The AI revolution isn’t coming—it’s already here. And businesses that fail to adapt are at risk of falling dangerously behind.
In this compelling conversation with Dr. Iyanu Odebode (CEO of Wokkah) and Jason Wright (COO of Wokkah), we explore how artificial intelligence has evolved from niche academic ideas like “pattern recognition” and “data mining” into real-world business tools that are transforming industries.
Ever wondered what an API actually is or how it works?
Iyanu and Jason break it down in simple terms—showing how APIs, the “digital connectors” of our time, enable powerful automation and productivity across platforms. As Jason puts it:
“The businesses that get it, the businesses that embrace this, are going to have such a competitive edge over the businesses that don’t. It’s not even going to be fair.”
From producing 900+ educational videos in just three days to streamlining content creation and social media workflows, this episode showcases the incredible efficiency AI brings to marketing and operations.
Concerned about data privacy?
Dr. Odebode also discusses how AI can be used responsibly in sensitive industries like healthcare—through proper governance and secure, isolated environments.
And if you think AI is here to take your job—think again.
This discussion redefines AI not as a replacement for people, but as a force multiplier that frees up humans to become strategic orchestrators.
Want to see Iyanu’s passion for technology and leadership in AI? Watch his TEDx Talk: https://www.youtube.com/watch?v=8TzSs9ejxdY
Learn more about Wokkah at: www.wokkah.com
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Ready to do this?
Yeah, let's do it we don'tsmoke anything in here like Joe
Rogan, that's good.
Speaker 3 (00:04):
Life Unscripted with Kevin Shipp.
Speaker 1 (00:09):
Welcome to Life.
Unscripted.
Thanks, kevin.
So this is like a podcast thatsometimes goes off the rails.
It's not scripted, it's nothing, I don't plan anything, but I'm
really fortunate to have youguys here today.
Iannu Odibodi, you got it right.
Yes.
Speaker 2 (00:26):
Jason.
Speaker 1 (00:27):
Jason Wright From Walker.
Yeah, and tell me a little bitabout Walker.
Speaker 3 (00:35):
So we saw a gap in the market and we realized that,
as AI begins to become a thing,that as AI begins to become a
thing, that there will bedisplacement of jobs for many
people, and that gave us a lotof concern, and that's one of
(00:57):
the reasons why we chose tocreate a company that would be
really focused on equippingtalents to fill this gap and
then positioning them to getopportunities with companies all
around the country.
So our passion is driven by that.
But the core of what we're ableto accomplish is, you know,
think about building AI agents,vertical AI agents, think about
(01:19):
automation and where automationis going.
I mean, you own a marketingcompany, so you can imagine how
automation can just helpaccelerate the work that you
currently do.
So these are the issues that wesaw and we realized hey, look,
if we can train multiple or morepeople to be able to address
this problem, then the worldwould be a better place and we'd
(01:42):
be able to see acceleration ingrowth in business growth from a
small business perspective, andeven from a mid-sized to
large-sized companies also canbenefit from it.
Speaker 1 (01:53):
So backing up a little bit too.
First of all, a lot of peoplethey're still scared of AI and I
think there's a lot of unknownbecause they're not familiar
with it.
Yes, ai coming out is verycomparable to when electricity
came out right, and it's soprofound and it's made such a
(02:17):
large impact at scale to thisdate that people are still
scared because there's a lot ofunknown.
So that's kind of why I wantedyou guys to hang out with me
right this episode.
Um, because I'm down here anduh, but I'm like the everyday
consumer.
Like you said, I have amarketing business and I have
clients in multiple industries.
(02:39):
So I'm trying to come up hereand knock on your door in the
cloud and say, hey, how can Iuse what you're doing in my
everyday processes?
So the way you've explainedwhat WACA does is the way you
(03:00):
explained it to me is you justtake a lot of the AI tools and
then you develop APIs and pullthem all together.
So my first question is whatthe hell is API?
Speaker 3 (03:11):
So great, great question.
So API is sort of an interfacebetween whatsoever you're trying
to do and a database system.
What's a database system?
A database system is a systemthat helps store records.
So let's say you wanted toenter information about your
(03:33):
company, your customers andthings of that nature.
You put it in a form.
It sends it to a SQL database.
Now we have like Google Sheets,where people can actually
visually see some of the thingsthat they're updating via Google
Forms right, but a databasestores all this information.
Now, for us to have access tothat information, then we have
(03:55):
to go through an API system.
If we would have to accesssomeone else's you know data or
someone else's information.
So companies that are providingthese services right now,
typically you go in and then youcan use whatever services they
provide you.
But if you wanted to develop itat scale or if you wanted to
(04:18):
solve problems at scale, youwould have to drag and drop all
of these things within thecontext of the software.
So now you don't have to dragand drop all of these things you
know within the context of thesoftware.
So now you don't have to dothat, since you can, you know,
interact with the exact sameservice via an API.
So you make an API request andessentially, the request then
produces a response that helpsyou to solve that exact problem.
Speaker 2 (04:42):
I kind of think of it as like a backdoor into, kind
of behind, the software, so thatway you don't have to go and
type it in from the front of it,you can all access it from the
backend through code.
And so a lot of it, a lot of itis it's it's speeding up the
process, it's you know cause.
Right now, say you wanted touse a piece of software like,
for instance, even chat GPT.
You got to go into a, you gotto go to chatgbtcom, you got to
(05:04):
log in, you got to type in whatyou want.
An API would allow you toaccess it from the back end of
it.
So you didn't actually have togo to the website.
So you have another tool,another piece of software.
You can engage with chatgbtwithout ever going to the
website through that one tool.
And so imagine if you couldhook into all these other
different pieces of softwarethrough one tool.
That's what APIs unlock thepower of.
Speaker 1 (05:26):
So what are some current tools?
You mentioned ChatGVT, and thenI was showing you some of the
video applications I've used.
I think Reveal was one of them,where we insert a static image
and it turns it into a littlevideo.
So I still are using all thesetools.
So can you give me someexamples of what tools currently
(05:49):
exist that you could pulltogether by doing the code and
creating the APIs?
Speaker 3 (05:57):
So I mean one is GPT.
It's an easy one to explain and, by the way, API means
application programminginterface, just for some people.
Speaker 1 (06:07):
I'm gonna need that and I'm gonna put it on the
screen when I edit this.
Speaker 3 (06:10):
Okay okay, awesome.
So.
So, for example, gpt is oneright.
Typically, when you want to aska question, you put in your
information into a chat and thenyou you add your data and you
say, hey, tell me about numberof sales that I had this month
from the data that I imputedinto you and it just does it
(06:34):
right.
But now, rather than having todo that, you could actually
interface through APIs where youactually write a code or write
code in Python, and then it hitsthat API and would ask multiple
data sets, the same exactquestion and give you responses
that could be actually visuallydisplayed and things of that
nature.
(06:54):
So typically you would alwaysstill need someone with some
type of coding experience tointerface with it.
But what it unlocks for a smallbusiness is it unlocks
opportunities to do things atscale and to do things pretty
quickly.
So the things that you used tohave to do and would take you
five hours to do, now you can doit in five minutes because you
have access to one GPT to do theprompting, the data, but also
(07:20):
the API that requests thatinformation directly from GPT,
so that you can get the responsewithout having to interface
with GPT all the time.
Speaker 1 (07:28):
Very cool and I know we talked the other night about,
you know, where does AI comeinto place with human operations
, and you said it kind of atscale.
So that's where this seems tobe really valuable is being able
to create, yes, 900 videos intwo days, versus the human, yes,
(07:52):
and there is a lot of.
You know, the people want thathuman talking to them, oh, yes,
but sometimes we can't make thatin that time, so like,
especially when it comes tovideo editing.
Well, you know, I was tellingyou all my little processes the
other night that I was doing tocreate 90 short videos and that
(08:14):
took me like four or five hours.
Yes, and um, ai could have, youknow, I could have used, like
opus was as a tool, but I stillhave to go to opus and still
have to feed it that video,which is, I'm not complaining
because, by the way, has its apisystem too.
Yeah, yeah, so that, becausethat's light years ahead of you
(08:35):
know what we used to have to doto make all of these clips.
Speaker 2 (08:37):
That's correct, that's correct.
Speaker 1 (08:39):
Um so, but what you're doing, and by connecting
all the dots, you're speeding upfaster and faster and faster
and making that a more efficientprocess.
Speaker 2 (08:50):
Yeah, and I just wanted to throw in a comment.
There you talk about what arethe different tools.
There's probably what?
Over 1,000 different tools outthere available right now, and
my own experience is I came froma, a manufacturing company, and
I ran our marketing department,and you know, one of the
challenges that I had was Ididn't have time to learn all
these new tools.
(09:11):
I was too busy doing my job, andso every day I would come in
and like, ok, how do I get thisstuff done using the knowledge
that I have?
And I was using stuff like chat,gpt, but aside from that, I
wasn't really using a whole lot,but aside from that, I wasn't
really using a whole lot.
And so one of the I think thatthe things that's coming, that
it's already happening right now, is really more of an
automation to where you come inand I no longer have to go do
(09:32):
the things.
We actually use AI to do thethings and I just kind of
conduct the orchestra so we cancreate like automated workflows
that actually, you know, insteadof me going in and creating the
social media posts, I actuallyhave an AI agent that oversees
that, and so how does it knowwhat to do.
Well, I train it, I give it allof my data, I dump everything
(09:52):
in there about my brand andabout what I'm doing as a
company and say here's yourinstructions.
And then I have another AIagent that acts as more of a it
summarizes and it bringseverything into a usable format
and then it passes along.
So we create workflows thatactually step-by-step.
You've got different AI, soimagine chat GBT as just one of
(10:13):
your workers and it has aspecific task, and then you have
.
You can actually line thesethings up and create workflows
that get all of your work donefaster and then so really what
your job is at the end of theday is more of like a manager or
an overseer that says, okay, isthe quality good or do I need
to?
And you can even train it to goback and edit its own work and
(10:34):
figure out what it you knowmight do better, and then at the
end of the day, you basicallysay, okay, what I want from you
is I want you to create allthese social media posts and add
them as drafts in my Facebookaccount.
And then I can go in andmanually approve them to make
sure that they're on point.
So that's really.
You know, a lot of people thinkof chat, gbt, as you know, ai,
but really it's it's kind of thetip of the iceberg, it's it's.
(10:57):
It goes so much deeper than that, especially when you talk about
automation and real worldapplications, because for me, ai
was always this mysterious boxand, like you're, trying to
understand what's in it.
But I think this year it'salready happening.
But this year we're going tosee more of it where it becomes
(11:22):
practical, where businesses cansay, oh, you mean, this could
save me 10 hours per week.
Yes, I'll take that.
Speaker 1 (11:31):
I'm starting to see it creep into I shouldn't say
creep when it comes to AI,because then people but I've
heard of grant writers startingto use it a little bit to apply
for grants and people arestarting to see, oh, this is
(11:51):
actually a wonderful tool whenused properly.
Oh, yes, um, I've seen it inhealth care.
So a client, uh, hospital upnorth, um, they went to a little
speech, little little deal, andwhere they were talking about
ai and health care andmanagement.
And I walked in on thisconversation the other day and I
had to go up there to do somestuff.
But I heard her talking aboutthat and just from talking to
(12:12):
you and you guys, I was like, ohyeah, that's so possible.
I said, hey, I work with thiscompany.
You know we meet every Mondayand they just explained that
whole process to me and I shewas like that was like that's
exactly everything.
I shouldn't have went to thatthing.
Speaker 3 (12:27):
Right, right, I mean right now you could use AI for
training.
You could access tons of datavia API from multiple databases.
Train AI to be able to explaineducational materials, company
materials, company documentation, fact sheets and things of that
(12:52):
nature where employees canactually start to gain real-time
access to how to do their jobsmore effectively.
So this is really powerful.
Speaker 1 (13:02):
With that educational aspect.
A lot of companies don't wanttheir information out there, so
can you explain how you protectthat?
Speaker 3 (13:09):
Yes, I like what you said earlier when you talked
about responsible use of AI.
Not so many people think aboutthis and how to responsibly use
it.
There's a lot of work aroundresponsible and ethical AI lot
(13:29):
of work around responsible andethical AI and lots of the
government put out a sort of apolicy, a NIST policy, around
how to responsibly use AI forall the different, I would say,
cycles of machine learningdevelopment, from data
extraction to data training todeveloping the models to
inference.
All of that might sound likecrazy for some people listening,
but we'll definitely get intosome of that conversation as we
(13:52):
move along, um, and discuss someof this and definitely open to
sharing more of that, you know,on your your podcast and getting
people up to speed with that,but around responsible use.
Now companies are moving towardsusing more open source models,
so the idea behind that is youcan actually bring in a model
(14:17):
like Olamer into yourenvironment and train on Olamer,
and Olamer is completelyblocked from the outside, so
restricted within yourorganization, used specifically
for your own data, has nointerference with the outside
world at all, and you're stillusing this information to help
(14:38):
your staff to do, to createtraining courses, to build
systems and things of thatnature, and so the open source
route is becoming a good optionfor companies that are thinking
about the security of their dataand being able to have that
helps you know a lot, so we'reworking together on a proposal
(15:02):
pretty soon for thoseeducational videos, so that
would be using open source andthen private network.
Speaker 1 (15:09):
Yes, that's correct and you've mentioned before,
basically you're putting a wallbetween that cloud and that
cloud.
Speaker 3 (15:16):
We need to do that, gotcha.
We need to do that to be ableto build responsibly.
Otherwise, you give your databack to the know, the world, to
train and right.
Speaker 1 (15:26):
That's.
That was the big thing.
That's what we kind of talkedabout up there.
Um that hospital was you know,um they, you know they were like
well, how can they not get ourdata, how can they not get our
hip?
Speaker 3 (15:39):
this is the reason why you need people that have
expertise in this specific area.
Speaker 1 (15:44):
So they could do that , so they could do that.
You know anybody?
Speaker 3 (15:49):
we do.
I mean, we, we have thatexpertise.
Um, we have that expertise, umwe've we've done a lot of work
in this.
In this space, I particularlyhave also done quite a number of
work um in this space.
I have a phd in artificialintelligence myself, so the
ability to know where thingsstart to get iffy is really
(16:10):
really critical Also outside ofjust the AI expertise.
I think it's also critical tohave all the stakeholders
involved in the conversation, sofrom the business executives to
the subject matter expertsinvolved in that conversation as
well, because policy doesn'tlie only in the hand of the AI
(16:33):
expert.
You have to sort of think of itas a more collective effort to
get there so the businessmanager might see some, you know
, sides that you wouldn't see.
And this is where people likeJason come in and start saying,
hey, how do we think about this?
How do we look at this?
Speaker 2 (16:54):
And you know.
Speaker 3 (16:55):
So that's my thought process around.
It is yes, have your AIinfrastructure in place, but
also have subject matter expertsthat can tell you hey, this is
what data governance looks likein this space.
This is what we want to expose.
This is what we don't want toexpose.
What does our RBAC?
Rbac meaning, you know, ourrole-based access control.
What does that look like?
(17:16):
You know, who are we givingaccess to?
Who are we not giving?
Speaker 1 (17:18):
access to.
Speaker 3 (17:19):
Right.
Speaker 1 (17:20):
So that's what that's kind of.
What that marketing manager upthere was talking about is
they're allowed to use ChatsBTfor a lot of their stuff, but
they are restrictive on whatthey can tell it.
Oh, yeah, so that's why theyneed you guys to kind of come in
and be able to make sure theyare protected.
Speaker 3 (17:42):
Yes, yes, yes, you could be completely blocked out.
And also another thingcompanies do is they might just
choose to use pre-trained modelsto fine-tune what they already
have inside.
So you use your regular opensource models, you train on your
data, but then it's not talkingto anything outside, but also
(18:02):
it's restricted to what's inside.
You get what I mean, and so allthe questions you ask are
focused on what's inside and notnecessarily what's outside.
Speaker 1 (18:13):
So you got a PhD in artificial intelligence?
Yes, sir.
So how long did that take?
Seven years.
So you knew about AI sevenyears ago, Because I never heard
of it until three years agomaybe so I got into the
information systems program atUMBC Before that.
Speaker 3 (18:34):
While I was doing my master's, I was working on
protein sequences, and so at thetime, I was using AI neural
networks to be specific as amatter of fact it wasn't called
AI.
Speaker 1 (18:46):
You used AI to get your PhD.
Got it, got it.
I get one of those.
I get one of those too.
I'll get it today.
Let's do it.
I'm on chat, gpt right now.
Speaker 3 (18:58):
Yeah, I mean chat.
Gpt was taking tests, you knowPhD legal tests and this thing
was doing like was acing thetest.
So you can imagine.
Speaker 1 (19:12):
You don't need to go for a PhD anymore now.
Speaker 3 (19:17):
All 10 viewers are going to get this confession
from you.
Yeah, but like I got into mymaster's program and we were
already, it wasn't even calledAI.
It was called patternrecognition and pattern mining,
data mining.
Speaker 1 (19:29):
Really yes.
Speaker 3 (19:31):
Yes, okay, so the name has actually largely
evolved from data mining.
We started calling it machinelearning and then we started
calling deep learning and all ofa sudden you had this whole
generative AI thing come out.
And then everybody startscalling it AI, which is
perfectly fine.
I can spell that.
But if you read some of thoseold books, you know artificial
(19:53):
intelligence would be comparedto like expert systems and
things of that nature.
Speaker 2 (19:59):
Now how did you get involved?
So we, you know it wasinteresting, because it's
probably been about a littleover two years ago that, um, I
met jan it was actually throughour church and, um, you know, we
, we, you know he knew that Iwas in business and we started
talking and it was like, youknow, I wonder if there's
something here.
And at the time he was tryingto get me as a client to you
(20:22):
know, and it was interestingthis is just a barter.
Speaker 3 (20:24):
Yeah, it was interesting.
Speaker 2 (20:24):
This is just a barter , yeah.
Speaker 3 (20:26):
It was interesting Six months back and forth.
Speaker 2 (20:28):
Yeah, we were having meetings and you know, I'll
admit, you know, my struggle atthe time was because he was
talking, you know, and, andYano's world is like this big,
and I was living in a world thatwas this big and I couldn't get
there, I couldn't get what hewas talking about, cause I'm
like I don't even know what'spossible.
And you're, you know, when Iask you, you know, what can we
do?
His answer was anything.
Speaker 3 (20:47):
And I'm like I need more specifics.
Speaker 1 (20:49):
I'll take two yeah.
Speaker 2 (20:51):
You know, and so my challenges at the time were in
the realm of marketing and I hadall these manual processes that
I was going through every day.
You know, ad managing, adwordcampaigns, like doing all of
this stuff.
That was very labor intensive,you know from or you know it was
.
Speaker 1 (21:07):
I wasn't actually, it wasn't hard labor, but let's
get it straight, man, cause I dosome of that.
It wasn't that hard, but it'scommitment though.
Speaker 2 (21:15):
Well, I was starting to.
It started to get me thinkingabout, you know, maybe maybe I'm
not doing things in the mostefficient way, maybe there's a
better way.
Long story short, you know, Iended up leaving my job last
last summer and then we startedtalking about like hey, maybe we
should do something, like maybewe should actually start, start
start something, and so that'skind of how I got involved.
(21:37):
But you know, since then, youknow and this is why I say this
because when I was working Iknew that this, that some of
this stuff was coming, but I didnot have time, I didn't have
time to learn it.
And so since I left, I'vestarted to understand a lot more
, because I have a lot more timeto really dig in and kind of go
(21:58):
down rabbit holes and figureout, like, okay, what is he
talking about?
Speaker 1 (22:02):
Is he on the phone at night with you for an hour and
a half?
Speaker 2 (22:04):
Yeah, so after you know five months or six months
now, I'm just now starting tolike get the gravity of what
we're dealing with here, andsome of this stuff is happening
faster than anyone realizes,like I think most businesses
would be shocked to understandhow fast this is.
It's like a freight traincoming at them and they don't
see it coming.
And, in my mind, what thisreally means is the businesses
(22:27):
that get it, the businesses thatembrace this, are going to have
such a competitive edge overthe businesses that don't.
It's not even going to be fair.
I mean, they're going to.
A lot of businesses are goingto go under, not because AI is
replacing their jobs, butbecause they don't embrace it,
because what it empowers you todo is to do things better,
faster and cheaper, and yourcompetitors are going to do it,
(22:49):
and so it's going to be.
It's going to depend on howwilling you are.
Are you to embrace this newtechnology and and to accept
that it's part of it's part ofthe way forward?
Speaker 1 (22:59):
before.
Before um ai, I had no idealike what the dental industry is
like, what the car automotiveindustry is like or the
automotive collision repair andall this stuff.
So I could create content.
I could still do videos andstuff, but I couldn't post shit
on their social media pagesbecause I didn't know anything
about it.
But then AI comes along.
(23:20):
I start sitting there andhaving conversations with
ChatGPT and then it gets to knowme and how I talk.
Ai comes along, I start sittingthere and having conversations
with chat GPT and then it getsto know me and how I talk so it
can understand me, and then Ican start developing content
based on that Brand packages,everything On AI.
Now it's a full-scale marketingbusiness, right, and it's just
(23:41):
me and a contractor.
Speaker 2 (23:45):
I think the next level is really in the realm of
automation because, you've gotthe tools, now you're able to do
stuff better and maybe faster.
But where automation comes inis it almost takes a lot of what
you're doing out of the loopand so now it does all the stuff
that you were doing and now youjust kind of oversee it.
Speaker 1 (24:00):
Yeah, so we spoke off camera about this before a
couple, last week or whatnot.
But I want to talk on cameraabout this because I kind of
want to explain to these peoplehow this would be very efficient
.
Just in my world, there theother day, I had scripts that I
(24:22):
got off of chat gbt 30 scriptsfor 30 videos, 20 second long,
and I had them on a teleprompter.
So I went up there and I mikedthem, lights, everything, filmed
it, all of that.
None of that would reallychange and then so then I had my
a roll, which was all thatvoice, and then I got b roll
(24:42):
footage of the golf carts anddrone footage and everything
else.
So my current process is tocome back and put it on my
server, my little nasa home, andthen of course, it's already
set up to where it backs up togoogle, a google file.
But then I still got to pullthose videos and that's where I
sat there for four or five hoursat night and edited 90 videos.
(25:06):
I had three um duplicated,three people, but um, that still
took.
That took a lot of time,because then I have to a roll b
roll call, a grade adjustment,layer, captions, then I have to
export, do it in and out inadobe Premiere and export each
one to that file.
(25:26):
Okay, so then I got my videos.
But now I still have to go tochat jbt and say create a
caption based on this video, andI could slide my video in and
they can analyze it, or I couldjust say what that video is
about.
You know the title.
Basically I'll just take thattitle.
Anytime it's a short clip,it'll get what it needs create
(25:47):
my caption.
And then I have to open up asoftware like Metricool to
create my post and schedule itor post it.
But you could make somethingthat pulls from the Google file.
Yes, literally.
So as soon as it backs up tothe Google file, I'm hands off.
Speaker 3 (26:06):
Yes, and you could tag your videos.
You could make the AI analyzethe video and tag the videos for
you based on certain words orcertain voice commands that you
put in there.
You can see A-roll, shot one,shot two, and it would recognize
(26:33):
that and use that to actuallyeven build the entire sequence
for you.
Now you could also take AI anduse it to edit the video and say
, hey, take out second 0.001 to0.10.
So the idea is you need to havea workflow for what it is that
(26:53):
you're doing and for each ofthose steps, there are systems
in place that we built thatwould also help you to do it.
Speaker 1 (27:02):
So you could take it from.
As soon as it backs up togoogle, pulls it, edits it, it
creates the copy for the socialmedia post and then it actually
schedules the post as a draft.
Yes, to where?
Later on in the day?
Absolutely, I'm gonna get moresleep yes, take my credit card.
Speaker 2 (27:18):
Now, let's do this let's do it so we had a, we had
a customer that, uh, it was acouple of weeks ago that they
approached us and they said wewant, we want to build a um, a
nano learning platform which isjust small videos, very like 30
second minute long videos, thatthat their customers would pay
for subscription and they would,they would be able to consume
(27:41):
kind of like different contentvideos over like leadership and
motivation and all thesedifferent topics.
And they said so we want youguys to create the platform.
We also want you to create allof the content.
So develop all the content andalso create the videos.
So basically they had nothingat the time.
They said can you guys developall of this for us?
(28:02):
We created 900 plus videos inthree days.
three days and developed the app.
The entire thing was done inthree days that's crazy.
Yeah, yes, that's the power ofautomation, and it wasn't just
like, hey, there's a tool thatyou can go out there and do this
with, because there wasn't.
Speaker 1 (28:18):
We had to hook like using apis, we had to hook like
10 different tools together soapis are just like little chain
links that you can just cliptogether in all the different
ways.
Speaker 3 (28:28):
You can imagine to have that and then have an
AI-powered system in front of it.
Speaker 1 (28:34):
Mm-hmm.
Speaker 3 (28:35):
Game-changing.
Speaker 1 (28:36):
That's interesting.
So is there anything, as wekind of wrap this up, that you
can tell, you can kind of passalong to the everyday consumer,
because I feel like every daysomebody should be playing with
some type of ai tool just to getfamiliar with capabilities?
(28:57):
Yes, you know, and we've seenopen ai pretty much dominate.
Um, google can't keep up.
And here's a.
Here's something I like to askpeople.
Do you think, as I start to usechat gpt more and more and more
, I don't use google foranything as a search engine?
(29:18):
I use chat gpt as a searchengine and uh, do you think this
is just off the cuff, casualthat one day open ai will take
out google, or because I knowthey keep trying to catch up
with jim and I.
Yeah, but jim and I keepsputting out some awkward um,
(29:38):
giving you about some awkwardanswers and stuff, so so I have
a very interesting take on this.
Speaker 3 (29:45):
I think Google is going to continue to stay ahead
of the curve here.
Yes, because Google.
It was a paper that was writtenout of Google.
Attention is all you need.
That's how GPT got created.
So a lot of the things that wesee on GPT the foundational
(30:05):
knowledge was learned fromGoogle.
Now I don't know what Google'ssearch engine optimization or
how they run their ads wouldlook like in the future, but
that's not to say they wouldn'tfind new opportunities.
Speaker 1 (30:19):
They got to step it up because their Gemini.
They'll run their Gemini for awhile and then they'll take it
back down because they can't getit to perform like chat gpt
does.
Yeah, but here lately I'venoticed they've kept it up, so I
don't know if it's doing betterchat.
Speaker 3 (30:34):
Gpt2 does have its own challenges, though, because,
um, like, there's no ai toolthat really has that reasoning
capability all figured out.
Because it's all text, so it'snatural language processing,
right.
So it's all text.
It learns by sentencecompletion, right?
So you know, two plus two goesinto the database you know, and
(30:58):
then it tries to see, okay, whatmight be the answer to that.
Speaker 2 (31:00):
The other thing is OpenAI.
They don't actually own any oftheir own data.
Like google has youtube, gmail,all of the you know, google has
their own set of data that theycan utilize.
Um versus chat, gbt is usingother people's data oh yeah, it
scrapes the entire internet.
Speaker 1 (31:19):
Right, that's it.
I mean right.
That's why I use it as a searchengine instead of google,
because this thing's gonnascrape the entire internet now.
Will it eventually catch up towhere there'll be copyright
issues with, uh, with like chat,gbt and other tools?
I think?
Speaker 3 (31:39):
the uh.
Data is now gold, like data isthe new gold.
Literally, data is a newcurrency.
So I think every company shouldkeep their data really
proprietary to them because thatinformation would be so gold in
in the new age of ai.
(32:00):
And so yes, but I think thatcopyright issues is going to be
a problem because now, withaccess to books and access, to
all that information.
You can just feed that into anLLM.
There's a lot of issues there.
I think there would need to bemore laws around how people's
(32:25):
books get used.
There will be some ethicalstuff that would come out of
that.
Speaker 1 (32:32):
I wonder, because there's a few podcasts that I
listen to that talk about someof this and you know they talk
about will Google keep up withopen AI, that kind of stuff?
They talk about the lawsuitsAll of them get sued all the
time there was deep seek orsomething like that.
That's an interesting one.
(32:52):
They've been brought up before.
Speaker 3 (32:54):
Yeah, yeah, an interesting one.
Speaker 1 (32:57):
I don't know if you want to share.
Not, I don't get a whole lot ofviewers, so don't worry about
that.
Well, so, so don't worry aboutthat.
Speaker 3 (33:03):
Well, so DeepSeq and there's been a lot of concern
around this DeepSeq and aroundwhere it's trained, how it is
trained, what data they're usingfor training and things of that
nature.
But it goes back to what I wassaying earlier is, if you're
using DeepSeq within yourcontained environment, then run
(33:27):
as much security tests around itbefore you use it.
Otherwise, ensure that,whatever platform you're using,
you're not feeding it privateinformation or data that is
sensitive, because all of thatwhether they say they train it
or not, we don't know right.
We still have to be careful, andso it's really critical that
(33:50):
that's done.
And running these models arenot really hard to do right now.
In a day or two we could takecare of that kind of thing for
any system.
Speaker 1 (34:00):
So I think we're getting it to where I'm going to
understand all of this.
So I think we're getting it towhere I'm going to understand
all of this.
So they're basically like.
Chatgpt is just like a big boxof millions of APIs all talking
to each other.
Speaker 3 (34:19):
No, GIPT is using a bunch of people's data training
it, using AI to generate thismodel that we all go in and ask
it questions how?
Speaker 1 (34:31):
do they get that data , though?
What was that?
That's going to be my ringtonefor you from now on, oh man.
So they script the internet.
Speaker 3 (34:43):
So they script the internet, you know.
Speaker 1 (34:47):
Are they using APIs to script the internet, though?
Speaker 3 (34:49):
No, not necessarily.
They might be getting some datafrom internal systems, you know
, via API.
So ChatGPT might have reachedout to this company that does.
There's this company that haslots of images.
They might have reached out tothem and say hey, would you give
us access to your API so thatwe can get all your images, so
that we can use that to trainour model to do.
Speaker 2 (35:10):
X and Z.
They collect data for training.
Speaker 3 (35:13):
Yeah, they collect data through APIs for training.
Speaker 1 (35:17):
But what you can do is come in in a more secured
fashion and create the AIs andcode it to get more secured
product end product.
Speaker 3 (35:29):
So what we could do is isolate that model, put it
within your environment, use anopen source version and not
necessarily an open AI system.
Open AI has its own so it'sjust totally up to any customer
once you use that.
Speaker 1 (35:50):
Open AI has good application, but not in every
single industry, Like healthcarewhen we were talking about
healthcare and everything elsethat is correct.
That's where that would reallycome into play is you can come
in and build a more securesystem that they can still use.
Speaker 3 (36:05):
Also, some of these models have their limitation in
terms of their responses.
So all of them run between, Ithink, 4,000-something tokens.
So even if you put in inputs interms of questions, there's a
limit to which the number ofanswers they can produce, which
definitely also limits the waythat the responses come out, and
(36:28):
things of that nature, becausesome of it might appear somewhat
summarized and things of thatnature.
There are ways around it, youknow, and chunking and things of
that nature, but I thinkultimately the core of it is,
you know, being able to isolateit, bring it into your
environment.
The core of it is being able toisolate it, bring it into your
environment, use your data tofine-tune what already exists
(36:51):
and supercharge your operations.
That's crazy.
Speaker 1 (36:55):
Yeah, I'm ready to start seeing it.
Let's do it, let's do it.
I don't even know where westart, See Caffeine.
Speaker 3 (37:09):
Well, see, I always say, like, try to look at your
process and see which of yourprocesses brings you a lot of
pain, and then ask how can AIhelp to do that?
And the best place for peopleto start is to use ChartGPT to
be honest with you.
Just to get it, just to get itjust to get familiar yes, and
don't think of gpt as just aquestion and answer platform.
(37:32):
Think of gpt as your companion,who is there to give you some
guidance, wow yeah well, theother night.
Speaker 1 (37:41):
So, randolph county, they wanted that marketing
proposal, yeah, so I kind ofknew the nuts and bolts of what
I wanted to pitch, but I neededhelp.
So for it was about two and ahalf hours I sat there and
talked to ChatGBT, put it on,talk to text.
That way I'm not just gettingarthritis.
But so I had that longdrawn-out conversation,
(38:04):
conversation with chat gbt, andit got me thinking about certain
points and then I would be like, okay, so what if we change
this a little bit?
Or you know, um, how can weconnect the employers more?
How and it literally helped mewrite that entire proposal um,
I've done a lot of that whereclients have wanted proposals.
(38:28):
Well, they don't want them in amonth, they want them as soon as
they text you, as soon as theycall you.
They want that proposal forthose videos or marketing.
I literally plug that in andget a proposal.
And then I will talk to ChatGVTand sit there and say, okay,
but let's change to these termsnet 30, net 90, something like
(38:48):
that, and put a cover letter toit.
And there's my proposal formarketing services, send it off,
they like it.
I literally just take thatdocument and put it back in
ChatGPT and I say now, make thisa purchase agreement, gives me
a purchase agreement.
Throw that in Adobe and get itsigned.
Speaker 3 (39:06):
So, Kevin, imagine if you now had the opportunity to
basically have a form right infront of your website.
The customer puts what theywant and the proposal gets
created.
Speaker 1 (39:21):
That would be crazy.
That would be crazy.
Speaker 3 (39:24):
Yeah, and so you have the proposal right there.
You just review the proposal,like Jason would do, check for
all the issues that exist in thecontext of the proposal.
Update it, feed it right back,generate your PDF, send it right
off While I'm driving down theroad on my phone.
Speaker 2 (39:46):
You know what I'm saying Automation, automation,
right.
So you brought up aninteresting point.
I wanted to touch on that for asecond, because, you know, a
lot of people are concernedabout ai replacing jobs, and
while, while that may be true tosome degree, I think it's going
to be more of a shift whereit's it's it's going to change
the nature of work, and so youactually brought up a really
good example, because what Ithink there's the most power is
(40:09):
really going to come in the formof AI combined with subject
matter experts.
So you already knew what youwanted, you already knew what
you were doing, you knew how todo it, you understood, you know
almost you know you'refunctioning as like a CEO or a
manager.
You know the work that needs tobe done.
Now you just need to get itdone, and that's where AI comes
in and helps, but it still needsyou to guide it, because if
(40:30):
you've got somebody who doesn'thave a clue what they're doing,
then they're just going to haveto hope that AI got it right,
and it doesn't always.
That's where it gets its badreputation.
Yeah, it doesn't always.
It makes mistakes, and so itneeds it needs a human in the
loop, human in the loop to beable to oversee it, to make sure
it's doing the correct things.
So I think you know, you knowthe.
(40:50):
The buzz is, you know, oh, aiis going to replace people.
It's actually not.
It's going to give I think it'sactually going to give people
the more important jobs, whichis really acting almost as like
an orchestra, like anorchestrator or a conductor,
somebody who's looking at thewhole picture and saying, okay,
you know, making sure all thejobs are getting done correct,
they're, they're providingoversight to the AI, which AI is
(41:12):
just doing you know the, thegrunt work a little bit, where
the stuff that we're really noteven that good at in the first
place.
But I think there is value inhaving still people with very
specific skill sets.
I mean you can look at likedevelopers.
You know and I'll just usemyself as an example I'm not a
developer, I'm a marketing guy.
So I don't know how to writecode.
Speaker 3 (41:32):
I mean, I know a little bit of.
Speaker 2 (41:33):
HTML.
But other than that, you know,I've been using AI to help me
build things just because I'vebeen practicing with it and I
get stuck a lot.
It can't, it can't quite do it,and so.
But where the real value is isactually having training a
developer who already knows howto do all of it to use the AI to
(41:54):
help him do his job better andfaster, because he knows when it
gets stuck.
Oh, you're doing this wrong, orI see, I see what the issue is
versus a guy like me, like Ihave to, I have to go to Yano's
like help, I'm stuck.
I don't know enough about codeand development work to be able
to figure out what the problemis.
I'm just practicing, but it mayget to that point at some point
.
But right now I think there isa.
(42:15):
You know I always tell people alot of times it's like, you
know, ai isn't quite as scary asyou think it is.
It's actually going to assistus and help us be better and do
more important work.
Speaker 1 (42:29):
Like I said, it's the new electricity, it's the new
cell phone.
Everyone was scared of both ofthose when they came out.
So I think grasping it as soonas you can, learning it as soon
as you can, that's the biggesttakeaway, absolutely.
(42:49):
Well, I appreciate you guyshopping on here with me.
Thanks, kevin, this is an honor.
Speaker 3 (42:55):
Thanks, especially, you've done a TED Talk when did
you do that TEDx Many years withthat, like maybe four or five
years ago.
Speaker 2 (43:07):
So you just keep going down because now you're on
this.
Speaker 1 (43:08):
You started a TEDx Now.
That'd be cool.
Speaker 3 (43:10):
Yeah, I'm watching.
It's good it's.
I'm really excited to be here.
Thanks for having us.
Speaker 1 (43:15):
Oh, I appreciate you guys hanging out with me, cause,
um, I'm just kind of like amediocre marketer and whatnot,
but or whatnot.
But I'm real fascinated withwhat you're doing.
I look forward to working withyou guys and growing together
and scaling together.
Yeah, I look forward to thesame as well.
Speaker 3 (43:32):
No-transcript.
Ready to do this?
Yeah, let's do it we don'tsmoke anything in here like Joe
Rogan, that's good.
Speaker 3 (00:04):
Life Unscripted with Kevin Shipp.
Speaker 1 (00:09):
Welcome to Life.
Unscripted.
Thanks, kevin.
So this is like a podcast thatsometimes goes off the rails.
It's not scripted, it's nothing, I don't plan anything, but I'm
really fortunate to have youguys here today.
Iannu Odibodi, you got it right.
Yes.
Speaker 2 (00:26):
Jason.
Speaker 1 (00:27):
Jason Wright From Walker.
Yeah, and tell me a little bitabout Walker.
Speaker 3 (00:35):
So we saw a gap in the market and we realized that,
as AI begins to become a thing,that as AI begins to become a
thing, that there will bedisplacement of jobs for many
people, and that gave us a lotof concern, and that's one of
(00:57):
the reasons why we chose tocreate a company that would be
really focused on equippingtalents to fill this gap and
then positioning them to getopportunities with companies all
around the country.
So our passion is driven by that.
But the core of what we're ableto accomplish is, you know,
think about building AI agents,vertical AI agents, think about
(01:19):
automation and where automationis going.
I mean, you own a marketingcompany, so you can imagine how
automation can just helpaccelerate the work that you
currently do.
So these are the issues that wesaw and we realized hey, look,
if we can train multiple or morepeople to be able to address
this problem, then the worldwould be a better place and we'd
(01:42):
be able to see acceleration ingrowth in business growth from a
small business perspective, andeven from a mid-sized to
large-sized companies also canbenefit from it.
Speaker 1 (01:53):
So backing up a little bit too.
First of all, a lot of peoplethey're still scared of AI and I
think there's a lot of unknownbecause they're not familiar
with it.
Yes, ai coming out is verycomparable to when electricity
came out right, and it's soprofound and it's made such a
(02:17):
large impact at scale to thisdate that people are still
scared because there's a lot ofunknown.
So that's kind of why I wantedyou guys to hang out with me
right this episode.
Um, because I'm down here anduh, but I'm like the everyday
consumer.
Like you said, I have amarketing business and I have
clients in multiple industries.
(02:39):
So I'm trying to come up hereand knock on your door in the
cloud and say, hey, how can Iuse what you're doing in my
everyday processes?
So the way you've explainedwhat WACA does is the way you
(03:00):
explained it to me is you justtake a lot of the AI tools and
then you develop APIs and pullthem all together.
So my first question is whatthe hell is API?
Speaker 3 (03:11):
So great, great question.
So API is sort of an interfacebetween whatsoever you're trying
to do and a database system.
What's a database system?
A database system is a systemthat helps store records.
So let's say you wanted toenter information about your
(03:33):
company, your customers andthings of that nature.
You put it in a form.
It sends it to a SQL database.
Now we have like Google Sheets,where people can actually
visually see some of the thingsthat they're updating via Google
Forms right, but a databasestores all this information.
Now, for us to have access tothat information, then we have
(03:55):
to go through an API system.
If we would have to accesssomeone else's you know data or
someone else's information.
So companies that are providingthese services right now,
typically you go in and then youcan use whatever services they
provide you.
But if you wanted to develop itat scale or if you wanted to
(04:18):
solve problems at scale, youwould have to drag and drop all
of these things within thecontext of the software.
So now you don't have to dragand drop all of these things you
know within the context of thesoftware.
So now you don't have to dothat, since you can, you know,
interact with the exact sameservice via an API.
So you make an API request andessentially, the request then
produces a response that helpsyou to solve that exact problem.
Speaker 2 (04:42):
I kind of think of it as like a backdoor into, kind
of behind, the software, so thatway you don't have to go and
type it in from the front of it,you can all access it from the
backend through code.
And so a lot of it, a lot of itis it's it's speeding up the
process, it's you know cause.
Right now, say you wanted touse a piece of software like,
for instance, even chat GPT.
You got to go into a, you gotto go to chatgbtcom, you got to
(05:04):
log in, you got to type in whatyou want.
An API would allow you toaccess it from the back end of
it.
So you didn't actually have togo to the website.
So you have another tool,another piece of software.
You can engage with chatgbtwithout ever going to the
website through that one tool.
And so imagine if you couldhook into all these other
different pieces of softwarethrough one tool.
That's what APIs unlock thepower of.
Speaker 1 (05:26):
So what are some current tools?
You mentioned ChatGVT, and thenI was showing you some of the
video applications I've used.
I think Reveal was one of them,where we insert a static image
and it turns it into a littlevideo.
So I still are using all thesetools.
So can you give me someexamples of what tools currently
(05:49):
exist that you could pulltogether by doing the code and
creating the APIs?
Speaker 3 (05:57):
So I mean one is GPT.
It's an easy one to explain and, by the way, API means
application programminginterface, just for some people.
Speaker 1 (06:07):
I'm gonna need that and I'm gonna put it on the
screen when I edit this.
Speaker 3 (06:10):
Okay okay, awesome.
So.
So, for example, gpt is oneright.
Typically, when you want to aska question, you put in your
information into a chat and thenyou you add your data and you
say, hey, tell me about numberof sales that I had this month
from the data that I imputedinto you and it just does it
(06:34):
right.
But now, rather than having todo that, you could actually
interface through APIs where youactually write a code or write
code in Python, and then it hitsthat API and would ask multiple
data sets, the same exactquestion and give you responses
that could be actually visuallydisplayed and things of that
nature.
(06:54):
So typically you would alwaysstill need someone with some
type of coding experience tointerface with it.
But what it unlocks for a smallbusiness is it unlocks
opportunities to do things atscale and to do things pretty
quickly.
So the things that you used tohave to do and would take you
five hours to do, now you can doit in five minutes because you
have access to one GPT to do theprompting, the data, but also
(07:20):
the API that requests thatinformation directly from GPT,
so that you can get the responsewithout having to interface
with GPT all the time.
Speaker 1 (07:28):
Very cool and I know we talked the other night about,
you know, where does AI comeinto place with human operations
, and you said it kind of atscale.
So that's where this seems tobe really valuable is being able
to create, yes, 900 videos intwo days, versus the human, yes,
(07:52):
and there is a lot of.
You know, the people want thathuman talking to them, oh, yes,
but sometimes we can't make thatin that time, so like,
especially when it comes tovideo editing.
Well, you know, I was tellingyou all my little processes the
other night that I was doing tocreate 90 short videos and that
(08:14):
took me like four or five hours.
Yes, and um, ai could have, youknow, I could have used, like
opus was as a tool, but I stillhave to go to opus and still
have to feed it that video,which is, I'm not complaining
because, by the way, has its apisystem too.
Yeah, yeah, so that, becausethat's light years ahead of you
(08:35):
know what we used to have to doto make all of these clips.
Speaker 2 (08:37):
That's correct, that's correct.
Speaker 1 (08:39):
Um so, but what you're doing, and by connecting
all the dots, you're speeding upfaster and faster and faster
and making that a more efficientprocess.
Speaker 2 (08:50):
Yeah, and I just wanted to throw in a comment.
There you talk about what arethe different tools.
There's probably what?
Over 1,000 different tools outthere available right now, and
my own experience is I came froma, a manufacturing company, and
I ran our marketing department,and you know, one of the
challenges that I had was Ididn't have time to learn all
these new tools.
(09:11):
I was too busy doing my job, andso every day I would come in
and like, ok, how do I get thisstuff done using the knowledge
that I have?
And I was using stuff like chat,gpt, but aside from that, I
wasn't really using a whole lot,but aside from that, I wasn't
really using a whole lot.
And so one of the I think thatthe things that's coming, that
it's already happening right now, is really more of an
automation to where you come inand I no longer have to go do
(09:32):
the things.
We actually use AI to do thethings and I just kind of
conduct the orchestra so we cancreate like automated workflows
that actually, you know, insteadof me going in and creating the
social media posts, I actuallyhave an AI agent that oversees
that, and so how does it knowwhat to do.
Well, I train it, I give it allof my data, I dump everything
(09:52):
in there about my brand andabout what I'm doing as a
company and say here's yourinstructions.
And then I have another AIagent that acts as more of a it
summarizes and it bringseverything into a usable format
and then it passes along.
So we create workflows thatactually step-by-step.
You've got different AI, soimagine chat GBT as just one of
(10:13):
your workers and it has aspecific task, and then you have
.
You can actually line thesethings up and create workflows
that get all of your work donefaster and then so really what
your job is at the end of theday is more of like a manager or
an overseer that says, okay, isthe quality good or do I need
to?
And you can even train it to goback and edit its own work and
(10:34):
figure out what it you knowmight do better, and then at the
end of the day, you basicallysay, okay, what I want from you
is I want you to create allthese social media posts and add
them as drafts in my Facebookaccount.
And then I can go in andmanually approve them to make
sure that they're on point.
So that's really.
You know, a lot of people thinkof chat, gbt, as you know, ai,
but really it's it's kind of thetip of the iceberg, it's it's.
(10:57):
It goes so much deeper than that, especially when you talk about
automation and real worldapplications, because for me, ai
was always this mysterious boxand, like you're, trying to
understand what's in it.
But I think this year it'salready happening.
But this year we're going tosee more of it where it becomes
(11:22):
practical, where businesses cansay, oh, you mean, this could
save me 10 hours per week.
Yes, I'll take that.
Speaker 1 (11:31):
I'm starting to see it creep into I shouldn't say
creep when it comes to AI,because then people but I've
heard of grant writers startingto use it a little bit to apply
for grants and people arestarting to see, oh, this is
(11:51):
actually a wonderful tool whenused properly.
Oh, yes, um, I've seen it inhealth care.
So a client, uh, hospital upnorth, um, they went to a little
speech, little little deal, andwhere they were talking about
ai and health care andmanagement.
And I walked in on thisconversation the other day and I
had to go up there to do somestuff.
But I heard her talking aboutthat and just from talking to
(12:12):
you and you guys, I was like, ohyeah, that's so possible.
I said, hey, I work with thiscompany.
You know we meet every Mondayand they just explained that
whole process to me and I shewas like that was like that's
exactly everything.
I shouldn't have went to thatthing.
Speaker 3 (12:27):
Right, right, I mean right now you could use AI for
training.
You could access tons of datavia API from multiple databases.
Train AI to be able to explaineducational materials, company
materials, company documentation, fact sheets and things of that
(12:52):
nature where employees canactually start to gain real-time
access to how to do their jobsmore effectively.
So this is really powerful.
Speaker 1 (13:02):
With that educational aspect.
A lot of companies don't wanttheir information out there, so
can you explain how you protectthat?
Speaker 3 (13:09):
Yes, I like what you said earlier when you talked
about responsible use of AI.
Not so many people think aboutthis and how to responsibly use
it.
There's a lot of work aroundresponsible and ethical AI lot
(13:29):
of work around responsible andethical AI and lots of the
government put out a sort of apolicy, a NIST policy, around
how to responsibly use AI forall the different, I would say,
cycles of machine learningdevelopment, from data
extraction to data training todeveloping the models to
inference.
All of that might sound likecrazy for some people listening,
but we'll definitely get intosome of that conversation as we
(13:52):
move along, um, and discuss someof this and definitely open to
sharing more of that, you know,on your your podcast and getting
people up to speed with that,but around responsible use.
Now companies are moving towardsusing more open source models,
so the idea behind that is youcan actually bring in a model
(14:17):
like Olamer into yourenvironment and train on Olamer,
and Olamer is completelyblocked from the outside, so
restricted within yourorganization, used specifically
for your own data, has nointerference with the outside
world at all, and you're stillusing this information to help
(14:38):
your staff to do, to createtraining courses, to build
systems and things of thatnature, and so the open source
route is becoming a good optionfor companies that are thinking
about the security of their dataand being able to have that
helps you know a lot, so we'reworking together on a proposal
(15:02):
pretty soon for thoseeducational videos, so that
would be using open source andthen private network.
Speaker 1 (15:09):
Yes, that's correct and you've mentioned before,
basically you're putting a wallbetween that cloud and that
cloud.
Speaker 3 (15:16):
We need to do that, gotcha.
We need to do that to be ableto build responsibly.
Otherwise, you give your databack to the know, the world, to
train and right.
Speaker 1 (15:26):
That's.
That was the big thing.
That's what we kind of talkedabout up there.
Um that hospital was you know,um they, you know they were like
well, how can they not get ourdata, how can they not get our
hip?
Speaker 3 (15:39):
this is the reason why you need people that have
expertise in this specific area.
Speaker 1 (15:44):
So they could do that , so they could do that.
You know anybody?
Speaker 3 (15:49):
we do.
I mean, we, we have thatexpertise.
Um, we have that expertise, umwe've we've done a lot of work
in this.
In this space, I particularlyhave also done quite a number of
work um in this space.
I have a phd in artificialintelligence myself, so the
ability to know where thingsstart to get iffy is really
(16:10):
really critical Also outside ofjust the AI expertise.
I think it's also critical tohave all the stakeholders
involved in the conversation, sofrom the business executives to
the subject matter expertsinvolved in that conversation as
well, because policy doesn'tlie only in the hand of the AI
(16:33):
expert.
You have to sort of think of itas a more collective effort to
get there so the businessmanager might see some, you know
, sides that you wouldn't see.
And this is where people likeJason come in and start saying,
hey, how do we think about this?
How do we look at this?
Speaker 2 (16:54):
And you know.
Speaker 3 (16:55):
So that's my thought process around.
It is yes, have your AIinfrastructure in place, but
also have subject matter expertsthat can tell you hey, this is
what data governance looks likein this space.
This is what we want to expose.
This is what we don't want toexpose.
What does our RBAC?
Rbac meaning, you know, ourrole-based access control.
What does that look like?
(17:16):
You know, who are we givingaccess to?
Who are we not giving?
Speaker 1 (17:18):
access to.
Speaker 3 (17:19):
Right.
Speaker 1 (17:20):
So that's what that's kind of.
What that marketing manager upthere was talking about is
they're allowed to use ChatsBTfor a lot of their stuff, but
they are restrictive on whatthey can tell it.
Oh, yeah, so that's why theyneed you guys to kind of come in
and be able to make sure theyare protected.
Speaker 3 (17:42):
Yes, yes, yes, you could be completely blocked out.
And also another thingcompanies do is they might just
choose to use pre-trained modelsto fine-tune what they already
have inside.
So you use your regular opensource models, you train on your
data, but then it's not talkingto anything outside, but also
(18:02):
it's restricted to what's inside.
You get what I mean, and so allthe questions you ask are
focused on what's inside and notnecessarily what's outside.
Speaker 1 (18:13):
So you got a PhD in artificial intelligence?
Yes, sir.
So how long did that take?
Seven years.
So you knew about AI sevenyears ago, Because I never heard
of it until three years agomaybe so I got into the
information systems program atUMBC Before that.
Speaker 3 (18:34):
While I was doing my master's, I was working on
protein sequences, and so at thetime, I was using AI neural
networks to be specific as amatter of fact it wasn't called
AI.
Speaker 1 (18:46):
You used AI to get your PhD.
Got it, got it.
I get one of those.
I get one of those too.
I'll get it today.
Let's do it.
I'm on chat, gpt right now.
Speaker 3 (18:58):
Yeah, I mean chat.
Gpt was taking tests, you knowPhD legal tests and this thing
was doing like was acing thetest.
So you can imagine.
Speaker 1 (19:12):
You don't need to go for a PhD anymore now.
Speaker 3 (19:17):
All 10 viewers are going to get this confession
from you.
Yeah, but like I got into mymaster's program and we were
already, it wasn't even calledAI.
It was called patternrecognition and pattern mining,
data mining.
Speaker 1 (19:29):
Really yes.
Speaker 3 (19:31):
Yes, okay, so the name has actually largely
evolved from data mining.
We started calling it machinelearning and then we started
calling deep learning and all ofa sudden you had this whole
generative AI thing come out.
And then everybody startscalling it AI, which is
perfectly fine.
I can spell that.
But if you read some of thoseold books, you know artificial
(19:53):
intelligence would be comparedto like expert systems and
things of that nature.
Speaker 2 (19:59):
Now how did you get involved?
So we, you know it wasinteresting, because it's
probably been about a littleover two years ago that, um, I
met jan it was actually throughour church and, um, you know, we
, we, you know he knew that Iwas in business and we started
talking and it was like, youknow, I wonder if there's
something here.
And at the time he was tryingto get me as a client to you
(20:22):
know, and it was interestingthis is just a barter.
Speaker 3 (20:24):
Yeah, it was interesting.
Speaker 2 (20:24):
This is just a barter , yeah.
Speaker 3 (20:26):
It was interesting Six months back and forth.
Speaker 2 (20:28):
Yeah, we were having meetings and you know, I'll
admit, you know, my struggle atthe time was because he was
talking, you know, and, andYano's world is like this big,
and I was living in a world thatwas this big and I couldn't get
there, I couldn't get what hewas talking about, cause I'm
like I don't even know what'spossible.
And you're, you know, when Iask you, you know, what can we
do?
His answer was anything.
Speaker 3 (20:47):
And I'm like I need more specifics.
Speaker 1 (20:49):
I'll take two yeah.
Speaker 2 (20:51):
You know, and so my challenges at the time were in
the realm of marketing and I hadall these manual processes that
I was going through every day.
You know, ad managing, adwordcampaigns, like doing all of
this stuff.
That was very labor intensive,you know from or you know it was
.
Speaker 1 (21:07):
I wasn't actually, it wasn't hard labor, but let's
get it straight, man, cause I dosome of that.
It wasn't that hard, but it'scommitment though.
Speaker 2 (21:15):
Well, I was starting to.
It started to get me thinkingabout, you know, maybe maybe I'm
not doing things in the mostefficient way, maybe there's a
better way.
Long story short, you know, Iended up leaving my job last
last summer and then we startedtalking about like hey, maybe we
should do something, like maybewe should actually start, start
start something, and so that'skind of how I got involved.
(21:37):
But you know, since then, youknow and this is why I say this
because when I was working Iknew that this, that some of
this stuff was coming, but I didnot have time, I didn't have
time to learn it.
And so since I left, I'vestarted to understand a lot more
, because I have a lot more timeto really dig in and kind of go
(21:58):
down rabbit holes and figureout, like, okay, what is he
talking about?
Speaker 1 (22:02):
Is he on the phone at night with you for an hour and
a half?
Speaker 2 (22:04):
Yeah, so after you know five months or six months
now, I'm just now starting tolike get the gravity of what
we're dealing with here, andsome of this stuff is happening
faster than anyone realizes,like I think most businesses
would be shocked to understandhow fast this is.
It's like a freight traincoming at them and they don't
see it coming.
And, in my mind, what thisreally means is the businesses
(22:27):
that get it, the businesses thatembrace this, are going to have
such a competitive edge overthe businesses that don't.
It's not even going to be fair.
I mean, they're going to.
A lot of businesses are goingto go under, not because AI is
replacing their jobs, butbecause they don't embrace it,
because what it empowers you todo is to do things better,
faster and cheaper, and yourcompetitors are going to do it,
(22:49):
and so it's going to be.
It's going to depend on howwilling you are.
Are you to embrace this newtechnology and and to accept
that it's part of it's part ofthe way forward?
Speaker 1 (22:59):
before.
Before um ai, I had no idealike what the dental industry is
like, what the car automotiveindustry is like or the
automotive collision repair andall this stuff.
So I could create content.
I could still do videos andstuff, but I couldn't post shit
on their social media pagesbecause I didn't know anything
about it.
But then AI comes along.
(23:20):
I start sitting there andhaving conversations with
ChatGPT and then it gets to knowme and how I talk.
Ai comes along, I start sittingthere and having conversations
with chat GPT and then it getsto know me and how I talk so it
can understand me, and then Ican start developing content
based on that Brand packages,everything On AI.
Now it's a full-scale marketingbusiness, right, and it's just
(23:41):
me and a contractor.
Speaker 2 (23:45):
I think the next level is really in the realm of
automation because, you've gotthe tools, now you're able to do
stuff better and maybe faster.
But where automation comes inis it almost takes a lot of what
you're doing out of the loopand so now it does all the stuff
that you were doing and now youjust kind of oversee it.
Speaker 1 (24:00):
Yeah, so we spoke off camera about this before a
couple, last week or whatnot.
But I want to talk on cameraabout this because I kind of
want to explain to these peoplehow this would be very efficient
.
Just in my world, there theother day, I had scripts that I
(24:22):
got off of chat gbt 30 scriptsfor 30 videos, 20 second long,
and I had them on a teleprompter.
So I went up there and I mikedthem, lights, everything, filmed
it, all of that.
None of that would reallychange and then so then I had my
a roll, which was all thatvoice, and then I got b roll
(24:42):
footage of the golf carts anddrone footage and everything
else.
So my current process is tocome back and put it on my
server, my little nasa home, andthen of course, it's already
set up to where it backs up togoogle, a google file.
But then I still got to pullthose videos and that's where I
sat there for four or five hoursat night and edited 90 videos.
(25:06):
I had three um duplicated,three people, but um, that still
took.
That took a lot of time,because then I have to a roll b
roll call, a grade adjustment,layer, captions, then I have to
export, do it in and out inadobe Premiere and export each
one to that file.
(25:26):
Okay, so then I got my videos.
But now I still have to go tochat jbt and say create a
caption based on this video, andI could slide my video in and
they can analyze it, or I couldjust say what that video is
about.
You know the title.
Basically I'll just take thattitle.
Anytime it's a short clip,it'll get what it needs create
(25:47):
my caption.
And then I have to open up asoftware like Metricool to
create my post and schedule itor post it.
But you could make somethingthat pulls from the Google file.
Yes, literally.
So as soon as it backs up tothe Google file, I'm hands off.
Speaker 3 (26:06):
Yes, and you could tag your videos.
You could make the AI analyzethe video and tag the videos for
you based on certain words orcertain voice commands that you
put in there.
You can see A-roll, shot one,shot two, and it would recognize
(26:33):
that and use that to actuallyeven build the entire sequence
for you.
Now you could also take AI anduse it to edit the video and say
, hey, take out second 0.001 to0.10.
So the idea is you need to havea workflow for what it is that
(26:53):
you're doing and for each ofthose steps, there are systems
in place that we built thatwould also help you to do it.
Speaker 1 (27:02):
So you could take it from.
As soon as it backs up togoogle, pulls it, edits it, it
creates the copy for the socialmedia post and then it actually
schedules the post as a draft.
Yes, to where?
Later on in the day?
Absolutely, I'm gonna get moresleep yes, take my credit card.
Speaker 2 (27:18):
Now, let's do this let's do it so we had a, we had
a customer that, uh, it was acouple of weeks ago that they
approached us and they said wewant, we want to build a um, a
nano learning platform which isjust small videos, very like 30
second minute long videos, thatthat their customers would pay
for subscription and they would,they would be able to consume
(27:41):
kind of like different contentvideos over like leadership and
motivation and all thesedifferent topics.
And they said so we want youguys to create the platform.
We also want you to create allof the content.
So develop all the content andalso create the videos.
So basically they had nothingat the time.
They said can you guys developall of this for us?
(28:02):
We created 900 plus videos inthree days.
three days and developed the app.
The entire thing was done inthree days that's crazy.
Yeah, yes, that's the power ofautomation, and it wasn't just
like, hey, there's a tool thatyou can go out there and do this
with, because there wasn't.
Speaker 1 (28:18):
We had to hook like using apis, we had to hook like
10 different tools together soapis are just like little chain
links that you can just cliptogether in all the different
ways.
Speaker 3 (28:28):
You can imagine to have that and then have an
AI-powered system in front of it.
Speaker 1 (28:34):
Mm-hmm.
Speaker 3 (28:35):
Game-changing.
Speaker 1 (28:36):
That's interesting.
So is there anything, as wekind of wrap this up, that you
can tell, you can kind of passalong to the everyday consumer,
because I feel like every daysomebody should be playing with
some type of ai tool just to getfamiliar with capabilities?
(28:57):
Yes, you know, and we've seenopen ai pretty much dominate.
Um, google can't keep up.
And here's a.
Here's something I like to askpeople.
Do you think, as I start to usechat gpt more and more and more
, I don't use google foranything as a search engine?
(29:18):
I use chat gpt as a searchengine and uh, do you think this
is just off the cuff, casualthat one day open ai will take
out google, or because I knowthey keep trying to catch up
with jim and I.
Yeah, but jim and I keepsputting out some awkward um,
(29:38):
giving you about some awkwardanswers and stuff, so so I have
a very interesting take on this.
Speaker 3 (29:45):
I think Google is going to continue to stay ahead
of the curve here.
Yes, because Google.
It was a paper that was writtenout of Google.
Attention is all you need.
That's how GPT got created.
So a lot of the things that wesee on GPT the foundational
(30:05):
knowledge was learned fromGoogle.
Now I don't know what Google'ssearch engine optimization or
how they run their ads wouldlook like in the future, but
that's not to say they wouldn'tfind new opportunities.
Speaker 1 (30:19):
They got to step it up because their Gemini.
They'll run their Gemini for awhile and then they'll take it
back down because they can't getit to perform like chat gpt
does.
Yeah, but here lately I'venoticed they've kept it up, so I
don't know if it's doing betterchat.
Speaker 3 (30:34):
Gpt2 does have its own challenges, though, because,
um, like, there's no ai toolthat really has that reasoning
capability all figured out.
Because it's all text, so it'snatural language processing,
right.
So it's all text.
It learns by sentencecompletion, right?
So you know, two plus two goesinto the database you know, and
(30:58):
then it tries to see, okay, whatmight be the answer to that.
Speaker 2 (31:00):
The other thing is OpenAI.
They don't actually own any oftheir own data.
Like google has youtube, gmail,all of the you know, google has
their own set of data that theycan utilize.
Um versus chat, gbt is usingother people's data oh yeah, it
scrapes the entire internet.
Speaker 1 (31:19):
Right, that's it.
I mean right.
That's why I use it as a searchengine instead of google,
because this thing's gonnascrape the entire internet now.
Will it eventually catch up towhere there'll be copyright
issues with, uh, with like chat,gbt and other tools?
I think?
Speaker 3 (31:39):
the uh.
Data is now gold, like data isthe new gold.
Literally, data is a newcurrency.
So I think every company shouldkeep their data really
proprietary to them because thatinformation would be so gold in
in the new age of ai.
(32:00):
And so yes, but I think thatcopyright issues is going to be
a problem because now, withaccess to books and access, to
all that information.
You can just feed that into anLLM.
There's a lot of issues there.
I think there would need to bemore laws around how people's
(32:25):
books get used.
There will be some ethicalstuff that would come out of
that.
Speaker 1 (32:32):
I wonder, because there's a few podcasts that I
listen to that talk about someof this and you know they talk
about will Google keep up withopen AI, that kind of stuff?
They talk about the lawsuitsAll of them get sued all the
time there was deep seek orsomething like that.
That's an interesting one.
(32:52):
They've been brought up before.
Speaker 3 (32:54):
Yeah, yeah, an interesting one.
Speaker 1 (32:57):
I don't know if you want to share.
Not, I don't get a whole lot ofviewers, so don't worry about
that.
Well, so, so don't worry aboutthat.
Speaker 3 (33:03):
Well, so DeepSeq and there's been a lot of concern
around this DeepSeq and aroundwhere it's trained, how it is
trained, what data they're usingfor training and things of that
nature.
But it goes back to what I wassaying earlier is, if you're
using DeepSeq within yourcontained environment, then run
(33:27):
as much security tests around itbefore you use it.
Otherwise, ensure that,whatever platform you're using,
you're not feeding it privateinformation or data that is
sensitive, because all of thatwhether they say they train it
or not, we don't know right.
We still have to be careful, andso it's really critical that
(33:50):
that's done.
And running these models arenot really hard to do right now.
In a day or two we could takecare of that kind of thing for
any system.
Speaker 1 (34:00):
So I think we're getting it to where I'm going to
understand all of this.
So I think we're getting it towhere I'm going to understand
all of this.
So they're basically like.
Chatgpt is just like a big boxof millions of APIs all talking
to each other.
Speaker 3 (34:19):
No, GIPT is using a bunch of people's data training
it, using AI to generate thismodel that we all go in and ask
it questions how?
Speaker 1 (34:31):
do they get that data , though?
What was that?
That's going to be my ringtonefor you from now on, oh man.
So they script the internet.
Speaker 3 (34:43):
So they script the internet, you know.
Speaker 1 (34:47):
Are they using APIs to script the internet, though?
Speaker 3 (34:49):
No, not necessarily.
They might be getting some datafrom internal systems, you know
, via API.
So ChatGPT might have reachedout to this company that does.
There's this company that haslots of images.
They might have reached out tothem and say hey, would you give
us access to your API so thatwe can get all your images, so
that we can use that to trainour model to do.
Speaker 2 (35:10):
X and Z.
They collect data for training.
Speaker 3 (35:13):
Yeah, they collect data through APIs for training.
Speaker 1 (35:17):
But what you can do is come in in a more secured
fashion and create the AIs andcode it to get more secured
product end product.
Speaker 3 (35:29):
So what we could do is isolate that model, put it
within your environment, use anopen source version and not
necessarily an open AI system.
Open AI has its own so it'sjust totally up to any customer
once you use that.
Speaker 1 (35:50):
Open AI has good application, but not in every
single industry, Like healthcarewhen we were talking about
healthcare and everything elsethat is correct.
That's where that would reallycome into play is you can come
in and build a more securesystem that they can still use.
Speaker 3 (36:05):
Also, some of these models have their limitation in
terms of their responses.
So all of them run between, Ithink, 4,000-something tokens.
So even if you put in inputs interms of questions, there's a
limit to which the number ofanswers they can produce, which
definitely also limits the waythat the responses come out, and
(36:28):
things of that nature, becausesome of it might appear somewhat
summarized and things of thatnature.
There are ways around it, youknow, and chunking and things of
that nature, but I thinkultimately the core of it is,
you know, being able to isolateit, bring it into your
environment.
The core of it is being able toisolate it, bring it into your
environment, use your data tofine-tune what already exists
(36:51):
and supercharge your operations.
That's crazy.
Speaker 1 (36:55):
Yeah, I'm ready to start seeing it.
Let's do it, let's do it.
I don't even know where westart, See Caffeine.
Speaker 3 (37:09):
Well, see, I always say, like, try to look at your
process and see which of yourprocesses brings you a lot of
pain, and then ask how can AIhelp to do that?
And the best place for peopleto start is to use ChartGPT to
be honest with you.
Just to get it, just to get itjust to get familiar yes, and
don't think of gpt as just aquestion and answer platform.
(37:32):
Think of gpt as your companion,who is there to give you some
guidance, wow yeah well, theother night.
Speaker 1 (37:41):
So, randolph county, they wanted that marketing
proposal, yeah, so I kind ofknew the nuts and bolts of what
I wanted to pitch, but I neededhelp.
So for it was about two and ahalf hours I sat there and
talked to ChatGBT, put it on,talk to text.
That way I'm not just gettingarthritis.
But so I had that longdrawn-out conversation,
(38:04):
conversation with chat gbt, andit got me thinking about certain
points and then I would be like, okay, so what if we change
this a little bit?
Or you know, um, how can weconnect the employers more?
How and it literally helped mewrite that entire proposal um,
I've done a lot of that whereclients have wanted proposals.
(38:28):
Well, they don't want them in amonth, they want them as soon as
they text you, as soon as theycall you.
They want that proposal forthose videos or marketing.
I literally plug that in andget a proposal.
And then I will talk to ChatGVTand sit there and say, okay,
but let's change to these termsnet 30, net 90, something like
(38:48):
that, and put a cover letter toit.
And there's my proposal formarketing services, send it off,
they like it.
I literally just take thatdocument and put it back in
ChatGPT and I say now, make thisa purchase agreement, gives me
a purchase agreement.
Throw that in Adobe and get itsigned.
Speaker 3 (39:06):
So, Kevin, imagine if you now had the opportunity to
basically have a form right infront of your website.
The customer puts what theywant and the proposal gets
created.
Speaker 1 (39:21):
That would be crazy.
That would be crazy.
Speaker 3 (39:24):
Yeah, and so you have the proposal right there.
You just review the proposal,like Jason would do, check for
all the issues that exist in thecontext of the proposal.
Update it, feed it right back,generate your PDF, send it right
off While I'm driving down theroad on my phone.
Speaker 2 (39:46):
You know what I'm saying Automation, automation,
right.
So you brought up aninteresting point.
I wanted to touch on that for asecond, because, you know, a
lot of people are concernedabout ai replacing jobs, and
while, while that may be true tosome degree, I think it's going
to be more of a shift whereit's it's it's going to change
the nature of work, and so youactually brought up a really
good example, because what Ithink there's the most power is
(40:09):
really going to come in the formof AI combined with subject
matter experts.
So you already knew what youwanted, you already knew what
you were doing, you knew how todo it, you understood, you know
almost you know you'refunctioning as like a CEO or a
manager.
You know the work that needs tobe done.
Now you just need to get itdone, and that's where AI comes
in and helps, but it still needsyou to guide it, because if
(40:30):
you've got somebody who doesn'thave a clue what they're doing,
then they're just going to haveto hope that AI got it right,
and it doesn't always.
That's where it gets its badreputation.
Yeah, it doesn't always.
It makes mistakes, and so itneeds it needs a human in the
loop, human in the loop to beable to oversee it, to make sure
it's doing the correct things.
So I think you know, you knowthe.
(40:50):
The buzz is, you know, oh, aiis going to replace people.
It's actually not.
It's going to give I think it'sactually going to give people
the more important jobs, whichis really acting almost as like
an orchestra, like anorchestrator or a conductor,
somebody who's looking at thewhole picture and saying, okay,
you know, making sure all thejobs are getting done correct,
they're, they're providingoversight to the AI, which AI is
(41:12):
just doing you know the, thegrunt work a little bit, where
the stuff that we're really noteven that good at in the first
place.
But I think there is value inhaving still people with very
specific skill sets.
I mean you can look at likedevelopers.
You know and I'll just usemyself as an example I'm not a
developer, I'm a marketing guy.
So I don't know how to writecode.
Speaker 3 (41:32):
I mean, I know a little bit of.
Speaker 2 (41:33):
HTML.
But other than that, you know,I've been using AI to help me
build things just because I'vebeen practicing with it and I
get stuck a lot.
It can't, it can't quite do it,and so.
But where the real value is isactually having training a
developer who already knows howto do all of it to use the AI to
(41:54):
help him do his job better andfaster, because he knows when it
gets stuck.
Oh, you're doing this wrong, orI see, I see what the issue is
versus a guy like me, like Ihave to, I have to go to Yano's
like help, I'm stuck.
I don't know enough about codeand development work to be able
to figure out what the problemis.
I'm just practicing, but it mayget to that point at some point
.
But right now I think there isa.
(42:15):
You know I always tell people alot of times it's like, you
know, ai isn't quite as scary asyou think it is.
It's actually going to assistus and help us be better and do
more important work.
Speaker 1 (42:29):
Like I said, it's the new electricity, it's the new
cell phone.
Everyone was scared of both ofthose when they came out.
So I think grasping it as soonas you can, learning it as soon
as you can, that's the biggesttakeaway, absolutely.
(42:49):
Well, I appreciate you guyshopping on here with me.
Thanks, kevin, this is an honor.
Speaker 3 (42:55):
Thanks, especially, you've done a TED Talk when did
you do that TEDx Many years withthat, like maybe four or five
years ago.
Speaker 2 (43:07):
So you just keep going down because now you're on
this.
Speaker 1 (43:08):
You started a TEDx Now.
That'd be cool.
Speaker 3 (43:10):
Yeah, I'm watching.
It's good it's.
I'm really excited to be here.
Thanks for having us.
Speaker 1 (43:15):
Oh, I appreciate you guys hanging out with me, cause,
um, I'm just kind of like amediocre marketer and whatnot,
but or whatnot.
But I'm real fascinated withwhat you're doing.
I look forward to working withyou guys and growing together
and scaling together.
Yeah, I look forward to thesame as well.
Speaker 3 (43:32):
No-transcript.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
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.