Are you curious about how artificial intelligence is reshaping our lives? Want to know how businesses can tap into the potential of AI? Join us on the Loupe, where we have the pleasure of unpacking these subjects and more with Daniel Sirois, CPO & Co-Founder of Punchmark, and Alex Schlindwein, also known as Diamond Diaries. They'll help unravel the mysteries of AI, the impact it has on content creation, and how it's revolutionizing our workflows with the latest features like Punchmark's Ask AI tool.
Imagine being able to create engaging blog posts, landing pages, and other content effortlessly. That's the promise of AI. We explore the benefits and potential pitfalls of AI in content creation, discussing how Punchmark's Ask AI feature can help save time and resources. We also delve into the importance of the human touch in editing AI-generated content, ensuring it resonates uniquely with your audience. Daniel will share invaluable insights on how to make the most of AI, and how Punchmark decided to embrace this groundbreaking technology.
But the potential of AI extends far beyond content creation. It's a game-changer for businesses, able to assist in everything from customer onboarding to custom product creation. We'll guide you through how AI can be integrated into a retail business, the legal implications, and how your company can harness its power to gain a competitive edge. It's a brave new world out there, and we're here to help you navigate it. So, join us on this exhilarating journey through the realm of AI, and discover how it's poised to revolutionize various aspects of our lives.
Follow Alex on Instagram for more Retail Jeweler Tips: instagram.com/_diamond_diaries_/
Send feedback or learn more about the podcast: punchmark.com/loupe
Learn about Punchmark's website platform: punchmark.com
Inquire about sponsoring In the Loupe and showcase your business on our next episode: podcast@punchmark.com
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
Welcome back everybody to In the Loop.
What is up everybody?
My name is Michael Burpo.
Thanks again for listening toIn the Loop.
This week we're back.
We had a two week break, weposted a new trailer and now
we're ready for the rest of theyear.
Hope you guys are ready.
This week we're going to becovering something very exciting
, which is going to be just AIartificial intelligence in
(00:24):
general and we're going to betalking about the new Ask AI
feature that Punchmark justrolled out, and I'm joined by
Daniel Saroi, the Chief ProductOfficer at Punchmark, as well as
Alex Schlinlein, also known asDiamond Diaries and we're going
to be talking all about howwe're using artificial
intelligence in our workflows,as well as how you can keep
(00:45):
humanizing it, and also someinteresting kind of offsprings
of the very popular chat GBT andhow you can start leveraging
that as well.
So, for example, with MidJourney, which is good for
creating new images, it's goingto be a fun one.
This is very topical, and I'msure we'll have to make another
one of these in about six months, when AI has progressed even
more, but for the meantime, thisis everything you need to know
(01:07):
about AI Cheers and for yourfree demo today at punchmarkcom.
(01:41):
While enjoying this week'sepisode, take a moment and leave
us a star rating on the Spotifymobile app.
Or, if you're on Apple Podcasts, leave us a star rating and a
review.
It's the best way to help usgrow into show that you really
enjoying the show.
Speaker 2 (01:56):
Thanks, hey everybody .
This is Jason Adams, Directorof Education at Punchmark.
If you'll be at the RJO Phillyshow, we'd love to chat with you
.
Stop by our booth so we canlearn about what's working well
for you, Share some marketingadvice on what's working for our
other retailers and you caneven check out our Answer to
Marketing campaign that we'reunveiling there.
Go to punchmarkcom slash RJO tomake an appointment and visit
(02:19):
us at booth 325 at the show.
See you there.
Speaker 1 (02:24):
And now back to the show.
Welcome back everybody.
My name is Michael Burpo.
I'm joined by Dan Serro, theCPO and co-founder of Punchmark,
as well as Alex, also known asDiamond Diary, our recurring
correspondent for In the Loop.
(02:45):
How you guys doing today?
Good, Good to be here Doinggreat.
Awesome.
We're talking about the hottestword on the internet right now
artificial intelligence, alsoknown as AI.
We did a kind of a looserepisode on this about the AI
chat feature.
That podium had rolled out.
That was kind of ourintroduction to it.
This is going to be more, alittle bit more tailored to like
(03:07):
artificial intelligence, notjust specifically, but also in
general and where we kind of seeit going.
Punchmark has been using it alot.
Maybe, dan, you could startwith kind of introducing us to
the concept and how you know whyyou're so excited about it.
Speaker 4 (03:23):
Sure.
So I think the one that I gotexcited with the most right away
was ChatGPT, the one that wealready heard about.
It's free, it's accessible,it's easy to use and you can
basically have a conversationwith a robot and ask it to do
tasks and shores and help youthink through complex topics,
and it's a game changer.
For anyone who's not using it,I highly recommend getting in
(03:45):
there, playing with it, talkingto it, even if it's about silly
things like numerology orastrology questions.
It has such a wealth ofknowledge about topics that it's
very, very interesting,definitely.
Speaker 1 (03:57):
And Alex, I mean and you've used it a couple of times
, I believe, for your blog andbeing able to kind of churn out
landing pages, Is that right?
Speaker 3 (04:06):
Yeah, I have mostly with the built-in feature that
Punchmark has now.
But I was curious, dan.
I just listened to anotherwebinar the other day talking
more about all this and thedifferent.
I guess there's differentlevels of ChatGPT, like 3.5 and
4.0.
Maybe you can help describesome of the differences between
(04:26):
that a little bit better.
Speaker 4 (04:28):
Sure.
So 3.5 is the public versionthat most people have access to
when you create your freeaccount.
However, once you kind of getin there and you create your
business account, they do haveaccess to version 4, which is in
data right now, so you canrequest access to it.
There's going to be more, lesslimitations on it, a little more
power.
Obviously, every year thisthing is alive I hate to say the
(04:49):
word alive, but every year thisthing is here.
So that's where it keepsgrowing and getting smarter.
So that's where the versioningis going to come into play.
I imagine once we're getting tolike version 5, 6 of this thing
, it's going to be churning outsome pretty incredible stuff
that we can't even fathom rightnow.
So that's kind of what thosenumbers come from, and 3.5 is
current.
Speaker 1 (05:10):
Yeah, and the Chat feature is.
I think the reason why it'staken everything by storm is
because it is so easy to use.
It's no secret that artificialintelligence.
It learns a lot of time bymachine learning, which is where
you're just basically feedingit a ton of information and then
it is learning by giving it apositive or negative kind of
(05:31):
correlation, like a 0 or 1.
And then it can determinesuccess or the more successful
option through that and that hasbeen implemented very often.
I mean, even you know it's justa long way of explaining what
an algorithm is.
But, for example, google searchis just an algorithm.
(05:52):
I think that it's so enchanting.
I think is like the term.
It's like I'm just, you know,you go on there.
I mean I used it to write likewrite me a bedtime story.
That was like the first time Iwas like give me like a five
minute bedtime story and it doesit for you and it's a
completely original one.
Dan, do you remember like whatyour kind of early experiments
(06:15):
were with it?
Speaker 4 (06:16):
So one of the ones that kind of opened my eyes the
most.
I'm also a reef tank enthusiast.
I have a reef tank at the house,like you know, do tropical fish
and coral and all that.
It's a lot of water parametersand dosing this, that and the
other chemical.
So it'd be surprised on howmuch knowledge it has, just from
its short time being around tohave, you know, absorbed all
(06:37):
that.
Where you can ask it a question, as if you're standing in the
aquarium shop, and it'll giveyou answers on why this
parameter should be what andwhat chemicals.
It won't necessarily recommendproducts or anything, but you
know it can help you with thescience behind things and it's
super cool.
It's just so helpful.
That was my first like personaluse of it where I was just like,
(06:57):
okay, this thing is prettydifferent.
The next one I would say,without getting too far ahead,
was being able to throw code atit and watch and manipulate or
refactor a piece of code.
Say, you've got a littleJavaScript snippet that you know
changes the phone number tobold or whatever.
You can pop that in and ask itto refactor and it'll rewrite
the whole thing with less impacton your site or whatever you're
(07:19):
adding it to.
It's super impressive.
Speaker 1 (07:22):
Yeah, it's like the problem solving is.
What's fascinating to me islike I mean, it's for a really
long time when I was learninghow to code.
I still don't know how to code,but I would have to ask Dan
every single step of the HTMLprocess.
You're like, hey Dan, I havethis image.
I want to add a very thinstroke to it for this blog that
we have on punchmarkcom.
How do I do that?
(07:43):
And Dan would have to turnaround, be like, okay, it's
going to be style equals quotes,blah, blah, blah, blah.
And then I would do that and itwould work.
And then I would have to askthe next thing and it would
always pull Dan's attention awayfrom what he was working on For
a recent blog that we made.
It was announcing, you know,some promotions at punchmark.
I was just like, oh, let mejust see if I can just ask you
(08:03):
to chat to you this, and that iswhat it is very good at.
There are some things.
Maybe, Dan, can you like kindof shine a little bit of light
on some things it's not great atyet.
Is that possible?
Speaker 4 (08:15):
So it has a stop of knowledge after 2001 or 2021.
So one thing I try to do thisyear was I'm sure a lot of
people play fantasy football.
I was going to start asking itabout.
You know which players should I?
Speaker 3 (08:28):
draft.
Speaker 4 (08:29):
You know, here's the mode I like to get it out, but
it doesn't have access toanything current.
It doesn't have access to theinternet at all.
It only had access while it wasin its sort of hibernation
state.
So anything you ask it sort ofhas to be either timeless, data
problem solving or creative.
It's not going to give you,like you know, current events or
(08:49):
interpretations of anythingcurrent or stuff like that.
So that's where it kind offalls short For a good reason,
though I think that's where itstarts to get pretty scary on
the kid of power If it did haveaccess to the internet and what
could happen with that.
So you know, that's for a goodreason.
Speaker 1 (09:05):
So I guess let's dive into how Punchmark has
implemented it into Site Managerand then we'll get into all the
exciting different things andthe possibilities and how we're
using it.
For example, dan, walk methrough the Ask AI tool that
Punchmark rolled out.
We made a big deal about it.
It was right before JCK Realbig win for us.
(09:26):
I was really excited about itbecause we were one of the first
to kind of bring it to market,if not the first.
Can you just kind of give me anoverview and like what that
means?
Speaker 4 (09:35):
Sure, so our first goal at Punchmark was to add
content creation capabilitieswith AI, so we incorporated this
into Page Builder.
So when you add a header, youcan ask it to regenerate or
generate you a header based oninformation on the page, as well
as take a header and thengenerate a description text to
go with it and then, in the blogas well, you know, allowing it
(09:57):
to generate full article textbased on a prompt that you can
add.
At the moment, we've met keptthe prompts pretty simple.
It's nothing too complex, but Ithink as we go forward, we'll
probably start to feed morecustomer information into that,
those prompts behind the scenes,so that it can get a little bit
more specific to the customerasking and not just the same for
(10:18):
all.
So it's, you know, while it's abig win for Punchmark that we
have this capability on thefront end, I think there's a lot
more cool stuff we're going tobe able to do with it now that
it's available.
Speaker 1 (10:28):
Yeah, and Alex, I know you were one of the first
people that I actually saw thatI hadn't explained the tool to,
that actually took the tool,learned how to use it and then
started implementing it.
Can you talk me through, firstof all, what the learning
process was and also what thefirst thing you started building
with it was?
Speaker 3 (10:47):
Yeah.
So I mean I have a little bitof knowledge because I have
played around with chat GPT.
So I will say I had a littlebit of background knowledge.
My husband, like loves thiskind of stuff.
So he told me about chat GPTprobably at least six months ago
, if not even a, maybe a littlebit further back.
But I saw the email blast thatyou know this was going live,
and I was like, oh my gosh, thisis so cool, I can't wait to try
(11:09):
it out.
I just jumped in there.
I really didn't know exactly howit was all going to work, but
it's really, I mean, easy tofigure out and even for someone
that I think is not techie, Ithink could figure it out.
So it's on the page builder.
You just click the button inthe top toolbar.
It says ask AI.
And I first just put reallysimple stuff in there just to
(11:30):
see what it would come out likeyellow gold ring, what is it
going to put out with for that?
And it did it like itelaborated, and I don't know how
it pulled some of it, becauseit was it started saying like
five grams with diamonds and I'mlike, where did you get that
from?
Speaker 4 (11:45):
I didn't say how much this weighs or anything.
Speaker 3 (11:48):
So it is kind of funny.
But so you definitely have toedit it and proofread it and all
that.
But it is interesting that itcan prompt you on what details
could further the description orthe page as well.
So even though I didn't put inthere the weight of the piece,
it now put something in thereand I'm like maybe I should put
the weight of the piece.
(12:08):
Let me go figure out how muchthat piece actually weighs, and
maybe that is useful information.
So I think it's good for that.
But now I've been using it a lotfor writing my blog posts and
copy on my landing pages andit's been.
It's saving me so much time.
I think it would take meprobably a good two hours to
write a blog post because I liketo be very thorough and I do
(12:31):
like to do the research, andthen I'm going to proofread it
and I'm going to like go, youknow, simmer on it and come back
to it again.
I wrote a blog post in like 30minutes the other day using the
ask AI and I do see a little bitof repetition sometimes with my
paragraphs in there.
So I just go, you know, you gotto go, fluff it and everything
and take out the redundancies,but overall it was really a
(12:55):
great basis of where to startanyways.
So, for example, I wrote a blogpost the other day, using it
that I did the top weddingjewelry trends of 2023.
Speaker 1 (13:05):
That was the one I saw.
I loved it.
Speaker 3 (13:07):
Yeah, and I already had kind of some ideas in my
head of where I wanted to gowith it, so I didn't let it do
the whole thing from scratch,but I just said write me and
write an intro paragraph abouttop wedding jewelry trends of
2023.
It wrote the intro for me, thenI fluffed it and edited it.
Then I said, okay, write me aparagraph about rose gold
wedding jewelry.
(13:28):
It wrote me the paragraph, Iedited it.
So that's kind of how I wentabout it.
So I gave it, you know, someguidelines to look for, and
sometimes I gave it more or lesslike write me a paragraph about
wearing black diamonds for analternative wedding jewelry,
look.
And so the more you give it,the more you're going to get out
of.
It is definitely what I'mfinding too, because it can only
(13:50):
go so far.
I can't read your mind, youknow.
So I am curious Does the ask AI?
Does it have the capability oflearning or following any kind
of patterns that you might useit for?
Speaker 4 (14:02):
So we can.
We have not done any model,which is called this training
data, so we haven't starteddoing training data yet on this.
However, it is all using thesame API key, so that means all
of our customers are hooked intothe same exact bot.
Basically, probably the waywe're going to do it is is we're
going to break it down byfeature type and then start to
train the models based on eachone.
(14:24):
So, for instance, contentcreation you know all the
different things that jewelersare putting in there.
It can start to learn.
You know that these are alljewelers for one, or that you
know where their store existsand things like that.
And then there's things likeproduct descriptions, where you
know it's not going to be thatat all, but it's going to be
more like how do we pull thisinformation out of their
database, their cataloginformation, their edge data,
(14:48):
and make sure we can utilizethat data when we're training
that model?
So each one is more specific tothe task that it's trying to
achieve.
So, yeah, that's kind of ournext.
Our next trick at a bunch morehere is starting to
individualize and train themodel so that they all start to
learn from each other.
We probably don't want them tobe so co like, so intermingled,
where your creativeness is thenbleeding into other you know of
(15:12):
our customers, because thatstuff, I think, is private to
you.
That's your prompts, that's theresults you got.
However, you know the thingsthat we can do to help you get a
leg up.
We definitely want to befeeding feeding to the model.
Speaker 3 (15:23):
That's interesting, and so what was really the I
don't know the mindset behindyou guys wanting to put it
straight into punch mark versusjust, I don't know, helping us
figure out how to use chat GPTfor stuff like this?
Speaker 4 (15:38):
For me.
I was an 80s kid, right, soprobably the very end of Gen X.
Maybe you call me a millennial.
On the first year, you know, Igot playing in the dirt and
computers came out when I waseight or nine years old and I
was in love right away and Ichanged everything I was doing.
I immediately stopped doingnormal art, started doing
computer art and started myjourney on that and what an
(15:58):
impact it had on the world,right.
So I kind of see AI as anotherone of those that's happening in
my lifetime.
So, as the CPO of Punchmark, I'mgoing to try to make it my
mission to rebuild our platformto utilize AI technology in
every single nook and crannythat we can possibly put it into
, because I think a lot ofpeople are maybe nervous about
(16:20):
AI and implications of it.
Where this is going am I goingto lose my job, all that kind of
stuff but I think the peoplethat are utilizing it now and
starting to understand it thatit's just a tool, that it's not
going to take over the world andthen be a Terminator 2, skynet,
it's really just.
You know, it's a reallypowerful tool and if you
understand how to use it and howto talk to it.
It can double your capacityeasily, within no time, so you
(16:44):
can bring that same power theentire platform, from page
creation to products, to imaging, to everything that we do.
Speaker 1 (16:51):
I saw this quote.
It was from Henry Ford, ofcourse, you know, inventor like,
I guess, the guy who madepopular the automobile.
He said if I had asked themwhat they wanted, they would
have responded faster horses.
And I was like you know whatit's a fair point, like if
people, what are people askingfor us?
We want to make you know, wewant more website sales.
(17:13):
But it's like okay, like let'sthink beyond that.
And to me it's like let's takethe manualness out of a lot of
these things that are just hypermanual.
Like you said about writing ablog post, like you know, about
you know, two hours, two and ahalf hours, for for me, like
it's also kind of goes almostbeyond that where it's about
topics that not only am I justnot a super expert about, but
(17:36):
I'm the guy who usually writesthese blog posts for punch mark.
So it's going to take time fromother people and a lot of times
I just have to drag on them.
So, for example, like we'reworking on this thing about ADA
compliance I don't know anythingabout ADA compliance.
I'm going to need to ask Dan.
Dan is in the middle ofbuilding all these websites, so
(17:58):
like it's going to take both ofus away to write this blog post
and instead it's like let me get95% of the way there, we can
tweak it a little bit and thenrun it by Dan, and then Dan can
play editor and it just is goingto put us on third base instead
of at that.
So we're going to take a quickbreak and hear a word from our
sponsors.
Stay with us.
(18:21):
Introducing Rembrandt Charms,the family owned company with
over 50 years of charm and charmbracelet design and
manufacturing experience.
Explore the world's largestcharm collection in sterling
silver, gold plate, 10 and 14 K,yellow and white gold.
The sterling silver merchandiseis also rhodium plated to
prevent tarnishing.
All Rembrandt Charms productscome with lifetime warranty
(18:45):
showcasing their dedication andcraftsmanship, available
exclusively through authorizedretail jewelers.
Find a store near you on theirwebsites.
Find a retailer section atRembrandtCharmscom To be listed
on their retail locator.
You'll need a retail jewelrystore with a Rembrandt Charms
display or shop and shop.
Our business building charmprogram offers inventory
(19:07):
controlled displays tailored toyour store, along with marketing
support, e-commerce assistance,stock balancing and more.
Contact Rembrandt Charms todayfor more information at
800-828-7840.
Again, that's 800-828-7840.
Until in the loop sent yaAlright and back to the show.
(19:30):
And we're back.
I guess I kind of want to bringthat through and talk about the
next step, which is, you know,ai driven imagery.
It is through this very popularAI model called mid-journey.
Then can you kind of give thelisteners like an idea of what
(19:52):
mid-journey is and also kind ofwhat makes you so excited about
it.
Sure, mid-journey, it is an AIimage creation tool.
Speaker 4 (20:00):
Right now it's exclusively on Discord Crazy.
So basically a message boardthat's hooked up to code and a
bunch of people working on it,pushing new features and
releases to it.
Anything you create on there ispublic domain, so nothing
you're creating there is goingto be able to be copywritten.
It can be stolen by anybody.
However, it's a great place togo in and brainstorm to.
(20:21):
You know, as a graphic designer, mike, you'll know this too,
but when you're coming up withlogos for somebody, you go
through a lot of concepts, a lotof ideas, a lot of like what if
we put a diamond on a K?
What if we put a?
You know all these differentthings that you're going to
think of, and sometimes you wantto see it before you commit to
(20:42):
the concept.
And you know back in the day theway we were trained.
You would do quick sketches.
You'd probably jump out of yoursketchbook and just start
drawing and stuff, which isstill obviously good, but to be
able to jump in there and justtalk to something and say, all
right, give me this, give methat, give me this, give me that
.
It really lets you move reallyfast on something and decide
when you hate something a lotfaster, you know, in a minute or
(21:02):
two.
So that's one example, but theway it can fake people and do
places and things is justabsolutely incredible.
So a good example of what I wasactually doing last night we're
working on a new campaign forAntwerp Goers and helping with
the marketing of that, so Istarted it and it's also one of
those things that's really hardto get images of because we're
(21:25):
not in Antwerp and stuff iscopyrighted in our editorial so
it's very difficult.
So just get in there and starttalking to it and say you know,
give me a picture of a jewelerin Antwerp meeting with diamond
cutters in a private office,touching round brilliant stones,
and immediately get into thebeautiful scene.
That's not perfect.
Similar to chat.
You can see it's not going tobe perfect right away.
Like the first ones, we weregiggling yesterday because the
(21:53):
rocks were like yeah, literallythe ones right now, right now.
And it's just like okay, likethat's a little crazy, but the
fact that it did that is awesomeand some of the scenes that it
made of like a late night inAntwerp with the moon in the
back, it's just, it's gorgeous.
So, like I saw some really coolstuff on JHJ as well, where
customers were creating customjewelry concepts with it.
(22:15):
You know where.
You know it's like here's thedifferent ring designs I might
do with somebody and they cankind of see it right away
without you guys having to gothrough and do a CAD and all
that extra work.
I mean, this one is stillrelatively new, so the API is
kind of difficult.
The API meaning the way weconnect it into the platform.
I think it's a little bitdifficult, but I think when it
gets there, this one's going tobe an absolute game changer, as
(22:36):
cool as ChatGPT is right now.
Speaker 1 (22:38):
Yeah, I just like the idea of being able to do quick
and dirty you know, concepts.
That's the thing is like.
The other part of it is justlike for you know, a social post
.
I'm sure that the next step forAlex is going to be like hey,
how can I get this to write,like you know, better
descriptions or more engagingcontent and stuff like that.
(22:59):
You know, do you see that kindof coming around Alex?
Speaker 3 (23:02):
Yeah, I've actually been playing around with that
already and actually my emailservice provider has a.
It's not completely AI, butit'll like suggest subject lines
for you and it's really nicebecause, I mean, I have to come
up with a lot of content all thetime, like at least one social
post every day.
Sometimes I'm doing multipleposts a day.
I'm usually sending out atleast one eblast almost daily.
(23:25):
I you know I start to feel likeI can't think of any more new
subject lines or captions orwhatever.
So it does help with thewriter's block, you know, when
it comes to all that.
So I'm really excited for thatto continue and to get easier to
use it for all of that.
Speaker 4 (23:40):
It really allows everybody to kind of be a
director, right.
So I mean, it used to be backin the day like you had to be
really good at doing the thingand coming up with the ideas,
and now it's kind of just, it'sgoing to start to level the
playing field where it's justlike, if you've got good ideas,
the execution side can kind ofbe handled pretty well by robot.
It doesn't mean it's going tobe perfect and you probably
(24:01):
still want a professional tocome in and do your final, but,
man, it can get you so close tothe finish line.
Speaker 1 (24:06):
Yeah, I guess maybe just staying on that point.
I think that there is somethingimportant about, like the.
I think that the human elementof this is the idea of editing
or refining or there's a lot ofthose kinds of concepts where
it'll get you I keep saying,it'll put you on third base, and
I think that with mid journey,you know you go through and if
(24:29):
you're not careful, you'll postsomething of like a model with
you know, six fingers on ourhand.
That's very common, because itdoesn't really quite know what
the hand looks like.
His hands are weird.
Same thing with but we've beenseeing it with models wearing
jewelry is kind of the thingthat I kind of hope will be the
(24:50):
next kind of step, becauseshooting models wearing jewelry,
especially jewelry that iseither a state or one off or you
know, et cetera, is verydifficult.
I guess I'll pose this questionopenly to you guys when do you
think the you know theresponsibility is, or the humans
, to be the editor?
I think that we need to makesure that the content that we're
(25:12):
putting out is quality and notrushed and is not entirely from
a robot, because I definitelyfeel like that's important.
Am I?
Am I crazy?
Speaker 4 (25:23):
No, not at all, and I think that's where you know
refining it as you go is veryimportant.
So, like with mid journey, youknow it is giving you a
finalized image.
So it's very tempting to use,but a lot of times it's, you
know, like you said, with notjewelry on models.
We did one where it put thediamond stud right in your cheek
and it's like I don't know ifanybody is getting that stud put
in.
you know so six fingers, youknow just all kinds of things.
(25:46):
It'll put words in thebackground that shouldn't be
there, like yeah, you definitelywant to play editor and keep
messing with it.
And that's where I think reallyyou know.
Why playing with it is soimportant is because we're all
really training ourselves tobecome prompt engineers, as they
would say, where you'relearning how to talk to the
robot.
So you're not just like give mea picture of something great
and then you just wait forsomething to come at you.
(26:07):
Like you have to really tell itspecifically what you want and
when it gets it wrong, you sayno, no, no, not like that, like
this, and you train it.
So that's, I mean.
I think that's really the keyis making sure you're always
doing that and then, when youget a final product, making it
your own.
Speaker 3 (26:21):
Yeah, and at some point you're going to your
audience can tell you know, likeif you don't edit it, they can
tell you're using stuff that'sbeen generated by a robot and
you're going to lose followers,you're going to get unsubscribed
, you're going to lose customers.
So you've got to still put thathuman element into it, or
people are going to be justglazing over.
Speaker 1 (26:42):
Yeah, yeah, related to that, alex, I know that like,
like a big part of marketing ishaving this voice, and it's one
of those things I actuallytried to explain to my parents
recently what what voices, andlike I tried to explain it
through the context of mypaintings where I was like, oh,
I have like a style, it's like avoice, and they really didn't
seem like they.
(27:02):
They got that.
I was like it's kind of like my, like a little bit of me,
that's in there, and I tried toexplain it about, especially for
writers and how, like certainwriters have like this wit.
That's not them trying to befunny, but it's like almost how
they frame the world thatthey're building.
I know that when you speak withjewelers, a lot of times you're
talking about you got to find,like, who you are.
(27:24):
Do you think that that kind ofbringing the voice into these
kind of generated content isgoing to be something that's
even achievable, or how do youthink that works?
Speaker 3 (27:37):
Yeah, definitely, and actually I think chat GPT has
the capability of doing this andyou can actually take, you know
, some text, feed it into chatGPT and say what is the voice or
the tone of this writing, andit can identify that for you.
And then you can go back andsay now write me something in
that tone or voice.
(27:57):
So that is important to knowwhat yours is.
And it's one of those thingslike no one can really help you
with.
You have to figure that out onyour own, especially for your
own brand.
Like do you want to be funnyand witty?
Do you want to be serious?
Do you want to be buttoned up?
So you kind of have to figurethat out before you move forward
with using this.
You know in a big picture, orelse you're not, you're going to
(28:18):
lose your identity 100%, andactually mid-journey does the
same thing.
Speaker 4 (28:23):
So there's actually a command you can feed to
mid-journey called describe,where you just upload the image
in and it'll just fit back textto you that gives you what a
prompt might look like to createthe image you just submitted.
Then I really teach you to belike okay, this is what it's
expecting when you want thatkind of result.
Like you said before, Mike,it's really like you want to
give it as much information aspossible and some of that is
(28:45):
stylistic.
Speaker 1 (28:47):
That was like the thing is.
So earlier today I just have itup right here Dan had put in
one of a headshot of himself anda series of professional
photography that we had done ofourselves.
It created these cartoon, coolbackgrounds that really
juxtaposed the image.
It was really interesting, Iguess.
(29:09):
I'm like would it be able tothen create a set of them?
Because part of somethingthat's hard to fake is, I guess,
consistency across multipleiterations.
A lot of the times I feel likethat might be where it falls
short.
I don't know.
It has almost too many optionsthat it can't really be
(29:33):
consistent.
Speaker 4 (29:35):
Yeah, I think when you're using the open source
tools right now, that'sdefinitely a problem, because
you can only train it so muchwithin the chat.
But again, that's where themodel training comes in, where
you can give it specific stuffand say this is your brain and
anything else is invalid.
So a good example of that isproduct description.
That's one we're doing where itsucks in all the different
(29:56):
variations of your product andsizes it's available in and
whatever, and generates a nicedescription that's unique to
that product alone.
Another one is image reading.
So in Page Builder I knowthere's another one that
everyone hates doing is addingalt tags to all their images.
Everybody, please do that.
It's very important.
We try to do some stuff in PageBuilder that'll maybe take the
(30:17):
headline of your widget and tryto create the alt tag.
But in the future one of thethings we're going to be hitting
is something just likemid-journey, where it can read
the image and tell me exactlywhat's in it, based on the
pixels.
Speaker 1 (30:30):
There's so many options for it.
I think that the timeline is alot shorter than what a lot of
people are expecting.
I think that the first time Iever heard that the phrase
chatGPT was probably about maybethree months ago four months
ago, it's tough to say, but thefact that we've come this far
(30:50):
where we can generate images injust about four months, it's not
going to get worse, it's onlygoing to get better.
Dan, where do you think thisimplementation falls on?
As far as the focus forPunchmark and also the timeline
it's tough to say timeline.
Speaker 4 (31:05):
We're trying to tread lightly, I would say we're
probably going to do a lot ofrefinements to what exists for
now and making sure theimplementations we already have
are getting better and we'refine tuning them.
As I mentioned, productdescriptions, alt tags those
will be some ones that will rollout in the near future One of
the most exciting ones, I think,beyond mid-journey, which I
would love to implement as well.
(31:25):
I don't know if the API isready for that, but if it will
be, I'll do that one as well.
But page builder being able totalk to site manager and say I
need a page about this and haveit, not only write it, but
generate a page with a herobanner and a call to action and
content in the middle and knowwhat the code of page builder
looks like, that's something I'mstarting to personally work on
(31:48):
now.
I mean, that's an absolute gamechanger for our customers, for
us at Punchmark, and that'sgoing to take just straight
model training, where we'refeeding it exactly how to build
a page, all the possibilities,what not to do, what you're not
allowed to do.
So it's going to take a while,but I think the end result is
going to be absolutely gamechanging for everybody.
Speaker 1 (32:08):
And Alex are you.
You know things are changingand, as a business owner, do you
see this as an opportunity foryou to be teaching your
employees how to leverage thistool?
Kind of a weird kind ofenvironment for it retail, you
know how I it's very wellequipped for the online presence
, but is there a way to sort offit it into the in store
(32:33):
approach, or what have youthought about that?
Speaker 3 (32:36):
That's a very interesting idea and I haven't
put much thought to it yet, tobe honest.
I mean, I do believe stronglythat everything I do more on the
back end supports my front line.
You know, the more complete ourwebsite is, the better the
descriptions are on our website,the more the customer is
already closed before they comein the door, you know, and it
(32:56):
answers a lot of those mostcommonly asked questions ahead
of time.
But that will be reallyinteresting to see Like.
I mean, we're so like in theinfancy of all this, so I'm
excited to see, like what allthe applications are going to be
of it.
I do have a couple of myassociates that are interested
in it and trying to figure itout, so I will definitely report
(33:17):
back if I find anything ofvalue Nice.
Speaker 4 (33:20):
Well, I'm curious if a customer came in with a bunch
of mid journey ideas for rings,would you guys take that?
Or would it be like, oh my God,get out of here?
Speaker 3 (33:31):
Yeah, I mean, why not ?
It's not really any differentthan if they sketch something on
a napkin and brought it in andasked if we could make it.
It does post some questions asfar as who owns the design and
stuff like that.
So you know, and luckily our,our custom designer that we work
with does some of that legworkfor us to make sure we're not
(33:52):
using copyrighted designs.
But I'm it's going to beinteresting to see, like, how
that's all going to play out,and I'm sure we're going to have
some new, some new laws andregulations based around all
this.
So we'll wait and see.
Speaker 4 (34:06):
We've toyed with the idea of implementing it,
whenever it's available, apiwise, into the our wizard where
you can kind of start a customproject on the website, sort of
giving them access to midjourney right there.
That would be so cool.
Speaker 3 (34:18):
It was starting point , like yeah, could you imagine
if you could actually type inthere like I want a three stone
ring with these birthstones andI want it yellow gold, but with
this with a swirl, and it justproduced an image for you right
there?
That would be mind boggling.
Speaker 4 (34:33):
It's super easy to do , but I think I think the
question for me as a developeris like is there they're going
to be submitting stuff that youguys are like this is impossible
, you got a giant.
Speaker 3 (34:46):
It would have to definitely have some disclaimers
.
Like you know, this is all preproduction or sales associate or
designers would need to lookover it, whatever you know,
before putting into production.
But I'm sure you know, I thinkpeople all understand to some
extent that this is all like Idon't know what's the word, like
it's aspirational, I guess youknow.
(35:07):
And it's just a place to start.
Speaker 1 (35:10):
Yeah, I think that it's again.
I am very focused, I think, onI'm trying to find new ways to
use this and, like Dan said,he's not just, you know,
preaching this to everybody else.
We have actually internalizedthis at Punchmark.
We've encouraged all thedifferent departments at
Punchmark to find ways thatthese tools might be able to
(35:33):
make their jobs easier.
Because it's kind of like, youknow, when there's a manual
approach to something and thenthere's like a scaled up way
that uses technology whetherthat's shoveling your driveway,
and it's like, oh, I only shovelby hand it's like, well, you
could just use technology anduse a snowblower, like you're
(35:54):
not proving anything by doing itby hand.
It's not like it makes it just,especially if it's a manual
thing.
But the one of the things thatI kind of wanted to explain
about how I'm using it just as areal life example, not just at
Punchmark, but actually I'musing it for a, for this podcast
, and I use a plugin that is inour podcast hosting tool called
(36:18):
buzz.
What we use is Buzzsprout andthey have a tool that they just
rolled out called co host AI.
And just as a kind of an ideaand I kind of want to just
mention this because then peoplewill be aware of why and how
we're using it, and I'm a host.
Ai does, and I've only done itfor one episode so far at the
time of recording this, but thisepisode will be on.
It is.
(36:39):
It writes my episode titles,episode description, transcript
and chapter markers.
So to break those down and whyI use them.
It as an example of why an AIis going to be able to help help
.
It is not changing the way thatwe are making this and producing
this content.
It's just taking away themanualness of it.
(36:59):
So, for example, chaptermarkers are one of the key
things that I've read and amillion places that will elevate
your podcast.
I've never done them, or Ican't keep up with them, because
you have to go through and youhave to listen to the entire
podcast again and then annotateit and write these anchors for
the episode.
You know, maybe you need fourchapter markers and it's just so
(37:23):
manual and I just can't do that, but it will do it for me.
But when it comes to episodetitles, it's like, oh, but maybe
you're going to be missing someof the spin.
It's like, yes, but what itwill allow me to do, it gives
you like six options.
Go in and then you can touchthem up from there.
That's the kind of use I amtrying to find is that it will
(37:43):
take a lot of the manualness outof it, or it will put me on
third base and allow me to justlightly edit and go from there.
So that's kind of like a reallife use case that is tangible
to a listener right now.
So all of that to say thank you,I think it's fun.
This episode will be on it.
I don't know if that's meta asheck, but whatever you're going
(38:04):
to do All that to say, I thinkthat maybe kind of leave it as
far as you know the generalconsensus and feeling around
this tool, I'm excited, but I'malso slightly apprehensive.
I really want to maintain thehuman element of everything I do
, because I do think that thatis what is special about you
(38:26):
know, the work I produce is thishuman element in the voice,
naturally, and you know, youknow metaphysically, do you guys
?
You know, alex, how will youapproach AI in the future?
Is it like, hey, let's divehead first, or is there some
kind of restraint that you'regoing to set up?
Speaker 3 (38:47):
Oh yeah, I mean, I think with anything that you
implement in business, you can'tjust jump in.
You've got to figure it out.
And especially with this,because it's so new, I'm dipping
my baby toe into it, you know,with the, with the ask AI, and
we'll go from there, you know,and it's might be a live and
(39:07):
learn kind of thing, you knowand kind of see how things go
and if I might have to pull itback at some point, you never
know.
So that's kind of my plan, justto kind of see, and I'm
actively watching what otherpeople are doing with it too.
There's a lot of people talkingabout this topic right now and
I think we can all learn fromeach other and, you know, see
where it goes, what about you,dan?
Speaker 4 (39:28):
Yeah, agreed, I mean I would say you know.
One thing I like to compare isyou know, people like you know
when Google came out or theinternet, I guess Google is kind
of people would say, oh,there's too much information out
here and you know why.
You know you can learn anythingonline.
Now, you can take a class, youcan do this, you can do that,
and the ask is really going tokind of be the same thing, right
(39:49):
, like I find myself alreadystarting to lean towards asking
chat.
You can see the answersquestions over Google as of
right now, and not because it's,you know, got better access to
things, as it does on Googledoes.
But the way it can kind ofsummarize things and simplify it
and make it accessible, I thinkis unique, and the more that
(40:11):
grows, the better it's going toget, and it's going to get.
It'll probably get to the pointwhere, if I'm sure, google will
roll out an AI at some pointthey're probably making it as
slow as possible because they'llhave the be all end, all of
information to feed that thing.
But I think you know, I thinkthat's that's the way of the
future.
This is a tool that we're allgoing to have to sort of get
used to, and it's kind ofamazing in a lot of ways.
(40:32):
It's scary in other ways, butI'll you know I think I'll leave
some of that to you know,whatever government entities are
going to get into their race ofcreating AI for nefarious
reasons, like, but you know,that is what it is and I think
everyone has a right to benervous about that and
apprehensive, but I wouldn'tdiscount the entire tool as a
(40:52):
result of that.
I would say that it's still atool, just like a hammer.
Pick it up, give it a shot.
It's going to help you.
Speaker 1 (40:58):
You know just like the internet there's like a lot
of is what you make of it.
You know the internet for everyexcellent example of, like how
I was able to show my paintingstream to my grandma, who lives
700 miles away, and have herwatch me paint.
I mean that's amazing.
But for every example of that,I'll show you an equally
horrible example of how theinternet has been used.
(41:20):
And we live I live perenniallyon the on the internet.
What are we going to say, alex?
Speaker 3 (41:26):
I was going to say the same thing.
You're saying, actually, thatjust like the internet.
Speaker 1 (41:31):
You know, just because there are terrible uses
of the internet I'm not doesn'tmean that we should stop making
websites for jewelers.
Speaker 4 (41:40):
So you don't throw the baby out with the bathwater
and there you go, there you go.
Speaker 1 (41:44):
Honestly, you know what.
Maybe we'll leave it out thatthat's a great point.
Alex Dan, thank you so much forbeing with me.
I think that this is kind oflike a more introspective
episode is very interesting.
I kind of hope everybody triesand even if you decide that you
don't like it and you know what,at least you know I think that
it is so accessible that youmight as well try, so that way,
(42:07):
when you form an opinion, you'llhave an informed opinion.
Speaker 4 (42:11):
We should do a chapter two someday when we when
we get to the next advancementnext year.
Speaker 3 (42:17):
When we've all been replaced by robots.
Speaker 1 (42:20):
Beep boop?
Absolutely, I think it would begreat.
I really like doing these kindof episodes.
Yeah, we'll set this.
We'll do another episode onthis Next time that there is
like a noticeable change whenGoogle comes out with their
whatever barred or whatever it'scalled.
It's going to be, I'm sure,better than everybody else's,
but in the meantime I'm excited.
I'm going to go play aroundwith mid journey, have it,
(42:41):
conjure up some cool rings andwe'll take it from there.
Take what we got.
All right, everybody.
Thanks so much for listening.
Appreciate you all.
New episode next week, tuesday.
Bye, all right, everybody.
That's the end of the show.
Thanks so much for listening.
This week's episode was broughtto you by punch mark and
produced and hosted by MichaelBurpo Our guest this week was
(43:03):
actually in wine, as well asDaniel Saroi, the chief product
officer at punch mark.
This episode was edited by PaulSuarez with music by Ross
Cochrum.
Don't forget to rate thepodcast on Spotify and Apple
Podcast and leave feedback forus at punchmarkcom.
Slash loop.
That's L-O-U-P-E.
We'll be back next week,tuesday, with another episode,
(43:23):
cheers.
Welcome back everybody to In the Loop.
What is up everybody?
My name is Michael Burpo.
Thanks again for listening toIn the Loop.
This week we're back.
We had a two week break, weposted a new trailer and now
we're ready for the rest of theyear.
Hope you guys are ready.
This week we're going to becovering something very exciting
, which is going to be just AIartificial intelligence in
(00:24):
general and we're going to betalking about the new Ask AI
feature that Punchmark justrolled out, and I'm joined by
Daniel Saroi, the Chief ProductOfficer at Punchmark, as well as
Alex Schlinlein, also known asDiamond Diaries and we're going
to be talking all about howwe're using artificial
intelligence in our workflows,as well as how you can keep
(00:45):
humanizing it, and also someinteresting kind of offsprings
of the very popular chat GBT andhow you can start leveraging
that as well.
So, for example, with MidJourney, which is good for
creating new images, it's goingto be a fun one.
This is very topical, and I'msure we'll have to make another
one of these in about six months, when AI has progressed even
more, but for the meantime, thisis everything you need to know
(01:07):
about AI Cheers and for yourfree demo today at punchmarkcom.
(01:41):
While enjoying this week'sepisode, take a moment and leave
us a star rating on the Spotifymobile app.
Or, if you're on Apple Podcasts, leave us a star rating and a
review.
It's the best way to help usgrow into show that you really
enjoying the show.
Speaker 2 (01:56):
Thanks, hey everybody .
This is Jason Adams, Directorof Education at Punchmark.
If you'll be at the RJO Phillyshow, we'd love to chat with you
.
Stop by our booth so we canlearn about what's working well
for you, Share some marketingadvice on what's working for our
other retailers and you caneven check out our Answer to
Marketing campaign that we'reunveiling there.
Go to punchmarkcom slash RJO tomake an appointment and visit
(02:19):
us at booth 325 at the show.
See you there.
Speaker 1 (02:24):
And now back to the show.
Welcome back everybody.
My name is Michael Burpo.
I'm joined by Dan Serro, theCPO and co-founder of Punchmark,
as well as Alex, also known asDiamond Diary, our recurring
correspondent for In the Loop.
(02:45):
How you guys doing today?
Good, Good to be here Doinggreat.
Awesome.
We're talking about the hottestword on the internet right now
artificial intelligence, alsoknown as AI.
We did a kind of a looserepisode on this about the AI
chat feature.
That podium had rolled out.
That was kind of ourintroduction to it.
This is going to be more, alittle bit more tailored to like
(03:07):
artificial intelligence, notjust specifically, but also in
general and where we kind of seeit going.
Punchmark has been using it alot.
Maybe, dan, you could startwith kind of introducing us to
the concept and how you know whyyou're so excited about it.
Speaker 4 (03:23):
Sure.
So I think the one that I gotexcited with the most right away
was ChatGPT, the one that wealready heard about.
It's free, it's accessible,it's easy to use and you can
basically have a conversationwith a robot and ask it to do
tasks and shores and help youthink through complex topics,
and it's a game changer.
For anyone who's not using it,I highly recommend getting in
(03:45):
there, playing with it, talkingto it, even if it's about silly
things like numerology orastrology questions.
It has such a wealth ofknowledge about topics that it's
very, very interesting,definitely.
Speaker 1 (03:57):
And Alex, I mean and you've used it a couple of times
, I believe, for your blog andbeing able to kind of churn out
landing pages, Is that right?
Speaker 3 (04:06):
Yeah, I have mostly with the built-in feature that
Punchmark has now.
But I was curious, dan.
I just listened to anotherwebinar the other day talking
more about all this and thedifferent.
I guess there's differentlevels of ChatGPT, like 3.5 and
4.0.
Maybe you can help describesome of the differences between
(04:26):
that a little bit better.
Speaker 4 (04:28):
Sure.
So 3.5 is the public versionthat most people have access to
when you create your freeaccount.
However, once you kind of getin there and you create your
business account, they do haveaccess to version 4, which is in
data right now, so you canrequest access to it.
There's going to be more, lesslimitations on it, a little more
power.
Obviously, every year thisthing is alive I hate to say the
(04:49):
word alive, but every year thisthing is here.
So that's where it keepsgrowing and getting smarter.
So that's where the versioningis going to come into play.
I imagine once we're getting tolike version 5, 6 of this thing
, it's going to be churning outsome pretty incredible stuff
that we can't even fathom rightnow.
So that's kind of what thosenumbers come from, and 3.5 is
current.
Speaker 1 (05:10):
Yeah, and the Chat feature is.
I think the reason why it'staken everything by storm is
because it is so easy to use.
It's no secret that artificialintelligence.
It learns a lot of time bymachine learning, which is where
you're just basically feedingit a ton of information and then
it is learning by giving it apositive or negative kind of
(05:31):
correlation, like a 0 or 1.
And then it can determinesuccess or the more successful
option through that and that hasbeen implemented very often.
I mean, even you know it's justa long way of explaining what
an algorithm is.
But, for example, google searchis just an algorithm.
(05:52):
I think that it's so enchanting.
I think is like the term.
It's like I'm just, you know,you go on there.
I mean I used it to write likewrite me a bedtime story.
That was like the first time Iwas like give me like a five
minute bedtime story and it doesit for you and it's a
completely original one.
Dan, do you remember like whatyour kind of early experiments
(06:15):
were with it?
Speaker 4 (06:16):
So one of the ones that kind of opened my eyes the
most.
I'm also a reef tank enthusiast.
I have a reef tank at the house,like you know, do tropical fish
and coral and all that.
It's a lot of water parametersand dosing this, that and the
other chemical.
So it'd be surprised on howmuch knowledge it has, just from
its short time being around tohave, you know, absorbed all
(06:37):
that.
Where you can ask it a question, as if you're standing in the
aquarium shop, and it'll giveyou answers on why this
parameter should be what andwhat chemicals.
It won't necessarily recommendproducts or anything, but you
know it can help you with thescience behind things and it's
super cool.
It's just so helpful.
That was my first like personaluse of it where I was just like,
(06:57):
okay, this thing is prettydifferent.
The next one I would say,without getting too far ahead,
was being able to throw code atit and watch and manipulate or
refactor a piece of code.
Say, you've got a littleJavaScript snippet that you know
changes the phone number tobold or whatever.
You can pop that in and ask itto refactor and it'll rewrite
the whole thing with less impacton your site or whatever you're
(07:19):
adding it to.
It's super impressive.
Speaker 1 (07:22):
Yeah, it's like the problem solving is.
What's fascinating to me islike I mean, it's for a really
long time when I was learninghow to code.
I still don't know how to code,but I would have to ask Dan
every single step of the HTMLprocess.
You're like, hey Dan, I havethis image.
I want to add a very thinstroke to it for this blog that
we have on punchmarkcom.
How do I do that?
(07:43):
And Dan would have to turnaround, be like, okay, it's
going to be style equals quotes,blah, blah, blah, blah.
And then I would do that and itwould work.
And then I would have to askthe next thing and it would
always pull Dan's attention awayfrom what he was working on For
a recent blog that we made.
It was announcing, you know,some promotions at punchmark.
I was just like, oh, let mejust see if I can just ask you
(08:03):
to chat to you this, and that iswhat it is very good at.
There are some things.
Maybe, Dan, can you like kindof shine a little bit of light
on some things it's not great atyet.
Is that possible?
Speaker 4 (08:15):
So it has a stop of knowledge after 2001 or 2021.
So one thing I try to do thisyear was I'm sure a lot of
people play fantasy football.
I was going to start asking itabout.
You know which players should I?
Speaker 3 (08:28):
draft.
Speaker 4 (08:29):
You know, here's the mode I like to get it out, but
it doesn't have access toanything current.
It doesn't have access to theinternet at all.
It only had access while it wasin its sort of hibernation
state.
So anything you ask it sort ofhas to be either timeless, data
problem solving or creative.
It's not going to give you,like you know, current events or
(08:49):
interpretations of anythingcurrent or stuff like that.
So that's where it kind offalls short For a good reason,
though I think that's where itstarts to get pretty scary on
the kid of power If it did haveaccess to the internet and what
could happen with that.
So you know, that's for a goodreason.
Speaker 1 (09:05):
So I guess let's dive into how Punchmark has
implemented it into Site Managerand then we'll get into all the
exciting different things andthe possibilities and how we're
using it.
For example, dan, walk methrough the Ask AI tool that
Punchmark rolled out.
We made a big deal about it.
It was right before JCK Realbig win for us.
(09:26):
I was really excited about itbecause we were one of the first
to kind of bring it to market,if not the first.
Can you just kind of give me anoverview and like what that
means?
Speaker 4 (09:35):
Sure, so our first goal at Punchmark was to add
content creation capabilitieswith AI, so we incorporated this
into Page Builder.
So when you add a header, youcan ask it to regenerate or
generate you a header based oninformation on the page, as well
as take a header and thengenerate a description text to
go with it and then, in the blogas well, you know, allowing it
(09:57):
to generate full article textbased on a prompt that you can
add.
At the moment, we've met keptthe prompts pretty simple.
It's nothing too complex, but Ithink as we go forward, we'll
probably start to feed morecustomer information into that,
those prompts behind the scenes,so that it can get a little bit
more specific to the customerasking and not just the same for
(10:18):
all.
So it's, you know, while it's abig win for Punchmark that we
have this capability on thefront end, I think there's a lot
more cool stuff we're going tobe able to do with it now that
it's available.
Speaker 1 (10:28):
Yeah, and Alex, I know you were one of the first
people that I actually saw thatI hadn't explained the tool to,
that actually took the tool,learned how to use it and then
started implementing it.
Can you talk me through, firstof all, what the learning
process was and also what thefirst thing you started building
with it was?
Speaker 3 (10:47):
Yeah.
So I mean I have a little bitof knowledge because I have
played around with chat GPT.
So I will say I had a littlebit of background knowledge.
My husband, like loves thiskind of stuff.
So he told me about chat GPTprobably at least six months ago
, if not even a, maybe a littlebit further back.
But I saw the email blast thatyou know this was going live,
and I was like, oh my gosh, thisis so cool, I can't wait to try
(11:09):
it out.
I just jumped in there.
I really didn't know exactly howit was all going to work, but
it's really, I mean, easy tofigure out and even for someone
that I think is not techie, Ithink could figure it out.
So it's on the page builder.
You just click the button inthe top toolbar.
It says ask AI.
And I first just put reallysimple stuff in there just to
(11:30):
see what it would come out likeyellow gold ring, what is it
going to put out with for that?
And it did it like itelaborated, and I don't know how
it pulled some of it, becauseit was it started saying like
five grams with diamonds and I'mlike, where did you get that
from?
Speaker 4 (11:45):
I didn't say how much this weighs or anything.
Speaker 3 (11:48):
So it is kind of funny.
But so you definitely have toedit it and proofread it and all
that.
But it is interesting that itcan prompt you on what details
could further the description orthe page as well.
So even though I didn't put inthere the weight of the piece,
it now put something in thereand I'm like maybe I should put
the weight of the piece.
(12:08):
Let me go figure out how muchthat piece actually weighs, and
maybe that is useful information.
So I think it's good for that.
But now I've been using it a lotfor writing my blog posts and
copy on my landing pages andit's been.
It's saving me so much time.
I think it would take meprobably a good two hours to
write a blog post because I liketo be very thorough and I do
(12:31):
like to do the research, andthen I'm going to proofread it
and I'm going to like go, youknow, simmer on it and come back
to it again.
I wrote a blog post in like 30minutes the other day using the
ask AI and I do see a little bitof repetition sometimes with my
paragraphs in there.
So I just go, you know, you gotto go, fluff it and everything
and take out the redundancies,but overall it was really a
(12:55):
great basis of where to startanyways.
So, for example, I wrote a blogpost the other day, using it
that I did the top weddingjewelry trends of 2023.
Speaker 1 (13:05):
That was the one I saw.
I loved it.
Speaker 3 (13:07):
Yeah, and I already had kind of some ideas in my
head of where I wanted to gowith it, so I didn't let it do
the whole thing from scratch,but I just said write me and
write an intro paragraph abouttop wedding jewelry trends of
2023.
It wrote the intro for me, thenI fluffed it and edited it.
Then I said, okay, write me aparagraph about rose gold
wedding jewelry.
(13:28):
It wrote me the paragraph, Iedited it.
So that's kind of how I wentabout it.
So I gave it, you know, someguidelines to look for, and
sometimes I gave it more or lesslike write me a paragraph about
wearing black diamonds for analternative wedding jewelry,
look.
And so the more you give it,the more you're going to get out
of.
It is definitely what I'mfinding too, because it can only
(13:50):
go so far.
I can't read your mind, youknow.
So I am curious Does the ask AI?
Does it have the capability oflearning or following any kind
of patterns that you might useit for?
Speaker 4 (14:02):
So we can.
We have not done any model,which is called this training
data, so we haven't starteddoing training data yet on this.
However, it is all using thesame API key, so that means all
of our customers are hooked intothe same exact bot.
Basically, probably the waywe're going to do it is is we're
going to break it down byfeature type and then start to
train the models based on eachone.
(14:24):
So, for instance, contentcreation you know all the
different things that jewelersare putting in there.
It can start to learn.
You know that these are alljewelers for one, or that you
know where their store existsand things like that.
And then there's things likeproduct descriptions, where you
know it's not going to be thatat all, but it's going to be
more like how do we pull thisinformation out of their
database, their cataloginformation, their edge data,
(14:48):
and make sure we can utilizethat data when we're training
that model?
So each one is more specific tothe task that it's trying to
achieve.
So, yeah, that's kind of ournext.
Our next trick at a bunch morehere is starting to
individualize and train themodel so that they all start to
learn from each other.
We probably don't want them tobe so co like, so intermingled,
where your creativeness is thenbleeding into other you know of
(15:12):
our customers, because thatstuff, I think, is private to
you.
That's your prompts, that's theresults you got.
However, you know the thingsthat we can do to help you get a
leg up.
We definitely want to befeeding feeding to the model.
Speaker 3 (15:23):
That's interesting, and so what was really the I
don't know the mindset behindyou guys wanting to put it
straight into punch mark versusjust, I don't know, helping us
figure out how to use chat GPTfor stuff like this?
Speaker 4 (15:38):
For me.
I was an 80s kid, right, soprobably the very end of Gen X.
Maybe you call me a millennial.
On the first year, you know, Igot playing in the dirt and
computers came out when I waseight or nine years old and I
was in love right away and Ichanged everything I was doing.
I immediately stopped doingnormal art, started doing
computer art and started myjourney on that and what an
(15:58):
impact it had on the world,right.
So I kind of see AI as anotherone of those that's happening in
my lifetime.
So, as the CPO of Punchmark, I'mgoing to try to make it my
mission to rebuild our platformto utilize AI technology in
every single nook and crannythat we can possibly put it into
, because I think a lot ofpeople are maybe nervous about
(16:20):
AI and implications of it.
Where this is going am I goingto lose my job, all that kind of
stuff but I think the peoplethat are utilizing it now and
starting to understand it thatit's just a tool, that it's not
going to take over the world andthen be a Terminator 2, skynet,
it's really just.
You know, it's a reallypowerful tool and if you
understand how to use it and howto talk to it.
It can double your capacityeasily, within no time, so you
(16:44):
can bring that same power theentire platform, from page
creation to products, to imaging, to everything that we do.
Speaker 1 (16:51):
I saw this quote.
It was from Henry Ford, ofcourse, you know, inventor like,
I guess, the guy who madepopular the automobile.
He said if I had asked themwhat they wanted, they would
have responded faster horses.
And I was like you know whatit's a fair point, like if
people, what are people askingfor us?
We want to make you know, wewant more website sales.
(17:13):
But it's like okay, like let'sthink beyond that.
And to me it's like let's takethe manualness out of a lot of
these things that are just hypermanual.
Like you said about writing ablog post, like you know, about
you know, two hours, two and ahalf hours, for for me, like
it's also kind of goes almostbeyond that where it's about
topics that not only am I justnot a super expert about, but
(17:36):
I'm the guy who usually writesthese blog posts for punch mark.
So it's going to take time fromother people and a lot of times
I just have to drag on them.
So, for example, like we'reworking on this thing about ADA
compliance I don't know anythingabout ADA compliance.
I'm going to need to ask Dan.
Dan is in the middle ofbuilding all these websites, so
(17:58):
like it's going to take both ofus away to write this blog post
and instead it's like let me get95% of the way there, we can
tweak it a little bit and thenrun it by Dan, and then Dan can
play editor and it just is goingto put us on third base instead
of at that.
So we're going to take a quickbreak and hear a word from our
sponsors.
Stay with us.
(18:21):
Introducing Rembrandt Charms,the family owned company with
over 50 years of charm and charmbracelet design and
manufacturing experience.
Explore the world's largestcharm collection in sterling
silver, gold plate, 10 and 14 K,yellow and white gold.
The sterling silver merchandiseis also rhodium plated to
prevent tarnishing.
All Rembrandt Charms productscome with lifetime warranty
(18:45):
showcasing their dedication andcraftsmanship, available
exclusively through authorizedretail jewelers.
Find a store near you on theirwebsites.
Find a retailer section atRembrandtCharmscom To be listed
on their retail locator.
You'll need a retail jewelrystore with a Rembrandt Charms
display or shop and shop.
Our business building charmprogram offers inventory
(19:07):
controlled displays tailored toyour store, along with marketing
support, e-commerce assistance,stock balancing and more.
Contact Rembrandt Charms todayfor more information at
800-828-7840.
Again, that's 800-828-7840.
Until in the loop sent yaAlright and back to the show.
(19:30):
And we're back.
I guess I kind of want to bringthat through and talk about the
next step, which is, you know,ai driven imagery.
It is through this very popularAI model called mid-journey.
Then can you kind of give thelisteners like an idea of what
(19:52):
mid-journey is and also kind ofwhat makes you so excited about
it.
Sure, mid-journey, it is an AIimage creation tool.
Speaker 4 (20:00):
Right now it's exclusively on Discord Crazy.
So basically a message boardthat's hooked up to code and a
bunch of people working on it,pushing new features and
releases to it.
Anything you create on there ispublic domain, so nothing
you're creating there is goingto be able to be copywritten.
It can be stolen by anybody.
However, it's a great place togo in and brainstorm to.
(20:21):
You know, as a graphic designer, mike, you'll know this too,
but when you're coming up withlogos for somebody, you go
through a lot of concepts, a lotof ideas, a lot of like what if
we put a diamond on a K?
What if we put a?
You know all these differentthings that you're going to
think of, and sometimes you wantto see it before you commit to
(20:42):
the concept.
And you know back in the day theway we were trained.
You would do quick sketches.
You'd probably jump out of yoursketchbook and just start
drawing and stuff, which isstill obviously good, but to be
able to jump in there and justtalk to something and say, all
right, give me this, give methat, give me this, give me that
.
It really lets you move reallyfast on something and decide
when you hate something a lotfaster, you know, in a minute or
(21:02):
two.
So that's one example, but theway it can fake people and do
places and things is justabsolutely incredible.
So a good example of what I wasactually doing last night we're
working on a new campaign forAntwerp Goers and helping with
the marketing of that, so Istarted it and it's also one of
those things that's really hardto get images of because we're
(21:25):
not in Antwerp and stuff iscopyrighted in our editorial so
it's very difficult.
So just get in there and starttalking to it and say you know,
give me a picture of a jewelerin Antwerp meeting with diamond
cutters in a private office,touching round brilliant stones,
and immediately get into thebeautiful scene.
That's not perfect.
Similar to chat.
You can see it's not going tobe perfect right away.
Like the first ones, we weregiggling yesterday because the
(21:53):
rocks were like yeah, literallythe ones right now, right now.
And it's just like okay, likethat's a little crazy, but the
fact that it did that is awesomeand some of the scenes that it
made of like a late night inAntwerp with the moon in the
back, it's just, it's gorgeous.
So, like I saw some really coolstuff on JHJ as well, where
customers were creating customjewelry concepts with it.
(22:15):
You know where.
You know it's like here's thedifferent ring designs I might
do with somebody and they cankind of see it right away
without you guys having to gothrough and do a CAD and all
that extra work.
I mean, this one is stillrelatively new, so the API is
kind of difficult.
The API meaning the way weconnect it into the platform.
I think it's a little bitdifficult, but I think when it
gets there, this one's going tobe an absolute game changer, as
(22:36):
cool as ChatGPT is right now.
Speaker 1 (22:38):
Yeah, I just like the idea of being able to do quick
and dirty you know, concepts.
That's the thing is like.
The other part of it is justlike for you know, a social post
.
I'm sure that the next step forAlex is going to be like hey,
how can I get this to write,like you know, better
descriptions or more engagingcontent and stuff like that.
(22:59):
You know, do you see that kindof coming around Alex?
Speaker 3 (23:02):
Yeah, I've actually been playing around with that
already and actually my emailservice provider has a.
It's not completely AI, butit'll like suggest subject lines
for you and it's really nicebecause, I mean, I have to come
up with a lot of content all thetime, like at least one social
post every day.
Sometimes I'm doing multipleposts a day.
I'm usually sending out atleast one eblast almost daily.
(23:25):
I you know I start to feel likeI can't think of any more new
subject lines or captions orwhatever.
So it does help with thewriter's block, you know, when
it comes to all that.
So I'm really excited for thatto continue and to get easier to
use it for all of that.
Speaker 4 (23:40):
It really allows everybody to kind of be a
director, right.
So I mean, it used to be backin the day like you had to be
really good at doing the thingand coming up with the ideas,
and now it's kind of just, it'sgoing to start to level the
playing field where it's justlike, if you've got good ideas,
the execution side can kind ofbe handled pretty well by robot.
It doesn't mean it's going tobe perfect and you probably
(24:01):
still want a professional tocome in and do your final, but,
man, it can get you so close tothe finish line.
Speaker 1 (24:06):
Yeah, I guess maybe just staying on that point.
I think that there is somethingimportant about, like the.
I think that the human elementof this is the idea of editing
or refining or there's a lot ofthose kinds of concepts where
it'll get you I keep saying,it'll put you on third base, and
I think that with mid journey,you know you go through and if
(24:29):
you're not careful, you'll postsomething of like a model with
you know, six fingers on ourhand.
That's very common, because itdoesn't really quite know what
the hand looks like.
His hands are weird.
Same thing with but we've beenseeing it with models wearing
jewelry is kind of the thingthat I kind of hope will be the
(24:50):
next kind of step, becauseshooting models wearing jewelry,
especially jewelry that iseither a state or one off or you
know, et cetera, is verydifficult.
I guess I'll pose this questionopenly to you guys when do you
think the you know theresponsibility is, or the humans
, to be the editor?
I think that we need to makesure that the content that we're
(25:12):
putting out is quality and notrushed and is not entirely from
a robot, because I definitelyfeel like that's important.
Am I?
Am I crazy?
Speaker 4 (25:23):
No, not at all, and I think that's where you know
refining it as you go is veryimportant.
So, like with mid journey, youknow it is giving you a
finalized image.
So it's very tempting to use,but a lot of times it's, you
know, like you said, with notjewelry on models.
We did one where it put thediamond stud right in your cheek
and it's like I don't know ifanybody is getting that stud put
in.
you know so six fingers, youknow just all kinds of things.
(25:46):
It'll put words in thebackground that shouldn't be
there, like yeah, you definitelywant to play editor and keep
messing with it.
And that's where I think reallyyou know.
Why playing with it is soimportant is because we're all
really training ourselves tobecome prompt engineers, as they
would say, where you'relearning how to talk to the
robot.
So you're not just like give mea picture of something great
and then you just wait forsomething to come at you.
(26:07):
Like you have to really tell itspecifically what you want and
when it gets it wrong, you sayno, no, no, not like that, like
this, and you train it.
So that's, I mean.
I think that's really the keyis making sure you're always
doing that and then, when youget a final product, making it
your own.
Speaker 3 (26:21):
Yeah, and at some point you're going to your
audience can tell you know, likeif you don't edit it, they can
tell you're using stuff that'sbeen generated by a robot and
you're going to lose followers,you're going to get unsubscribed
, you're going to lose customers.
So you've got to still put thathuman element into it, or
people are going to be justglazing over.
Speaker 1 (26:42):
Yeah, yeah, related to that, alex, I know that like,
like a big part of marketing ishaving this voice, and it's one
of those things I actuallytried to explain to my parents
recently what what voices, andlike I tried to explain it
through the context of mypaintings where I was like, oh,
I have like a style, it's like avoice, and they really didn't
seem like they.
(27:02):
They got that.
I was like it's kind of like my, like a little bit of me,
that's in there, and I tried toexplain it about, especially for
writers and how, like certainwriters have like this wit.
That's not them trying to befunny, but it's like almost how
they frame the world thatthey're building.
I know that when you speak withjewelers, a lot of times you're
talking about you got to find,like, who you are.
(27:24):
Do you think that that kind ofbringing the voice into these
kind of generated content isgoing to be something that's
even achievable, or how do youthink that works?
Speaker 3 (27:37):
Yeah, definitely, and actually I think chat GPT has
the capability of doing this andyou can actually take, you know
, some text, feed it into chatGPT and say what is the voice or
the tone of this writing, andit can identify that for you.
And then you can go back andsay now write me something in
that tone or voice.
(27:57):
So that is important to knowwhat yours is.
And it's one of those thingslike no one can really help you
with.
You have to figure that out onyour own, especially for your
own brand.
Like do you want to be funnyand witty?
Do you want to be serious?
Do you want to be buttoned up?
So you kind of have to figurethat out before you move forward
with using this.
You know in a big picture, orelse you're not, you're going to
(28:18):
lose your identity 100%, andactually mid-journey does the
same thing.
Speaker 4 (28:23):
So there's actually a command you can feed to
mid-journey called describe,where you just upload the image
in and it'll just fit back textto you that gives you what a
prompt might look like to createthe image you just submitted.
Then I really teach you to belike okay, this is what it's
expecting when you want thatkind of result.
Like you said before, Mike,it's really like you want to
give it as much information aspossible and some of that is
(28:45):
stylistic.
Speaker 1 (28:47):
That was like the thing is.
So earlier today I just have itup right here Dan had put in
one of a headshot of himself anda series of professional
photography that we had done ofourselves.
It created these cartoon, coolbackgrounds that really
juxtaposed the image.
It was really interesting, Iguess.
(29:09):
I'm like would it be able tothen create a set of them?
Because part of somethingthat's hard to fake is, I guess,
consistency across multipleiterations.
A lot of the times I feel likethat might be where it falls
short.
I don't know.
It has almost too many optionsthat it can't really be
(29:33):
consistent.
Speaker 4 (29:35):
Yeah, I think when you're using the open source
tools right now, that'sdefinitely a problem, because
you can only train it so muchwithin the chat.
But again, that's where themodel training comes in, where
you can give it specific stuffand say this is your brain and
anything else is invalid.
So a good example of that isproduct description.
That's one we're doing where itsucks in all the different
(29:56):
variations of your product andsizes it's available in and
whatever, and generates a nicedescription that's unique to
that product alone.
Another one is image reading.
So in Page Builder I knowthere's another one that
everyone hates doing is addingalt tags to all their images.
Everybody, please do that.
It's very important.
We try to do some stuff in PageBuilder that'll maybe take the
(30:17):
headline of your widget and tryto create the alt tag.
But in the future one of thethings we're going to be hitting
is something just likemid-journey, where it can read
the image and tell me exactlywhat's in it, based on the
pixels.
Speaker 1 (30:30):
There's so many options for it.
I think that the timeline is alot shorter than what a lot of
people are expecting.
I think that the first time Iever heard that the phrase
chatGPT was probably about maybethree months ago four months
ago, it's tough to say, but thefact that we've come this far
(30:50):
where we can generate images injust about four months, it's not
going to get worse, it's onlygoing to get better.
Dan, where do you think thisimplementation falls on?
As far as the focus forPunchmark and also the timeline
it's tough to say timeline.
Speaker 4 (31:05):
We're trying to tread lightly, I would say we're
probably going to do a lot ofrefinements to what exists for
now and making sure theimplementations we already have
are getting better and we'refine tuning them.
As I mentioned, productdescriptions, alt tags those
will be some ones that will rollout in the near future One of
the most exciting ones, I think,beyond mid-journey, which I
would love to implement as well.
(31:25):
I don't know if the API isready for that, but if it will
be, I'll do that one as well.
But page builder being able totalk to site manager and say I
need a page about this and haveit, not only write it, but
generate a page with a herobanner and a call to action and
content in the middle and knowwhat the code of page builder
looks like, that's something I'mstarting to personally work on
(31:48):
now.
I mean, that's an absolute gamechanger for our customers, for
us at Punchmark, and that'sgoing to take just straight
model training, where we'refeeding it exactly how to build
a page, all the possibilities,what not to do, what you're not
allowed to do.
So it's going to take a while,but I think the end result is
going to be absolutely gamechanging for everybody.
Speaker 1 (32:08):
And Alex are you.
You know things are changingand, as a business owner, do you
see this as an opportunity foryou to be teaching your
employees how to leverage thistool?
Kind of a weird kind ofenvironment for it retail, you
know how I it's very wellequipped for the online presence
, but is there a way to sort offit it into the in store
(32:33):
approach, or what have youthought about that?
Speaker 3 (32:36):
That's a very interesting idea and I haven't
put much thought to it yet, tobe honest.
I mean, I do believe stronglythat everything I do more on the
back end supports my front line.
You know, the more complete ourwebsite is, the better the
descriptions are on our website,the more the customer is
already closed before they comein the door, you know, and it
(32:56):
answers a lot of those mostcommonly asked questions ahead
of time.
But that will be reallyinteresting to see Like.
I mean, we're so like in theinfancy of all this, so I'm
excited to see, like what allthe applications are going to be
of it.
I do have a couple of myassociates that are interested
in it and trying to figure itout, so I will definitely report
(33:17):
back if I find anything ofvalue Nice.
Speaker 4 (33:20):
Well, I'm curious if a customer came in with a bunch
of mid journey ideas for rings,would you guys take that?
Or would it be like, oh my God,get out of here?
Speaker 3 (33:31):
Yeah, I mean, why not ?
It's not really any differentthan if they sketch something on
a napkin and brought it in andasked if we could make it.
It does post some questions asfar as who owns the design and
stuff like that.
So you know, and luckily our,our custom designer that we work
with does some of that legworkfor us to make sure we're not
(33:52):
using copyrighted designs.
But I'm it's going to beinteresting to see, like, how
that's all going to play out,and I'm sure we're going to have
some new, some new laws andregulations based around all
this.
So we'll wait and see.
Speaker 4 (34:06):
We've toyed with the idea of implementing it,
whenever it's available, apiwise, into the our wizard where
you can kind of start a customproject on the website, sort of
giving them access to midjourney right there.
That would be so cool.
Speaker 3 (34:18):
It was starting point , like yeah, could you imagine
if you could actually type inthere like I want a three stone
ring with these birthstones andI want it yellow gold, but with
this with a swirl, and it justproduced an image for you right
there?
That would be mind boggling.
Speaker 4 (34:33):
It's super easy to do , but I think I think the
question for me as a developeris like is there they're going
to be submitting stuff that youguys are like this is impossible
, you got a giant.
Speaker 3 (34:46):
It would have to definitely have some disclaimers
.
Like you know, this is all preproduction or sales associate or
designers would need to lookover it, whatever you know,
before putting into production.
But I'm sure you know, I thinkpeople all understand to some
extent that this is all like Idon't know what's the word, like
it's aspirational, I guess youknow.
(35:07):
And it's just a place to start.
Speaker 1 (35:10):
Yeah, I think that it's again.
I am very focused, I think, onI'm trying to find new ways to
use this and, like Dan said,he's not just, you know,
preaching this to everybody else.
We have actually internalizedthis at Punchmark.
We've encouraged all thedifferent departments at
Punchmark to find ways thatthese tools might be able to
(35:33):
make their jobs easier.
Because it's kind of like, youknow, when there's a manual
approach to something and thenthere's like a scaled up way
that uses technology whetherthat's shoveling your driveway,
and it's like, oh, I only shovelby hand it's like, well, you
could just use technology anduse a snowblower, like you're
(35:54):
not proving anything by doing itby hand.
It's not like it makes it just,especially if it's a manual
thing.
But the one of the things thatI kind of wanted to explain
about how I'm using it just as areal life example, not just at
Punchmark, but actually I'musing it for a, for this podcast
, and I use a plugin that is inour podcast hosting tool called
(36:18):
buzz.
What we use is Buzzsprout andthey have a tool that they just
rolled out called co host AI.
And just as a kind of an ideaand I kind of want to just
mention this because then peoplewill be aware of why and how
we're using it, and I'm a host.
Ai does, and I've only done itfor one episode so far at the
time of recording this, but thisepisode will be on.
It is.
(36:39):
It writes my episode titles,episode description, transcript
and chapter markers.
So to break those down and whyI use them.
It as an example of why an AIis going to be able to help help
.
It is not changing the way thatwe are making this and producing
this content.
It's just taking away themanualness of it.
(36:59):
So, for example, chaptermarkers are one of the key
things that I've read and amillion places that will elevate
your podcast.
I've never done them, or Ican't keep up with them, because
you have to go through and youhave to listen to the entire
podcast again and then annotateit and write these anchors for
the episode.
You know, maybe you need fourchapter markers and it's just so
(37:23):
manual and I just can't do that, but it will do it for me.
But when it comes to episodetitles, it's like, oh, but maybe
you're going to be missing someof the spin.
It's like, yes, but what itwill allow me to do, it gives
you like six options.
Go in and then you can touchthem up from there.
That's the kind of use I amtrying to find is that it will
(37:43):
take a lot of the manualness outof it, or it will put me on
third base and allow me to justlightly edit and go from there.
So that's kind of like a reallife use case that is tangible
to a listener right now.
So all of that to say thank you,I think it's fun.
This episode will be on it.
I don't know if that's meta asheck, but whatever you're going
(38:04):
to do All that to say, I thinkthat maybe kind of leave it as
far as you know the generalconsensus and feeling around
this tool, I'm excited, but I'malso slightly apprehensive.
I really want to maintain thehuman element of everything I do
, because I do think that thatis what is special about you
(38:26):
know, the work I produce is thishuman element in the voice,
naturally, and you know, youknow metaphysically, do you guys
?
You know, alex, how will youapproach AI in the future?
Is it like, hey, let's divehead first, or is there some
kind of restraint that you'regoing to set up?
Speaker 3 (38:47):
Oh yeah, I mean, I think with anything that you
implement in business, you can'tjust jump in.
You've got to figure it out.
And especially with this,because it's so new, I'm dipping
my baby toe into it, you know,with the, with the ask AI, and
we'll go from there, you know,and it's might be a live and
(39:07):
learn kind of thing, you knowand kind of see how things go
and if I might have to pull itback at some point, you never
know.
So that's kind of my plan, justto kind of see, and I'm
actively watching what otherpeople are doing with it too.
There's a lot of people talkingabout this topic right now and
I think we can all learn fromeach other and, you know, see
where it goes, what about you,dan?
Speaker 4 (39:28):
Yeah, agreed, I mean I would say you know.
One thing I like to compare isyou know, people like you know
when Google came out or theinternet, I guess Google is kind
of people would say, oh,there's too much information out
here and you know why.
You know you can learn anythingonline.
Now, you can take a class, youcan do this, you can do that,
and the ask is really going tokind of be the same thing, right
(39:49):
, like I find myself alreadystarting to lean towards asking
chat.
You can see the answersquestions over Google as of
right now, and not because it's,you know, got better access to
things, as it does on Googledoes.
But the way it can kind ofsummarize things and simplify it
and make it accessible, I thinkis unique, and the more that
(40:11):
grows, the better it's going toget, and it's going to get.
It'll probably get to the pointwhere, if I'm sure, google will
roll out an AI at some pointthey're probably making it as
slow as possible because they'llhave the be all end, all of
information to feed that thing.
But I think you know, I thinkthat's that's the way of the
future.
This is a tool that we're allgoing to have to sort of get
used to, and it's kind ofamazing in a lot of ways.
(40:32):
It's scary in other ways, butI'll you know I think I'll leave
some of that to you know,whatever government entities are
going to get into their race ofcreating AI for nefarious
reasons, like, but you know,that is what it is and I think
everyone has a right to benervous about that and
apprehensive, but I wouldn'tdiscount the entire tool as a
(40:52):
result of that.
I would say that it's still atool, just like a hammer.
Pick it up, give it a shot.
It's going to help you.
Speaker 1 (40:58):
You know just like the internet there's like a lot
of is what you make of it.
You know the internet for everyexcellent example of, like how
I was able to show my paintingstream to my grandma, who lives
700 miles away, and have herwatch me paint.
I mean that's amazing.
But for every example of that,I'll show you an equally
horrible example of how theinternet has been used.
(41:20):
And we live I live perenniallyon the on the internet.
What are we going to say, alex?
Speaker 3 (41:26):
I was going to say the same thing.
You're saying, actually, thatjust like the internet.
Speaker 1 (41:31):
You know, just because there are terrible uses
of the internet I'm not doesn'tmean that we should stop making
websites for jewelers.
Speaker 4 (41:40):
So you don't throw the baby out with the bathwater
and there you go, there you go.
Speaker 1 (41:44):
Honestly, you know what.
Maybe we'll leave it out thatthat's a great point.
Alex Dan, thank you so much forbeing with me.
I think that this is kind oflike a more introspective
episode is very interesting.
I kind of hope everybody triesand even if you decide that you
don't like it and you know what,at least you know I think that
it is so accessible that youmight as well try, so that way,
(42:07):
when you form an opinion, you'llhave an informed opinion.
Speaker 4 (42:11):
We should do a chapter two someday when we when
we get to the next advancementnext year.
Speaker 3 (42:17):
When we've all been replaced by robots.
Speaker 1 (42:20):
Beep boop?
Absolutely, I think it would begreat.
I really like doing these kindof episodes.
Yeah, we'll set this.
We'll do another episode onthis Next time that there is
like a noticeable change whenGoogle comes out with their
whatever barred or whatever it'scalled.
It's going to be, I'm sure,better than everybody else's,
but in the meantime I'm excited.
I'm going to go play aroundwith mid journey, have it,
(42:41):
conjure up some cool rings andwe'll take it from there.
Take what we got.
All right, everybody.
Thanks so much for listening.
Appreciate you all.
New episode next week, tuesday.
Bye, all right, everybody.
That's the end of the show.
Thanks so much for listening.
This week's episode was broughtto you by punch mark and
produced and hosted by MichaelBurpo Our guest this week was
(43:03):
actually in wine, as well asDaniel Saroi, the chief product
officer at punch mark.
This episode was edited by PaulSuarez with music by Ross
Cochrum.
Don't forget to rate thepodcast on Spotify and Apple
Podcast and leave feedback forus at punchmarkcom.
Slash loop.
That's L-O-U-P-E.
We'll be back next week,tuesday, with another episode,
(43:23):
cheers.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.