapper, of like, this
is, you could pick back and forth between these different models.

All these different models are in there.

And it's nice from the perspective of, you've just got one app that you're using that has
all of your conversations and your threads and.

Obviously you've used it more than I have, so maybe share a little bit about your
experience versus the individual apps.

Yeah, so this one requires having a an API account for whatever AI that you're using.

So in my case, I've got a cloud AI account or API account and then a chat GPT API account.

So instead of going to their, you know, their web interface, like, you know, just going to
chat GPT calm, I go to typing mine here and select like, want to use GPT four turbo, and

that will

access chat GPT for just like I was on their website, but it's through this couple
benefits to it is it's a little bit faster, especially with Claude, you don't run into

messaging limits.

Like, I don't know if you use Claude a lot, but you don't run into a lot of do you don't
you ever run into the messaging limits on there?

Not usually, no.

Sometimes it'll switch into the concise mode, which is kind of frustrating.

But yeah, I, for whatever reason, I don't know, maybe it's just the peak time of day that
I'm trying to use it, but it was almost always in concise mode for me, but I hit messaging

limits where I had to wait five or six hours before I could even talk to it again, which
was really frustrating.

So the AP using the API eliminates that.

the cost is a little bit cheaper for, you know, per chat as well.

And then there are additional plugins and stuff you can, you can add to this.

One of the, one of the coolest things though, that I really,

like me see if can find it here

Oops, wrong place, kind of forget where I'm at.

There we go.

So you can adjust for each different model or globally, the temperature and top P top K,
all these sliders here basically control how accurate creative it is, how much it

hallucinates, how technical versus how

you know, free format.

Like, creative, gets right and then you can put your own expand this.

So you can put these, you know, kind of like instructions for like, this is how I want you
to respond.

This is what I want you to do.

Basically, basically, yes, you're kind of telling it how you want it to interact with you
above just its default, you know, setting, I guess, which is which is pretty and you can

do all this in, you know, the

the regular web interface as well.

But I think you have a little more control over here.

So for one thing for me, it's just a little bit more convenient to have it all in one
right here.

I've got my cloud, my cloud chats and my my GPT chats, you know, right here together.

There's not a phone app for this, though, which kind of sucks because I use it on my phone
constantly as well.

And you know what mean?

So that I either have to go to the website for it I just end up opening the the cloud app
on my phone.

And then those chats don't show up here.

So that's kind of a kind of sucks.

Yeah, I think where this gets interesting is for like sign shops is obviously like the,
know, chat GPT and Claude, I think they both have like team accounts, but I don't think

it's like a shared workspace, right?

It's just like, like, hey, here's the team account.

And I think this typing mind does have like a team account where.

It does.

Yeah.

put a bunch of stuff together in one and it has like knowledge base type stuff where you
can upload documents and add that for context and have it be across the organization.

Because that's struggle in a business setting is like, how do we do this collaboratively
sort of thing.

Right.

Yeah, I haven't played around with this because I don't have a team to share it with here.

But yeah, it does.

I remember seeing this when I was signing up for it.

It's a paid add on for but yeah, I mean that I could see that being a barrier of entry to
especially businesses adopting this if you're you you have a creative team or something

and you're or just a, you know, a copywriting team or a web development team or something
like that.

I could see having the ability to collaborate on these.

across multiple people and teams would be hugely beneficial.

So that I think that that probably is a selling point for a lot of people.

But yeah, it's it's a it's a pretty slick app altogether.

And there's a there's a lot of these in typing.

Mine is just kind of the first one that I found when I asked chat GPT, what's the best AI
desktop platform to use?

And that's a that's actually a good one.

I don't I hardly ever use Google anymore.

I've just sort of naturally migrated to asking it of chat GPT or Claude, but Claude isn't
as good as a search engine as chat GPT because it doesn't have internet access.

But I find if I, know, something that I would go to Google and look up, I look up on chat
GPT and the answers I get are usually a lot more specific to what I'm looking for.

And obviously there's no, you know, ads trying to sell me shit I don't want.

So I like, obviously like, I think we talked about it maybe in the last one or one of the
previous episodes.

Like I'm working full time at a company called direct us right now.

This is unrelated to signage, but like one of the things that we're seeing like marketing
wise is a like Google search is still number one for how people find us.

And then, you you've got like referrals, which are a big piece of like, Hey, I worked for
a company that used direct us that used this previously and I brought to a new company.

or somebody at work gave me a recommendation.

But behind that, GPT and the LLMs are next as far as recommendations.

And it's crazy how much that has grown where it's almost like another layer of SEO that
I'm concerned about of how do you get your company indexed into these models so that you

come up in recommendations, right?

And that's gonna be, for the sign shop owner, that's gotta be an important thought as well
of how do I get access to this?

When people start asking questions like who's the best sign company in Bluefield, West
Virginia, which they type into Google now, but how do I get my name surfaced through that?

It's a really good, good question or good point.

I've never really even thought about that.

But yeah, I mean, there's a I don't know if it's science or witchcraft to getting rank
high in Google searches, you know, like, I don't think anybody knows that really talk

about something that nobody understands behind the scenes.

But you know, I mean, that's a massive industry, just SEO and, you know, trying to get
people on the first page of Google.

And yeah, like how

You know, on one hand, nobody knows how the Google algorithm works or how that really
happens.

But I think we all know that it's, you know, probably money that has something to do with
it.

But is that going to be the case with AI?

Like, can you monetize that?

Can you, is it a pay for play type of thing where the more you pay, the higher, the more
likely it's going to be?

How's that going to work?

I'm gonna go for this and see if this actually works.

If I get robotic, let me know.

All right, so it's been impossible to miss the AI overviews from Google, right?

So if you get a search like this where it's like, what is the sign company, it'll generate
this AI overview based on results it's come up with.

Now, as far as I could tell, it's still saying this is experimental.

But when you start doing like sign company in Bluefield, West Virginia, you get like the
traditional places widget, I guess you could call it, instead of like the AI overviews.

So what I'm assuming is happening right now because like ads and search are big Google
business, especially local, right?

it is refraining from doing like an AI overview that says, okay, here's the three sign
companies in Bluefield, West Virginia, and who, you know, based on reviews or something

like that, and analyzing the website content, who you should pick.

But, you know, to me, it's not necessarily a stretch to assume that like, Google may
eventually do that, or somebody else will do that, or, you know, people will just go

directly to chat GPT or Claude, right?

And I've

That's an interesting one.

I've not actually done this really.

Who's the best sign company in Bluefield, West Virginia?

And some of these models are probably trained not to answer this.

So I would be curious if you throw this through typing mine, what you would get out of it.

Or through the API.

some.

accounts.

Well, I've got, I've got, chat GPT pulled up here and just type that same, same exact,
question in there.

Let me share my screen.

I'll show you what it showed me.

there you go.

who is the best sign company in Bluefield, West Virginia.

And that's what we got.

So it kind of came up with a similar result to what Google did, gave us a map and then a
little brief.

Beckley, West Virginia.

That's a good sign company.

The old Beckley, West Virginia.

that's the fast signs location.

Okay.

me try that same thing in typing, man.

I'm just curious if that's going to give us a different result.

Change that to...

Yeah, and the distinction there is obviously like Claude is not searching the web.

it's, you know, like whenever they've done like the training for this and like the system
prompt is probably saying, hey, like if somebody asks for a recommendation and you don't

have context, don't provide it.

You know, I'm sure you could, I've seen like the Twitter threads or the X threads where
people like try to jailbreak the system prompts and get around it.

Yeah, I've been successful at that a few times.

To an extent, like there's a limit to what I think you can realistically get away with.

But if you talk to it, if you're really sweet to it, like it sounds hilarious and stupid.

But if you handle it with kid gloves and you talk to it really polite, this is actually I
find this more true with Claude than chat GBT.

But if you're really nice to it, really polite to it, and ask things a very specific way,
you can convince it to kind of skirt its

it's rules a little bit to a point.

And it's really interesting when that when you when you get to do that.

Crazy.

Yeah, I mean, you're basically asking it to roleplay a different in a scenario.

And when it when it thinks it's pretending or playing and not actually being, you know,
giving you a serious answer.

Sometimes you can get it to to jailbreak itself.

And it's fun.

So I put that exact same prompt into typing mind using GPT four, and

back was something totally different.

different, which is so weird.

I mean, same account, same model, same everything.

So that's strange.

I wonder if the API has different.

Yeah, like ChatGBT is a, like there's certainly like a wrapper around that that's using
the APIs, but it's basically like its own service with its own limitations.

And same for Claude, It's interesting though, like especially once you get into the web
search and the tool calling and like all of that stuff.

All right, not to get too far off track.

So let's circle back to text generation.

are you using this for?

Copy on your website.

What's your flow look like for that?

So I've got a Kind of a project started in chat GPT Actually, I have one in chat GPT in
claw depending on kind of the context of what I'm gonna be writing But that that projects

which anybody who doesn't know about projects in either in any of these tools a project is
it's like a it's like a chat that You self-contained that's kind of self-contained and you

can give that chat specific instructions

just for that particular project you can give it like reference files so if you like a
certain writing style say or you have a bunch of technical you know documents or reports

or something like that that you wanted to use as reference material you can add all of
that to your project and it instead of just going out in the broad web and it's you know

it's it's training it will heavily look at and consider your your project references.

cool, you got one.

Yeah, yeah, yeah.

Let's go.

So I did that AI workshop like eight months ago and it's it's totally outdated, but like
I've got Claude set up and I've got, Brian's Ghostwriter is the name of it, right?

And again, just like you said, what this does, it's available on both chat GPT and Claude
and like most of these have it now, but like for specific tasks, this is the best way that

I've found to get great output.

And it all starts with the system prompt where you're setting up the instructions.

typically, I'm leaning towards the role play scenarios that you're talking about.

Here's your role, here's your task, follow these guidelines.

So for example, the Bryant ghostwriter here is, hey, you're a world class ghostwriter, 15
years of experience.

You create compelling content that educates and entertains.

You will write whatever I tell you to, basically.

And then you give it more specifics, right?

And to get these rules, the funny thing is you can get meta on top of meta on top of meta
with it, and you can take something you've written or take a transcript of you talking to

somebody, paste it in here, ask it to analyze and give you the tone and the voice and the
style, and then feed that back in.

And you can get so close to...

just how you write.

It's almost, it's uncanny, Where, like, you're asking AI to write something, especially
Claude, I find it's great at writing, where I get really close to, like, I look at it I'm

like, I don't know how I would wrote this differently if I was writing it.

And I don't know if that's just, like, seeing it on the page already, or, but like, I read
it and I'm like, it's pretty close.

Right?

So like tone wise, use confident tone, be approachable, use plenty of real world
analogies.

I've got like a very distinct style.

I try to like in emails or anything, I write like I talk.

So that's what I've told it here.

One of the other big like tips I would say if you are using this for any type of writing
that you're to share, make sure you explicitly tell it like

Avoid buzzwords, avoid marketing language.

Call out some of the things like delve or revolutionize, streamline, like some of the
cliches, like tell it not to use that.

I've basically got this line in every GPT prompt probably of like avoiding prolix and
lengthy explanations of basic concepts and stuff like that.

Avoid like the stupid terminology.

then, like some workflows, I may even have it.

like a separate pass where like you tell it to write something and then come back and say,
like now, like humanize this, like break a few rules just to get it to sound more like me.

That's a good idea.

But as far as what I've given this, right, is just a sample article.

Like this is something that came from the direct...

and just giving it a sample of the writing.

And within that, once you do it, it's pretty close, right?

So I can go in here now and say write an outline for a blog post about, I think this was
trained on technical content, so the five tips for getting more SEO juice for your

website.

say?

no, I'm just agreeing with you.

Yeah, so like headline wise, right?

Like five SEO tips that actually work and won't make you feel gross.

Like, and that resonates with me.

Like that's something, that's how I would, that's almost how I would phrase something.

And this is just the outline.

And then, you know, to complete this thought, and I'll let you talk in a minute, Mike.

Basically like, I would, like if I was trying to get an article out of this or like copy
for a landing page, right?

Then you just proceed to, you know, either like, Hey, I'm good with this outline.

Or you know make whatever edits you want to the outline and then you know let's proceed
with the intro paragraph And you take it brick by brick instead of trying to get like a

one-shot type of scenario Because that's where you get better results or at least in my
experience

Totally agree.

Yeah.

And it's I mean, if anybody's tried to do this in any AI app, you know, a year ago, let
alone like two or three years ago, you can see how much better it is just by watching that

that write that I mean, it's, you know, it used to be we were we were talking about this
the other day that that was a Jasper AI Jasper.

Yes.

Yeah.

And it would it would write like two sentences at a time.

it was it was yeah, you can get it would it would spit out two sentences at a time they
were

barely coherent sentences, but we were still like, my god, this is amazing.

And now, I mean, look at it now, you know, what I was gonna say is, you know, with the
project, like you just showed there, I've actually got a project with a bunch of prompt

examples.

So whenever I need a, you know, beyond just like a surface level type question, if I
needed to do something really in depth for me, like write a blog post or that I was trying

get it to do that 3m vinyl color to Matthews paint color matching thing, you know, a
couple weeks ago,

stuff like that, where it's really more of an in depth thing you're working on, you can
have it write your prompts for you if you kind of give it this is what I'm trying to do.

Write a really extremely detailed comprehensive prompt for Claude or for chat GPT.

I'll tell it like I'll tell Claude write a prompt for chat GPT.

Weirdly, it doesn't get jealous.

I kind of thought it would.

Like honestly, like really like not joking, like I'm surprised it doesn't say like

something about why are you using another AI?

Right, but you know, it'll spit out this really detailed long prompt, you can copy and
paste.

And I find that that goes a really long way as well towards, you know, getting, getting
better information out of it or more concise or detailed responses.

So I do that a lot.

I have it write my prompts for me a lot, or rewrite my prompts, maybe would be a better
way to put it.

You know, I would be upset if we didn't just touch on the effect of something like this.

I guess this delves back into the doom and gloom, right?

Obviously, the speed and ease at which you can create content like this kind of sucks on
some level, right?

Because I'm...

I know probably a lot about SEO just because of my work and like the websites that I build
and things like that.

But like on some other topic, like I could presumably get you 1,000, 1,500 words that make
you think that I'm an expert in about five minutes or less with one of these tools, right?

So what does that do for...

Like writers, obviously, huge problem if you are a writer and you're listening to this and
that's your profession, big issue.

But you know, like business in general, if somebody can just whack a mold this stuff out,
article after article, like does the entire web just become noise and nonsense?

You know what?

I think a lot about that and it's a, it's a, it's a really, it's, it almost feels like a,
like we're standing on the edge of a precipice, right?

And like, we have a choice to turn around and go back to, go forward and fall.

Did we go to the level?

know, I think if I was a copywriter, there, there are some, there are definitely some
professions right now that I would say not.

not to be like a fear monger, but they should be afraid.

Like they really should.

And I think that copywriters are probably at the top of that list or pretty darn close to
it because at this point, like the writing is, is better than 99 % of copywriters out

there.

Like that's the, that's the cold hard truth about it is it's a better writer than people
are anymore.

And that's going to be a difficult thing for a lot of people to come to terms with, you
know, for that type of business or that

profession copywriting.

I think it's gonna, it's gonna really quickly.

don't know if decimate is the right word, but it's going to weed out the change farmers.

Yeah, and only the really, really good ones are going to be left in the ones, you know,
copywriting, I think is probably going to be more about mastering prompts than it's going

to be about, you know, at least marketing, copywriting and business copywriting, not
necessarily writing, right, The great American novel.

But, you know, that some, like I said, like think

graphic designers I think were safe for a while because it lacks the subjective human
element still.

don't know, researchers?

I don't know if you've really had a chance to look a whole lot at the new 03 research
model that's available in the research.

The deep research, yeah, in GPT Pro.

The $200 a month model, I don't have that, although I've been really tempted to subscribe
to it for a month just to play with it.

But I've played, I've read some of the deep research results that it generates.

And I mean, we're talking very, very detailed, extensive PhD level research papers that
generates like a hundred page documents, like, like not just a brief description, like

it's generating massive, massive documents that are so exhaustively researched and, and it
can generate something that would have taken, um, you know, a PhD candidate.

a year right, it can do that in 15 minutes and it's going to be better.

So I, you know what I mean?

I think it's going to, it's going to shift the way we consume knowledge, I think.

And there's a part of me that's trying to remain really hopeful and think that this is
going to push humans to be better at what we do.

It's going to ultimately give us more accurate information.

versus, you know, humans right now, we don't have guardrails the way you know, an LLM
does.

So you know, we can publish, say anything we want, like look at the state of the world
right now.

You know, I mean, will it will it rein in misinformation?

Will it will it

Yeah, like to play devil's advocate, right?

There's always been like those content farm websites that push out like dribble anyway.

like, obviously, yeah, like, that was a popular one for sure.

But like, yeah, like, it's like the great power, great responsibility.

think sure like you could use this.

And you know, when it comes to like Google, obviously, they've got both hands in

both jars, guess, of like, we're doing AI research, but we also get funds from ads on
keyword searches.

And it's like, which one do you prioritize?

like, like for, one question that.

somebody made the analogy I saw the other day of that exact thing Google, Google funding
AI research and in.

mean, if you were to if you were to bet money, I would I would put all my money on Google
because they've got the deepest pockets by far if anybody in this game.

But somebody made the analogy the day that like Google funding AI research is the
equivalent of like pharmaceutical companies funding cancer research, you know, right,

there's a significant conflict of interest there.

that shouldn't be ignored in google's case anyway

But it also kind of handicaps them as well, right?

If you are one of the leaders in AI generating content, and the new Google models are
phenomenal as far as the context that you could put into it.

It's like vastly, you can load a million words into, well, I don't know if there's a
million, but you can load a ton of words.

You could load a whole book into some of those new Google models and ask it to summarize
or...

you write a blog post outlining, et cetera.

But if their models are some of the lead-ering generative models, you've got to account
for AI-generated content in search results, right?

How do you square the two?

It feels like a snake eating its own tail, you know?

It's definitely weird.

And that makes me wonder, how much of our own data is Google using?

Think about how big, how much data Google has, right?

Between Gmail and Google Docs and Google Search, every keystroke that's ever been entered
into Google, they own.

And they can use that to train.

their AI model that chat GPT can't and Claude can't, you whether they are or not, we'll
probably never know.

think they claim they're not, imagine, you know, the additional level of information that
they have access to by that.

mean, that's, that's almost a scary thought when you think about it.

Yeah, yeah, but I mean, it raises a good point that it may be back to that precipice I
just mentioned, like, like maybe the bigger precipices.

who's going to own this technology, right?

Like, is it it is is it a race to eventually be one single unified AI to rule them all and
there's no longer GPT and chat clad clod and is there just gonna be one eventually?

Or will there be multiple like there is now will be more, but who's going to own them?

And who's going to say what the guardrails are as far as what kind of information they
disseminate?

And is this technology that belongs in

private hands are in the private sector, or is it technology that is so powerful that it
needs to remain open source and free and available to everybody?

Almost like the internet, you know, like, nobody, nobody owns the internet.

Because nobody saw it coming.

But I feel like if if there was an opportunity for somebody to buy the internet, you know,
early on, you bet your ass it would have happened, right?

Like, of course.

So

the hell I mean, Elon Musk has tried to buy open AI last week for what $98 billion and Sam

trying to raise what like five hundred billion dollars is what Sam Altman

Yeah, they're at like 200 billion now for their their whatever star Starship project where
they're building a new AI infrastructure But yeah, you know in rumors are that Elon tried

to buy open AI To force them to force their board of directors into keeping it Open source
versus go go in private.

Yeah, which I don't know.

Maybe that's a smart move on his part I'm not sure yet, but I feel like

The billionaire class is probably hustling right now to figure out who can who can own AI
control.

Yeah, and that's the race for me.

Like for us, the way that I'm looking at this is like, I don't even know that we're
climbing up a mountain or going down a mountain because a lot of my tasks are getting lot

easier with these things.

And I can do more of, like the way I'm looking at it is like,

let me get the noise or the busy work out of the way.

If these tools can help me with a lot of that so that I can get back to what I'm good at,
which is at the core, I think one of my skillsets is basically identifying a problem and

then just being tenacious about solving it, whatever that is, whether it's writing a bit
of code to solve a problem on a website or trying to figure out how to price.

set of channel letters with a spreadsheet or in ShopVox or whatever.

These things aren't totally autonomous to that point where you could just say, here's the
problem, go solve it.

on that respect, we're okay.

it feels like we're, I don't know, like the best analogy for it of like, I'm coming down
the mountain, like through some fog or something, and I can't see exactly where.

going.

Right, feels like things are getting easier and you know, also like, you know, there could
be like a desert between me and where we need to be, you know?

Yeah, there probably is I mean, there's so much I know I when ultimately, when when the
people that created these technologies openly say like, we don't really know exactly how

it works.

I think it's not a red flag, but I think it's something that should give us all pause and
and and be cautiously optimistic as we as we you know, hike down that mountain through the

fog, because we don't know what the next step is going to be.

And, you know, again, not to be

I'm trying not to be super political about this, but politics are becoming part of it.

And there, you know, we can see what's going on in Washington right now.

And people, people are jockeying for power around AI.

mean, if you look at Zuckerberg and Sam Altman and Elon Musk and Peter Thiel, mean,
they're all, they all have their hands very deep in, in political world now.

And that's a hundred percent driven by the potential to make profits off of this AI
technology.

So

It's definitely something that nobody should ignore for sure.

But yeah, I don't know.

mean, I think.

It's hard to like square the two, right?

Like at a high level is like, hey, I get like super concerned with it.

But like on a day to day level, it's like, I've got three kids.

I've got a full time job.

Like we do the podcast.

We do, like I do some consulting work.

Like we've got it.

You've got a couple of different revenue streams that you're managing.

Like, you know, at some point you're just like, I'm not, I'm going to use the tool like
that's in front of me to get the job done and like stop thinking about

Yeah, dude, it's a godsend.

It really, it really is just for like administrative stuff, day to day stuff, writing
tasks, having something to bounce ideas off of, like, there's, there's a lot of like

staunch anti AI holdouts out there, like, I'll never use this, it's going to take my job.

And I refuse to support something that's, you know, killing industries.

And I understand that sentiment, but it's not realistic.

mean, you know, people said that about steam engines, or, you know, the internal
combustion engine about

you know, it's going to take the steam engine locomotive engineers job and whatever.

Every time there's new technology, it changes things.

And some people might lose their jobs temporarily.

Some people might have to learn new skills.

I mean, it's just kind of how evolution in the world works.

To not embrace it in some way, I feel like is extremely foolish.

Because if you don't, the people that are embracing it are going to leave you in the dust
really quickly, like real quick.

to just make it Yeah, and you're just making your life harder by not using it like it's
it's here to stay people who don't believe it's here to stay and they think it's a fad or

something.

You know, how'd that work out for the people that said the internet was a fad?

And you know, this isn't going anywhere.

So to to actively avoid using it and not embracing it is just kind of kind of shoot
yourself in the foot.

I mean, it's it's not a wise move to make.

I don't think at this point, I mean, just have to, you know, be responsible about how you
use it.

But it will make your life so much better.

Really does.

guess that's the crazy thing about it is it you know, one of the things that I found
completely like unrelated to business with AI.

But I use chat a lot.

just call Chad for some reason.

I use Claude a lot like a personal advisor, almost like a therapist.

I mean, you know, I'm I'm I'm big on like the whole mental health thing and you know,
taking care of your somewhere.

Yeah, I'm excited for that one.

I'll wear my bunny suit for that one.

You know, it's it's great as like a therapist if you're struggling with something if
you're having a bad day if you're I don't know you're dealing with Trying to change habits

around something that you do or you know, like I'm trying to go to bed earlier I'm tired
of being up till three in the morning every night What are some actionable things that I

can do to get to bed earlier?

You know if you prompt it, right?

It's it's an incredible

I don't know whether term to use other than like therapists like that.

I therapists kind of makes it sound like you're pouring your soul out to it and

Yeah, no, I get it though.

For me on the, like just on developing websites, right?

Or like, you've got a couple projects in the works here, obviously at my full-time job,
but like this, more signage thing that I've been working on.

Like I'm not like a senior level architect, right?

Like I know enough to develop a solution, but like some of those questions about like,
hey, what's the best language to use for...

the back end of this or what's a mental model or a pattern to use for how to set up
something like personalization on a website.

It's nice to have somebody that can instantly give you an answer, whether it's, we could
debate whether it's a great answer or a good answer or a shitty answer, but closing that

feedback loop of like, hey, I have this question, I don't have the knowledge on it, let me
talk to somebody who.

who has more knowledge than me on it.

Yeah, it's like having an expert available 24 seven, you know, in most topics.

Or, if not an expert, pretty damn near an expert close enough to an expert that you can
you can get very far along on what you're trying to accomplish.

I mean, how we use it.

Like, two days ago, my wife was like, I don't know what to make for dinner.

All I have is like, some chicken breast.

Like, what do I do?

I'm like, she's like, I don't start pulling out recipe books.

I'm like, I don't know.

I'll ask chat GPT.

And it I'm like, give me

Give me 25 recipes for chicken breast that are outside the outside the box, not not super
perfunctory and every day.

I'm happy to run to the store if I need any ingredients.

The shit that it spit out like like crazy recipes.

You know, I mean, it's just it's just stuff like that.

you know, that something so simple and small instead of pulling out 20 recipe books and
spending the next two hours pouring over that.

you had fifty recipes to choose from in literally forty five seconds is that going to
destroy the publishing world

I don't know.

That stuff was helpful though, like outsourcing some of the decision making.

You got to be careful not to outsource too much of it.

you like like why like like what if we if we lighten our I mean think about you know
Albert Einstein working you know patent office so that he you know occupied his brain with

just boring mundane tasks throughout the day so you know in the background his brain could
you know synthesize information and work on problems that he was working on and they say

that was part of the key his brilliance was that he

you know, he immersed himself in something very boring and work a day.

So would that benefit humans to like offload as much of our mundane workload, like
cognitive processing, decision making basic stuff like that?

What do eat for dinner tonight?

Just let chat GPT tell us what to do.

Would that benefit humanity so then we can think about like creative things and like
solving humanity's bigger problems?

Or does that just usher in the era of terminators?

we let it think for us completely.

I mean, I don't know.

I think it would be fun to do an episode where it's just like chat GPT roulette, where you
just do something like this, where you just like, I want you to open up chat GPT and

you're going to like, we're just going to pick a message, right?

Just for context, I was working on a tool for my job, my full-time job, and I wanted to
have an ASCII art

bunny as part of like a command line utility.

I just have to say, your job is fucking ridiculous.

You get paid to do this.

Well, I mean, this is just like a small portion, right?

And it's like, well, like I wanted to have an animated bunny that like blinked and set a
message when you ran this thing.

yeah, like blending all the different experiences of like design and UX, but also like
human interaction or user experience, like all of that, like, you know, the way it

manifests itself sometimes is like, hey, create some ASCII art bunnies that I can use,
right?

But like, this is the type of shit that

Like it's very quick to, okay, now I got this.

I can put this in there, right?

I've got three or four different varieties.

You know, some of these are pretty wild as far as what they come up with, but you know,
this was something that I had to solve and this was like two minutes to solve it versus

like Googling or me like noodling in the text editor of like trying to craft these
characters.

So that would be a fun one to do, but.

And that's honestly like, that's a really great use case for it.

That it's just a mundane task that doesn't really necessarily involve a ton of skill.

It's just time consuming to find that crap.

It's out there somewhere, but

All right, let's bring it back to the tools.

There's a couple that I wanted to cover.

I know you mentioned DeepSeek.

That's one you should probably experience on your own.

Maybe we'll just do a...

I got this thing up, and we'll just see if I've got...

Let me just pull up that last one that we did, right?

If you've not used DeepSeek...

The hosted version does send your data to China, so you might want to be cognizant of
that.

This is one of the...

And don't put anything that's going to jeopardize national security in here, but for this
type of stuff, It's very interesting to watch it kind of think through this, right?

So I'm toggling this.

I think that's how you do it.

I'm asking it to write an outline.

And now it gives me like this stream of consciousness of like...

how it's thinking.

And it's very much like a stream of consciousness.

Okay, so the user wants an outline for a blog post about five tips.

Let me start by breaking down what they're asking for.

Maybe I should consider the target audience, main challenge, blah, blah.

So like it goes through this whole exercise.

wait, scroll, see what says wait, the user might want like that.

What amazes me about this is it's not linear, right?

Like you would think that, okay, it's a computer, it's a robot, like it's just going to go
in very linear order.

But it doesn't.

You can see it thinking through things and saying like, wait, that doesn't make sense.

Or wait, this might be a something that potentially could affect the outcome of this.

So I should maybe look at it.

You can see it step back.

and consider itself or consider what it's doing as is doing it.

And that to me is what the absolute mind boggling part about it is that I did not expect.

Yeah, wait, the user might want the outline to be SEO friendly itself.

like here it comes with, you know, like the listicle style outline.

Here's the five tips to boost website SEO juice, plus a bonus, plus a conclusion.

Yeah.

So it's super interesting.

Um, you know, I think one of the, one of the other interesting ones was a, like, I thought
this was super interesting where if you give it,

like a misspelling, right?

What is the capital of West Virginia?

I'm sorry.

So like it goes through this whole process.

I noticed that Virigina, it's misspelled.

It should be Virginia.

West Virginia is the capital.

Now I know the capital.

So like it goes through this whole like interesting process to like, at one point even
talking about the Civil War, right?

Like to get there and then like it does get there eventually, but it's just wild, man.

Yeah, yeah, it is it is really.

It's really I've seen a few examples to where people have put in like math problems where
depending on the your interpretation of of DOS or you know, which what do you perform

first and multiplication parentheses or the you know, and there's two different schools of
thought on how to answer that question.

Like if you give it one of those types of questions, watching it think through math
problems or more technical things like that is really wild to you because it like

it does some strange consideration.

It's really, this was, think this seeing how, how deep see things like I think I said this
earlier, like it was, it was a major aha moment for me when I saw that thinking process

and realized that it's, it's not an algorithm behind the scenes like Google.

That's, know, I the more I understand about how it functions, the more I understand about
the physical

the physical server farms and and and the neural network and you know the way GPUs process
these requests and The more I'm convinced that we've essentially figured out how to make a

synthetic human brain and it's just kind of still in the infancy learning phases because
it When you see the way something like deep seek thinks through problems like that

and you look at the way it's taking individual words or parts of words and parsing them
and turning them into tokens and then the way it processes information and thinks is at

its root exactly the same way we think.

Like a complicated mess that's hard to unpack,

Right, yeah.

it's, I'm more and more convinced that this literally is.

brain just in a different form the brain that's in our head.

I'm convinced though that the structure functions exactly the same way.

The the the way it thinks and arrives at conclusions is exactly the same way that our
brain thinks as far as sending neurons around and nobody really knows how brain thinks

either.

I genuinely believe that this is it's an actual brain.

You know, like, I don't at the end of the day, like,

what matters, what the brain is made of or what the brain does, right?

And if you look at what it does, there's very little difference between what it does and
what our brains do.

I mean, the biggest difference is, right, it's just not in a physical, biological body
that it has, you know, this brain body connection like we do.

Therefore it has no sense of self or self awareness, obviously, but I think it thinks, I
think it actually thinks.

I really do.

think it thinks I think it.

That's the title of this episode.

It's alive.

It's alive.

But but you know that brings up then the really creepy question What what does it mean to
be alive?

What does it mean to be?

What what does thinking mean what what?

You know it does that make it Sentient if it's thinking Sentient does that mean does that
have to mean that it's in a physical body and aware of that self or does it does just mean

that it?

Processes the information and stimuli around it

That's it.

That feels like another fucking episode.

I love these, man.

It's good to do this.

All right.

we even talked about science at the science industry at all.

I'm not sure that we have.

Let's see.

All right.

I'm going to run through some of these.

If you're consuming the audio version of this, I would highly recommend you check out
YouTube.

I know Mike mentioned the image generation side of it.

Still kind of like for what we do in signage, it's not super helpful.

Maybe for ideation.

You know, we should we should clarify it depends on what you're doing with it, right?

Like if you're a lot of digital printing and wrap design, or you're like doing a wall
mural, and you need more of like a, you know, I've definitely used it to generate some

absolutely phenomenal photograph type images and paintings and things like that, they're
using signs.

It is phenomenal at that.

Like, you know, it just like it's coming for copywriters.

It's also coming for artists, photographers, photographers.

Yeah, absolutely.

But again, like as far as like,

more like vector type things and technical things like that.

It's nowhere near where it would need to be to be useful in that realm for for sign
designers.

So just want to throw that out there.

I didn't want to sound like I was like, completely trashing its output.

It's just not not perfect.

And this is a good example.

For us, these are blog posts on the Directive's website.

These are all AI generated.

And the logo is obviously a hair, it's not a bunny.

But we have this bunny theme going on.

And it makes good thumbnail images for this.

And you can easily create a theme off of this.

I think these are all generated using mid-journey.

But it's great for stuff like this, like supporting elements for either blog posts,
websites.

If you're not using Mid Journey, you could check it out.

can give you more details on that, or we can add that in the show notes.

One of the other tools that I've been using, and you will have to sign up for a GitHub
account to do this, the tool is called Replicate.

And it looks very, it's a very developer-oriented tool at first, once you get into it.

But it does have like,

Playgrounds where you can go in and like test out like some of these models now like these
are like open source models that you know if you're super into a Like AI you can download

these and you know you can dive down that rabbit hole and learn how to set all this up
locally For me, I just like playing around with these and in like their playground or you

know checking out one of the nice things about replicate and

I may have showed some of this in the last one where I can go in and train my own model.

And this one is the flux model where basically I can enter something in here.

What's a good sign description?

Create a...

modern sign a backlit monument sign with a brick base.

Monuments on it with a brick bass for Mike's Barbecue Brick Bass.

And this will, we'll see what happens, right?

For four cents an image, we'll see what comes out of this.

And, you know, okay.

I don't know where fasty.com is coming from here, but you can see these models are getting
really good with, I'm almost scared to put that into the search bar.

like, for those who...

Like if you're listening to this, it's basically a monument sign with what looks like some
channel letters on it, some decorative elements and a brick base.

And it kind of looks like it's shot with a macro lens or something that makes it look like
it's a Lego type of scenario.

Yeah, you know, like, in all honesty, like that image is, is beautiful.

Like, it's a really, really, it's a decent image.

And in the design of that sign isn't terrible.

I'd probably do things a little bit differently.

But it, you know, so like, from an ideation standpoint, like, I'm just, have designers
block, and I'm not exactly sure what I should do.

Like, it, it's neat, it'll throw out ideas at you that you can kind of run with a little
bit.

And I use it for that a lot.

Yeah, you know, it's a magic of like prompting.

If you go in and you actually train one of your models, like you give it reference images.

So like you could train this on your own work and potentially use it to generate similar
concepts or, you know, there's probably some other more devious ways to use it.

But it is something that I use quite a bit to generate like supporting imagery and it's
good for that.

You know, I've never played around with this one.

I know you're always raving about it.

I think I'm going to try subscribing to this and see if I can train it on my design work.

Um, you know, I've got thousands and thousands of images I could feed into this thing as,
as training material.

I'm curious if I was to do that, how close it could emulate my design style.

Okay.

For part three of this discussion, I'll report back on.

One of the other models I've been using on here is called Recraft V3 SVG.

And basically, this one is very interesting because it will give you a high quality vector
output.

It is much harder to control.

again, like with most of the recommendations we've already made, you need to, like if
you're to try to use this, shrink the scope and like go like, hey, I need an element,

whether that's, you know, whatever it is.

versus give me a full logo because it's not quite there yet.

But the nice thing is, I think I shared a few of these with you.

The images that it generates are SVG and the lines are very clean compared to what I've
seen inside Adobe Illustrator when you generate vectors.

Yeah, you sent a few of these to me and I actually was like, blew my mind how good the
vectors actually were.

There, there were almost no weird stray points or odd funky jaggies or anything like that.

Like you would get with an auto traced image.

The ones that you sent me were every bit as clean as anything you would download from from
you like Adobe stock or I stock or something like they were, they were as perfect as it

could get, which was really surprising to me.

Like I didn't think it could do that, you know, with vectors.

Yeah, we'll see.

See what I can get out of this.

Again, this is just off the cuff.

I fully planned to prepare all of this, but maybe I'll put that in the next AI workshop
that we do.

right, so this guy is kind of missing a...

He's got like a turtle nose thing going on.

But as far as what's generated here, right?

This is pretty solid.

mean, considering that it's SVG.

and I could take this and quickly clean it up and edit it.

It's amazing that you can...

Yeah, it's amazing that that is almost usable like you could could cut that in vinyl right
out of that without any Modification without any cleanup at all like that's pretty it's

pretty incredible The on-demand clip art basically like it's it's wild

I'm not going to open Illustrator because then it will really go robotic, but just trust
me on this.

This is an actual SVG file that is downloaded.

can confirm.

Insane.

What are the other ones that I wanted to touch on?

I don't want to get into Replet just yet.

Notebook LM.

this one's cool.

This one is super interesting.

One kind of use case I would harp on this is training new people.

If you've got somebody new to the sign industry, notebook LM from Google, basically you
can add sources to it and it will basically create Mike and I wrapping on a specific topic

for you for half an hour's worth of podcast.

So training content.

You know, I like that immersion style of training, especially for hiring somebody new in
as a CSR or maybe a sales rep that doesn't have deep industry experience with very little

input.

can go very far with something like this where like remix a podcast on channel letters or
something like that.

You know, I was thinking about about this, as a supplement to a blog on a website or
something like, nobody reads blogs, right?

Like, but I feel like if somebody that I, you know, I followed for whatever reason, had a,
interesting business or blog that I would want to read, I just don't want to sit there and

actually read it.

Like, I feel like you could, you could put that blog into notebook and create.

embed embed, you know, that audio file on your blog as well.

So would you rather listen to this blog and actually have a podcast episode where, you
know, they're discussing that blog post, not just reading it verbatim.

I mean, I feel like that'd be really cool from a marketing standpoint.

And people would be more likely to continually go back to your website to listen to the
latest blog post.

I just generated this whole conversation based on your blog post from...

I saw it somewhere.

Where was it?

2019?

Managing a remote design team.

I should update my blog.

Yeah, so this will send me a notification and when this gets done, it'll be a full
conversation between two people, like discussing the content of this blog post in probably

a lot more detail than what Mike has gone into on his blog post.

And it'll sound just like you and me talking like well not like you me but like least you
would never know it wasn't a conversation between two people.

That's the cliffhanger we'll leave them with at the end like is this actually Mike and
Brian doing the podcast or is this just our AI avatars?

Are you able to train it on specific voices?

haven't really dug that far into this other than just-

So the the notebook LM I think you just kind of like like the voices that it uses it just
kind of gives you what you get you or you get what you get kind of thing There's another

tool that I have used called.

We'll just rattle off all these right here at the end 11 labs and basically like this one
is uncanny that Like they have voice cloning which used to be like super expensive and

like crazy

I can give this probably like a minute and a half of your voice, Mike, and it will come
out with something that's pretty close in the end.

I that.

I'm gonna have to stop sharing for a moment and catch up.

Yeah, you're a robot again.

again.

Hopefully all that came through.

Are you hearing me in real time again?

Close enough?

Yeah, yeah, I didn't lose your audio.

You're just a little dancy.

I got it.

We'll see, what are the other ones that I wanted to talk about?

Notebook LLM, 11 Labs is an interesting one.

One other, this one's not quite AI, it's not like an LLM.

And it is a little pricey for like a smaller shop that's doing, if you're doing like less
than.

500,000 a year or something, you're kind of limited on your tool budget.

But this app, Clay, basically what it is, it's kind of like Google spreadsheets, but for
outreach and sales and enrichment.

So imagine you've got a list of contacts in a spreadsheet and you push a couple buttons.

and you can enrich all those contacts and even generate personalized messaging for those
contacts and connect that to tools like Zapier or other places.

So that's what Clay is.

It's kind of like a mix between Airtable and Google Spreadsheets or Google Sheets, but
also with the ability to take an email address, find the person on LinkedIn, grab that

data, generate a personalized message to that person.

based on what you found.

And they've even got tools where you can actually send those messages out.

like sales-wise, like go to market, like revenue, ops, those are all terms in the tech
industry.

But like if you've got an outside sales reps and you want to feed them like relevant leads
with relevant information, Clay is definitely a tool to check out.

don't know if should show you this.

think I've never seen that one before.

That's really cool.

I I I can think of several sign companies that sales departments could use that right off
right out of the gate.

That's pretty slick.

It is kind of pricey.

But I mean, for a bigger shop that's, you know, if your shop has a sales team that that's
not that's just a drop in the bucket there.

Yeah, if you've got outside sales reps and you need to feed them leads and potential sales
versus kind of the inbound model that most shops swallow, this is a good tool for that.

Yeah.

What about perplexity?

Are you using that a lot for as a search?

Oh, did you have it open already?

didn't see it.

I think it's an interesting one, right?

Because the claim is, hey, it goes out, grabs all the sources, and gives you a nice
summary.

And I think it does that better than GPT for some use cases.

It's intended to kind of be an actual search engine or like as a replacement for Google,
right?

Is that my understanding?

Yeah, it's basically what it is.

It's like, hey, this is the AI search engine.

like, you know, what's the best?

Let's just try that.

Right.

What is the best sign companies in Bluefield, West Virginia?

And that's not my great grammar, but like it goes out and there's better sign shop.

We're number one.

So, you know, again, a better sign shop, technically not a not a sign company.

We serve sign companies.

But I do get the list here, but this is, again, what makes these different.

What makes these different?

Let's ask you that and see.

The main difference is Lamar focuses on, yeah, is this helpful?

Maybe, maybe not.

I think it depends on what you're trying to get out of it.

feel like it, you know, like, like we were talking earlier about Google, like I could see
this being helpful, because you don't know on Google anymore, what's paid ad and what's,

you know, legitimate, organic search result, something like that.

Like, I feel like this is sort of like, strips all that, that crap out of Google searches
and just gives you only the context you want.

The nice thing about this is it's like citing sources.

So at least you can kind of like verify some of the information of like, how did it
synthesize this?

Right?

It looks like a lot of it's coming from Yelp or the yellow pages, in turn, like these are
obviously like SEO optimized pages.

that's it's searching some search engine, whether that's Google or Bing or something, and
then summarizing the results.

right?

that's so is it actually giving you legitimate search results?

Or is it just synthesizing illegitimate paid search results?

Right.

you know, like most of these are from Yelp, which, you know, it's a local search.

So maybe that's why, but you know, it's something interesting to explore in your own
market.

you know, I would, I would definitely be like a takeaway, you know, we'll save like the
three takeaways at the end of this one because we're already way long, but yeah, I would

be running queries like this in chat GPT and something like perplexity and seeing what
comes back in your local market and just seeing

Number one, like are you even remotely close to the top of the list that it comes up with?

It's an interesting one.

see a lot of people using perplexity, so I feel like there's something to it that hasn't
clicked with me yet.

I need to try it again.

Alright, last button.

I don't even know if I want to get into this one.

Like we're already at like a minute, an hour, 20 minutes.

Maybe we save that one for something else.

Replet, this is the one that I just put you on.

We gotta show it, we gotta show it.

It's so cool.

All right, so Replet and there's a whole slew of these now, which as a person who does
identify as a developer or at least capable of doing development, these things are insane

to me where basically it's like an AI agent that will generate whatever tool that you ask.

And lovable is one that's more like non-tech oriented.

like if you don't fashion yourself a

want to be a developer or somebody who doesn't want to get too far into the weeds, you can
try lovable.

So that's just lovable.dev.

And basically you're interacting with chat GPT sort of thing, but it's actually generating
a working app for you that you can publish and share, which is super interesting.

So.

Yeah, and I just, uh, I, I tried lovable out recently just to see what it could do.

Um, and I'm constantly converting dimensions, right?

Like, you know, one 27.34, I need to figure out what that is in fractional feet and
inches.

Right.

So there's my calculator is always up and by my side and I must convert 50 dimensions on a
daily basis on average.

And it's tedious and I hate doing it.

So I had lovable write me a little web app that I could, you know, plug in whatever number
it is.

And it would just be the conversion for me automatically.

And I could click a copy button and I can paste it into CorelDRAW and it's done.

And the conversion, super fast, super simple.

Nothing like that existed on the market at all that I could find.

looked everywhere and I would have been willing to pay for it because it would have saved
me so much time.

within an hour on Lovable, and the majority of that time was just kind of like tweaking
and fine tuning the appearance of it.

I had this functional calculator.

Yeah, and I use this thing every day.

I think it's a great example of how, like we were talking earlier, the app development
world is probably going to change drastically because there's no longer a need for

somebody to make that product and sell that dimension conversion product.

I had a very specific need that would help me make my job easier.

And in an hour, I had a tool that I can use for it.

I made my own tool.

Yeah, and somebody else's needs might be slightly different than mine as far as converting
dimensions and they can build their own tool that works just for them that does just what

they needed to do.

I think that's going to be a really revolutionary thing that is when more business owners
kind of wrap their head around how this works and realize the value that that brings.

I think this part of it, these little app generators is going to be mind blowing to
people.

I guess I spoke too soon about lovable.

lovable.

It is not able to fix it.

Because it usually will.

Yeah, usually it's able to resolve it if you just say this, you know, fix this error,
which is kind of neat.

It's actually need to watch it troubleshoot itself.

And I had to do that a few times when it was doing my my little app and it worked through
the problem and

The interesting part here is this is actually generating source code that you could take
and run somewhere else.

I don't know if this thing will actually be able to write itself and actually show this.

It's not coming up with anything.

I don't want to waste too much more time, but considering this is one of the examples, I'm
disappointed this one didn't work right out of the gate.

Basically, it's writing code for you.

you know, it's kind of nice for like simple apps like Mike is talking about.

I wouldn't try to replicate something like Shotvox on here.

Like you're going to get frustrated, especially if you're like a non-developer.

The other one that I do think is more interesting of like for a technical person or more
technically inclined.

Again, like you're still going to be like looking at some code.

is this is called Replet.

So Replet is basically

It's like a full-on development environment, but they've added this agent to it where
again, I could just chat with this and say, we're creating a I Didn't even honestly.

I didn't even give it this prompt right so again if you're listening to this I Was trying
to show this to Mike and I said, okay like Mike is a freelance sign designer I want you to

build a freelance sign design collaboration web portal platform

It's like seven words, right?

That's what I gave it.

This agent goes in and creates a scope of work based on that.

And then it starts knocking this off one by one, basically autonomously.

Like I never had to touch any of this.

let's see if I can actually get to the...

I'm gonna hit run and see if we can actually see what it built in like 10 minutes.

I'm going to restart the server.

What does this do?

Good, I've broken all these.

Let's fix this thing.

Good job, right?

Let's demo.

Well, you know, I think this is actually kind of good to show people because this is part
of it.

I mean, you're going to have to troubleshoot as you build this.

mean, it'll do the troubleshooting for you, but you kind of have to like push it and hold
its hand like, Hey, this isn't working.

Can you find the error and fix this for me?

Um, you know, when I was playing around with this, I, know, every once in a while I'd get
hung up on something and I had to kind of show it that it was hung up, guess, for lack of

a better term, but, um, it is cool that it

You know, if something goes wrong, you're not on the hook and try and figure out what
happened.

Restarted start application restarted start application.

yeah, of course.

This is not gonna work either after I talk this shit up

It's the way these things go.

Thinking deeply.

Thank you.

Thank you for thinking deeply about the problem.

Basically, this was a super lightweight application that had like a queue, a job queue.

But the interesting part is like it spins up the infrastructure for you as well.

So like behind the scenes here.

Go to the database.

The database is paused.

that's probably, that's the issue.

Like I'm not on a paid account.

I've got a paid account.

want me to share my screen and show what?

Yeah.

Yeah.

I had to do the same thing.

Yeah.

Hang on here.

Yeah.

I think I use the exact same prompt when you share this to me.

Okay, yeah, so like you actually got to like a login page and shit.

Uh, yeah, yeah, I actually, I've got, see, that's probably gonna do the same thing to me
now.

Let's see.

How do I?

Hit run up there at the top and see what it does.

And that should, I want to say it shows in that web view pane.

there we go.

User error.

So.

And yours is actually...

Oh, shit, okay.

You got way farther than I did.

You spent more time with it.

So it creates a new project.

There's the project.

Yes, so that's as far as I got with it.

But you know, I've got a login page.

I've got a dashboard that'll show you all your projects in a card view.

You know, some filters here, a new project button, you know, some, some, you know, basic
stuff.

I mean, it's not like this is a functional app or anything, but this took 20 minutes to do
and just, you know, I told it what to do and that's what it did.

And as far as I just stopped, I was just kind of playing around with to see what it would
do.

And I mean, it's, it's pretty wild.

I mean, it's not like this isn't something where like, you know, it's going to build out
an entire app with one prompt and in an hour, right?

Like it's, if you wanted to build out an entire, you know, truly functional project
management tool that has a lot of bells and whistles.

And, you know, maybe not the complexity of something like shop box, but something that's
still full.

You're going to have dozens and dozens of hours and it's still, I mean, it's not going to
be fast, but

you won't have hundreds and hundreds or thousands of hours in it like if you were building
it from scratch.

Yeah, it's super interesting and being able to see where that's going to go as a developer
building stuff is interesting to me.

But also, there's this interesting space of, like you mentioned, of, hey, we need this app
to exist or even stuff like integrations between applications where I could leverage LLM

or an AI to build that for me instead of...

spending $25,000 to have somebody build that integration is super interesting.

Yeah.

Did our, did our notebook LM thing?

Did it finish?

Are you able to audio through this?

Yeah, I don't know.

Could be interesting.

Click to load the conversation.

How do I share?

Can I share audio?

I don't think so.

I can upload a potential file.

I don't know if I can make it listen to the computer audio or not.

If I...

You know, when you share your share that share your screen or share the tab, I think when
you share it gives you an option to share audio that that tab audio as well.

I feel like I saw that

I can download the audio.

Let's see.

Stop share.

Where's the share button at now?

Share screen.

To share audio, share a tab instead.

Okay.

All right, let's see if this is gonna work.

Can you hear that, Mike?

No.

you hear it?

I hear it in my microphone, or in my headphones here.

We'll just share this in the show notes.

We'll figure it out.

Yeah.

We need an AI podcast producer.

How about we stay on?

Yeah, sure it does.

Alright, we're an hour and a half in.

We can keep doing this for hours.

I feel like this is a good stopping point.

We've hit on a lot of different facets.

I feel like this needs to become a reoccurring thing every couple of months to try keep up
to speed.

Yeah, it'll change.

Always good to catch up any parting shots like

It's an overwhelming topic.

I don't even know where like, just, you know, I think that I genuinely think everybody
really should should embrace this and dedicate some time to learning it and understanding

it.

Hop on YouTube, you know, learn some, watch some AI, you know, introduction to AI videos
or basics of prompting, you know, learn, learn prompting, learn the art of prompting, get

really good at that.

And, you know, I think when people really

start to understand that part of it and play with it a little bit more.

They'll recognize how powerful of a tool this is and how much it has the potential to make
their lives easier.

If they, you know, approach it with an open mind.

I like it.

I like it.

Very diplomatic.

You no doom and gloom, so I'm trying to-

Doom and Glade.

These tools are super interesting, right?

If I was putting an ROI on the time I've invested in these tools, I would take the advice
of Mike and I and skip some of the deep rabbit hole stuff and figure out what are the use

cases.

I would try to figure out what are the use cases that I can leverage this for without a
ton of investment.

as far as like, personally, hours and hours and hours, and I'm like super deep into it.

I've dragged Mike into it.

It's super interesting, but also I'm like, I'm not building and selling signs on a daily
basis anymore either.

like I can imagine like if that was, if I wasn't at my computer all day, every day, it
would be hard to have this depth with it and know what's good and what's bad.

I would...

You know, start somewhere, obviously, but...

would encourage anybody who's you know, if you're dealing with some sort of a predicament
or business problem you're trying to solve or whatever it may be something related to your

business, just ask Claude or chat GPT, you know, hey, this is what I'm dealing with.

This is what I'm trying to figure out.

All you have to say this is this is the problem we're trying to figure out help me solve
it.

You don't even have to necessarily ask it a specific question.

Let it let it have a conversation with you just start having a conversation with these
things.

and and see where you end up i think that's kinda like the best way to you know maybe two
to learn about it or to take get more comfortable with it is just to start talking to

about whatever it is it's on your mind as strange as that sounds strike up a conversation
with your computer and and i think most people be really pleasantly surprised that that

where it goes and they'll start you know the gears will start turning ill searchy i could
use this in this part of my life for in this part of my business right it i could use it

to do this for me and

Yeah, that is true.

The deeper you go, the more you're like, hey, could see a use case for this or a use case
for that.

Yeah, totally.

Like I said, like I, I don't use it just as a business tool.

Actually, in all honesty, I probably use it more as a personal tool than a business tool
at this point.

And I use it as a business tool a lot for writing and things like that.

But it just anything where should I go for dinner tonight?

What should I have?

What should we make for dinner?

What's, you know, what, what whatever, it's incredible to have this thing in your pocket
that you can ask a question of, mean, hell, I'm at I could be at a restaurant and I'm

like, I don't know what that is on that menu.

And can, I can take a picture of the menu and ask chat GBT like, Hey, what is this?

What?

like it.

And it, and you know, like it's amazing, just stuff like that.

Like that you, you don't think of it first, but you know, the more you use it, the more
you get comfortable with it, the more you realize like, shit, can, this thing is like a

life assistant, you know,

bucks a month for most people.

It's the insane part.

Yeah, well, hell for free.

If you don't want to pay for me, you can get this out of it for free.

But paying for the, you know, the paid models is definitely worth it.

I recommend everybody pay for it.

Just to, you know, help fund the cause of nothing else.

All right, that's all I got to say about that.

Mike, always a pleasure, Let's do this again.

We'll have to get that mindfulness mental health episode fleshed out.

And I want to find an expert for that one.

But...

Yeah, I that'd be really good one.

Yeah, this has been great, man.

What am I supposed to say at the end of these things?

If you are interested in being a guest on the podcast, you know, we're looking for super
interesting stories from shop owners, sign industry experts, definitely hit us up.

Hey, at better sign shop.com.

We do have our Facebook group for shop owners only.

Make sure you check that out.

Just look for the Better Sign Shop community on Facebook.

I don't particularly like Facebook, but you guys do.

And that's where everybody is at.

So that's where we hang out.

Lots of great resources and connections there as well.

That's it.

Roll credits.

Boom.