How to Get Started with AI in Your Business (Practical Tips & Real Technologies) - Part 2 - The TechEd Podcast

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
TechEd Podcast Intro (00:09):
Announcer, this is the TechEd podcast,
where we feature leaders who areshaping, innovating and
disrupting technical educationand the workforce. These are the
stories of organizations leadingthe charge to change education,
to rethink the workforce and toembrace emerging technology.
You'll find us here everyTuesday on our mission to secure
the American Dream for the nextgeneration of STEM and workforce

(00:32):
talent.

Melissa Martin (00:35):
Welcome back to the TechEd Podcast. I'm Melissa
Martin, your producer, and thisweek, we're wrapping up our two
part series on how to getstarted with AI in your
business. Last week, in partone, Matt shared his insights
from his recent trip to China.
He talked about why small andmid sized businesses can't
afford to wait on AI adoption,and he introduced the first five

(00:56):
technologies every businessleader should know from his
manufacturer's guide to AI tech.
These include AI agents and MCPservers, digital twins, vector
databases, gpts and embeddedsmart technologies. If you
missed part one, be sure to goback and listen first. That sets
the stage for everything thatyou're going to hear in today's

(01:19):
episode. You can go back andcatch part one at TechEd
podcast.com/appliedAI. All right. In part two, Matt
will pick up where we left off,diving into the final seven
technologies that aretransforming how businesses
operate, automate and makedecisions, and most importantly,
he'll talk about what the futureof work looks like in this age

(01:40):
of AI, and what that means foryour job. Now, here's Matt.

Matt Kirchner (01:47):
All right, let's talk about intelligent data
prediction. You know, in myearly days of manufacturing, we
spent a lot of time looking overour shoulder, right? We would
get to the middle of get to themiddle of the month of November,
and we were looking back inOctober, and maybe we said, Wow,
that was a great month. I wonderwhat went well, or, man, we
really underperformed inOctober. What a disaster. And so
you're sitting there 15 minutesafter the end of the month,

(02:09):
wishing that you had it to doover again, right? Because
something went wrong, and youshould have probably played your
cards a little bit differentwhen it comes to financial
performance, is what I'mreferring to, by the way. So
we're looking at our financialstatements, you know, then we
got to a point where we werestarting to measure data in real
time. So in the world ofmanufacturing, you know, you're
pulling metrics in real time.
We, you know, we were doingIncome Statement projections

(02:31):
every single week. So we get tothe end of the second week of
the month, and we have a goodidea of what we think revenue
is. We know what our what ourspend is, and where we're you
know what our labor cost isgoing to be, materials cost,
maybe some of our energy costs,and so on. And so you could
start to predict out where yourbusiness is going to end up. And
if you didn't like what you wereseeing, you had time to fix it.
You had time to redirect thatwas a really good time in

(02:53):
business. And those of us thatfigured that out, and I was, I
was an early adopter, proud tosay, of that kind of a model
where we were, there were nosurprises in the manufacturing
businesses. I ran from about2005
until, you know, until I jumpedout in 20, 2015, we knew what
was happening as it washappening, and that was great.
Well, now we're at a point intime where that's not good

(03:16):
enough anymore, and we need tostart looking out into the
future. And if we don't likewhat we're seeing in the future,
making the change now so thatfuture never happens. Changing
in real time is no longer goodenough. As John Maxwell said
earlier in this podcast, or as Ireferenced earlier in this
podcast, quoting him, you know,it used to be you could just if
you're a leader, you were aleader because you could see
more, not anymore. Now you're aleader because you can see

(03:38):
before. We need to see ahead.
Let's talk about that. In one ofour businesses, we use a
CRM, a customer relationshipmanagement software called
Microsoft Dynamics, 365 we pullall kinds of data into that,
into that CRM, into thatplatform. And so one of the
things that we look at is everysingle order that, according to

(03:59):
our business development orsales teams is going to close in
the next 90 days, and we haveanother list of every order that
is going to close in the next180 days. And if we know that
what's going to close in thenext 90, what's going to close
in the next 180 have a prettygood idea of what. You know how
my business is going to perform,except that we all know that biz

(04:20):
dev. People aren't infallible,and their projections are not
always super precise. And we mayhave one person who is a
constant, you know, kind ofPollyanna person, who glasses
not just full, but like, triplefull, and everything's going to
close. Then we have somebodyelse that's like a total
sandbagger, and nothing is goingto close. And so we just take we

(04:40):
just accept the fact that peopleare going to project their
territories differently, but wetrack it over time. And so we
call it our sandbagger index.
And we know if you constantlyover project or under Project,
your expected performance. Weknow that, and so we just adjust
your projection. If you're anover projector, we'll back it
off based upon your previousperformance. And if you're a.
On our projector, we might juiceit up a little bit to get a

(05:02):
really good idea of what wethink in terms of orders, what's
going to close. We also knowwhat products those customers
are ordering. We know the leadtime, so we know when they're
going to ship. And then we alsoknow those customers payment
history, so we know when they'regoing to pay, so we know when
we're going to get our cash. Andwe really we put all that data
into and some other data sourcesas well, into Azure Data Factory

(05:23):
formatted the data for us useit. Used Microsoft. Azure
basically created an, what theycall an inference pipeline for
what that's worth, imputed newdata, output new predictions,
and here are the results, right?
So this is just pause on thisfor a minute and think back to
we used to look over ourshoulder on November 15 and say,
how we how did we perform? Andthen correct for the rest of
November if we didn't like whatwe saw for our October

(05:44):
performance. So we got to thepoint where we would correct in
real time, and that was great.
These are the results of what wedid through that project by
early 2024 so by the beginningof last year, that model was
predicting EBITDA, if you're notfamiliar with that, that term
earnings before interest, taxes,depreciation and amortization,
basically a financial measure ofthe cash flow of a business, not

(06:08):
quite the same as net income,but if you're if you're less
familiar with financial models,that's a good enough analogy. So
how much is the company going tomake in a specific period of
time? How much cash is it goingto produce? We're predicting
EBITDA 14 months in advance,plus or minus 4%
by early 2024 and then at theend of 2023 that same model made

(06:29):
a prediction for 2024 calendaryear EBITDA, and that prediction
for total cash flow for thatperiod of time, calendar year
2024 was with when within 1.5%so we literally know in our
businesses now, not just howthey're performing last month,
not not just how they'reperforming this month, but how
they're going to perform for thenext 14 months. How much more

(06:51):
risk can you take in a business,if you know between now and all
the way through 2026 how yourbusiness is going to perform,
and how can you take calculatedrisks in that business, knowing
that you're going to have thecash flow to be able to support
those risks that you're taking,and what you're able to do in
terms of innovation and reinvestin reinvestment in your
organization is light yearsahead of where we used to be

(07:12):
looking over our shoulder at theprior month. So really, really
cool stuff happening usingpredictive analytics and AI, you
know, we also use artificialintelligence to do all the show
notes for the podcast, at leastthe first draft of those, the
episode titles, chapter markers,social posts, a lot of those are
done transcripts and all therecordings. So so much of what
you're hearing today is doneusing artificial intelligence.

(07:34):
And so we've got phenomenalmarketing people in our in our
company, the TechEd media groupthat owns the podcast. We have
an amazing producer that I lovetalking about, Melissa Martin,
who really is the heart of thepodcast. Say that all the time,
and I mean it, but just a greatteam of people, but those people
are doing things that are valueadded, building relationships,
finding new guests, looking attrends and understanding our

(07:55):
markets, and a lot of theblocking and tackling that
happens in a podcast, including,like I said, all the production,
the social media, first pass ofthe show, notes and so on. All
that is done using AI backplatforms. We use them in our in
our marketing pieces as well. Imean, you get a picture
sometimes that is a beautifulpicture you want to use in a
marketing piece, but it's like,really busy or has a bunch of

(08:16):
stuff in the background. We canuse a platform like Canva is one
of them that we use. Just takethe background out of the
picture. Another great example,we'll get pictures sometimes
that aren't perfect, that wewant to use in a marketing
marketing materials, maybe apicture of a guest that is, you
know, a beautiful picture, anice, bright smile, but maybe
the shoulder is cut off in thepicture. We can use AI to give

(08:37):
that person a shoulder so itlooks normal when we put that
out in social media. One of myfavorite examples is, you know,
we get, we get photos every oncein a while from folks where
they're not quite high resenough. We can use AI just to
turn them into an into a highres image. It's not, not hard to
do. And then the other one is, Imight love this example too. Now
Secretary Doug Burgum, Secretaryof the Interior United States of

(09:00):
America. Is a friend of thepodcast, friend of Matt
kirkners. We were doing amarketing piece not too long
ago, and we had a great pictureof the two of us together, big
smiles, happy day. But I had ared tie on, and it was clashing
with the concept for the podcastor for the marketing materials.
I should say that we wereputting out. Hey, just used AI
to give me a blue tie. It's justthat easy. It's not, not

(09:22):
difficult at all. So really,really cool applications for
artificial intelligence, forgpts, practical applications in
manufacturing. And if you'relooking for an easy place to
start, some of those, some ofthose turnkey software
platforms, really, really greatplaces to start. I've got, we
use probably 40 of them. By theway, in our businesses,
altogether, the list may be evenlonger than that. I've got at

(09:43):
least 10 of them that I use todo slides, to do photos, to do,
you know, to generate imagery,all that kind of stuff when I
speak, just tapping into reallycool AI technology. All right,
staying on our manufacturingtopic, let's talk now about what
we call autonomous mobilerobots. So. So I spent some time
in the Nike of China, right? Soyou think about Nike and how

(10:03):
prolific that shoe brand is herein the United States, when I was
traveling to China earlier in2025 I spent some time in their
distribution facilities in theNike of China. And what I saw
absolutely, absolutelyfascinated me. So we saw all
kinds of shoe box after shoebox, package after package, and
they were all being movedautonomously. They had people at

(10:24):
the front end of the process atreceiving where product was
coming in, especially if it wascoming in in a non standardized
nature. In other words, youmight get a truck full of boxes
that weren't standard or so on,and it was a little hard to
automate that. We saw peopleworking in that part of the
business. We saw people workingon the back end of the business,
sending product out toindividual retail stores.
Everything in the middle,everything in the middle was

(10:45):
automated, wall climbing,autonomous mobile robots. So
they would pull up to a hugerack. Some of these racks that
we saw could literally be 1015,stories tall. Think about a rack
that tall in a distributionfacility, but these things would
drive up and down aisles, climbup and down racks, pick up
inventory, move it to where itneeded to be. We had, we saw
whole autonomous mobile robotsand AGVs, automated guided

(11:07):
vehicles. Difference being theamount of intelligence embedded
in the in the unit itself. AMRshave way more intelligence. But
we saw both and and these werethings that are moving inventory
all over a facility. No peopledriverless fork trucks just
moving tons and tons ofinventory all over that
facility. And as I suggest, theonly place that we really saw

(11:28):
people working after theyunloaded the truck at receiving
was at the final shippingfacility, where they were
loading up product at finalinventory. But we didn't see any
people other than those. Soreally cool applications for
autonomous mobile robots in allkinds of facilities, by the way,
when we asked about people aboutdeploying this technology in the

(11:48):
United States, so we said, Hey,this looks really cool. We'd
like to do it in the US. Thisautonomous. They call it ASRS,
so automated storage andretrieval systems. The people in
China literally laughed at us.
They said it would be four timesas expensive to do it here in
the United States as it was inChina. And they said it's, you
know, it's a combination ofsafety regulations. And look,

(12:10):
you're not going to get anyargument from me that we're
going to comply with safetyregulations here in the US, the
most important thing in abusiness is to send our people
home safe and sound healthy totheir families at the end of the
day. I'm all good with thesafety side of things. But they
said, look, the cost of ofexporting automation, whether
it's robot or robots or otherautomation, into the United
States super, super expensive,in part because of the the

(12:33):
implications of tariffs, andthen the other part of it is
you're building a big inventorystorage system. They said the
cost of steel in the US becauseof of tariffs. Huge implications
there too. They said it'd befour times more expensive to
deploy that technology here inthe US. So, you know, just a
little bit of a thought. It'snot a political thought, by the
way. I could, I could sit herefor 10 minutes and make a really

(12:54):
strong case for for tariffpolicy here in the United
States, and I could sit here andand make a really strong case
for not having tariffs here inthe United States for products
coming out of outside of thecountry. I'm generally a free
trader. I understand theadministration's position on
tariffs, and in manycircumstances, I accept the
logic and I don't argue with it.
The one where I think we have tokeep a really close eye on what

(13:18):
we're doing is with automation.
You know, we're not going tostand up automation companies
here in the United States fastenough to service the needs of
us, manufacturers, and if we aremaking it prohibitively
expensive for them to automateby placing tariffs on automation
equipment coming into ourcountry, we're putting putting
our manufacturers here in the USat a competitive disadvantage.

(13:39):
So that's not a politicalstatement I have, I have had
this conversation with peopleextremely high up about as high
as you can go in the USCongress, conversations with
people in the administration,and then they know how we feel
here at the TechEd podcast,about about making it
prohibitively expensive for usmanufacturers to automate. So
keep your eye on that. If you'rea manufacturer, we've got to

(14:01):
make sure that our manufacturerscan efficiently and cost
consciously, implementmanufacturing and automation
technology here in the UnitedStates of America. But let's
talk now a little bit more aboutautonomous mobile robots. I want
to mention a great company thatI've had the opportunity to get
to know well. Is auto This isdeployed artificial
intelligence, right? So whatyou're thinking about is a

(14:22):
driverless fork truck drivingaround a manufacturing facility,
driving around a distributionfacility, in many cases,
engaging with other people, whoare individuals who are driving
material handling equipment, butthey can take packages and
pallets and boxes and containersand move them wherever you want
them to go autonomously. There'sa ton of AI content on these

(14:43):
autonomous mobile robots. Sosmart sensors at the edge
gathering data in real time.
We've got LiDAR, we've got 2dcameras. We have a
microprocessor sendinginformation to a computer
network, sending thatinformation to the cloud. This
is the edge to cloud continuumat its most. Prolific here in
the United States, these thingswill literally they're
driverless fork trucks withoutoptimized material flow through

(15:04):
a manufacturing environment,through a distribution facility,
without a person doing anything.
So keep your eyes on AMRs. Oneof the people that I know that
had his eyes on AMRs is the CEOand founder of ClearPath
robotics company that owns automotors, the company that we were
just talking about autonomousmobile robots. His name is Matt
Randel. He, too has been a gueston the TechEd podcast. We'll

(15:27):
link that one up for you. He, bythe way, a couple years ago, a
company that he had foundedwith, I think it was three of
his college roommates. You canyou can validate that, confirm
that on the on the podcast, soldthat company two years ago to
another friend of mine by thename of Blake Moret, and Blake
is the CEO of RockwellAutomation, just an iconic
automation company, kind of thegodfathers of the programmable

(15:49):
logic controllers. Great UScompany and their CEO, Blake.
Blake Moret was on the podcast.
Point being, Matt Rendell soldhis company to Blake, to Blake's
company, Rockwell Automation,for $600 million so don't don't
tell me there aren'topportunities for young people
to innovate. Don't tell me therearen't opportunities in advanced
manufacturing, because MattRendell would probably have
something to say about that. AndI will also, while we're talking

(16:10):
about Blake Moret will make surethat his episode is also in what
may be perhaps the the episodeof The TechEd podcast with the
most links we've ever done inthe show notes. We'll make sure
Blake's episode is in there aswell. His company Rockwell
Automation will give them alittle bit of credit. Every year
they produce the annual State ofsmart manufacturing report. This
is the 10th one that they'vedone here in 2025 really, really

(16:31):
good content. And just a pullquote from that particular
report that was published hereearlier in 2025 95% of the
respondents to their survey, andthey had over 15,000
respondents. 95% of them haveeither invested in or plan to
invest in artificialintelligence, machine learning
and generative AI, or what wecall causal AI, little bit

(16:51):
outside the scope but another,another form of artificial
intelligence in the next fiveyears, AI adoption in the
manufacturing sector isoutpacing other industries, and
they point to large companies,especially among those companies
that are more than a billiondollars in revenue. But that is
coming to small to mid sizedcompanies, and it is coming
fast. So you can check thatreport out again. It's called

(17:13):
the 10th annual State of smartmanufacturing, and there's some
great data in there. Okay,number eight, smart drones in
manufacturing. Yet anotherepisode that, yes, we will link
up on the show notes, was ourepisode with Shawn Mitchell of a
company called gather AI. Now,if you've never worked in
manufacturing, if you've neverworked in accounting, if you've

(17:34):
never worked in a distributionfacility and had to do year end
inventory, or what we call cyclecounts. So doing inventory from
time to time to make sure thatwhat your books, what your
financial statements, what yourinventory list says you have, is
what you actually have ininventory, Count yourself lucky
if you've never done that,because it is a hassle and and
I've been through tons and tonsof physical inventories, and

(17:55):
they always end up with asurprise or two. You hope the
surprise is a small one and nota big one. But believe me, I've
had both. Sean Mitchell'scompany is solving for this
challenge, and they're basicallydeploying drones. So they have
drones flying in a manufacturingfacility, flying in a
distribution facility, andthey're gathering all kinds of
data, reading RFID, readingbarcodes, looking using what we

(18:15):
call classification machinelearning, which is a category of
AI, category of machinelearning, which is a category of
AI that's called classificationmachine learning, basically
looking at something andfiguring out what it is, using a
vision system. They're usingthat technology to do inventory,
to confirm what's in themanufacturing plant, to do piece
counts, to do cycle counts, tofigure out what pallets are
there, what boxes are there, inreal time. And boy, that just

(18:37):
solves all kinds of challengesfor a manufacturing environment.
But think about being able todeploy a drone in your plant and
have it count your inventory inreal time and feed all that data
to your ERP or MRP system on anongoing basis. Really cool
technology. So you're going tosee more and more applications
for both ground drones and andaerial drones in manufacturing,
because all the things that wecan do autonomously using

(18:59):
computer vision technology,which is a really, really key
along with LIDAR technology thatwe're seeing in terms of
allowing our machines to see inmanufacturing and then creating
inventory intelligence,understanding what's there,
where you may have an issue,where you may have a bottleneck,
where you may have lessinventory than you thought you
did, or, for that matter, maybeyou've got overstock or

(19:20):
something as well, having allthat in real time, super, super
valuable in manufacturing. Okay,now let's talk about what we
call AI powered industrialrobots. We will link Ariane
Kabir s episode up as well, graymatter robotics company out of
San Diego, California. I metAriane at a Automation
Conference in 2023we became fast friends.

(19:43):
Fascinating dude. He's raisedtons of money in his in his
company, gray matter robotics,from from investors that just
want to get a piece of the greatthings he's doing. Let's talk
about what Aryan is doing. Andwe'll start by talking about a
problem that I faced all thetime when trying to automate
Manufacturing Company. 90s. It'sjust one example. For 10 years,
I ran one of the largest,probably the largest, contract

(20:03):
metal finishing companies in theUnited States. So when we think
about metal finishing, if youthink about anything that's
plated using zinc plating orblack oxide or tin plating or
silver plating, like thecontacts in your phone, so many
things in automotive, forexample, and your vehicle are
plated in zinc to protect thesubstrate, to protect the steel

(20:24):
from rusting. You know, thinkabout paint or powder coat.
Anything that you have that'spainted or powder coated your
your patio furniture would be anexample. So we,
we ran a company though, a hugecompany, 16 automated production
lines that we're doing moremetal finishing and plating than
almost anyone else in the world,and we had so many different

(20:46):
parts that came through thatplant, right? So 16,000
different what we call skews. Sothat's not 16,000 units of
inventory. That is 16,000different distinct types or
designs of parts, right? Somillions upon millions of parts
running through that plant, but16,000 different versions of
them, so no two days were everalike. And so we were in what we

(21:06):
call a high mix, low volumeenvironment. So we've got every
order is a smaller order, huge,huge mix in terms of the
different products that we'rerunning in that manufacturing
plant. So automating thatprocess is hard. It's one thing.
I was in a plant in China wherethey make 8 million of the same
thing every day. That's reallyeasy to not easy. It's
relatively easy to automate,compared to automating something

(21:27):
where you have 16,000 differentthings you make every year and
every single day is different.
So it's really hard in the pastto automate processes with lot
sizes that are small. Well now,with lot sizes as small as one,
let me talk to you about whatArians product is doing. You
know, we used to have, forinstance, people using grinders,
grinding parts in a metalfinishing facility. So we've got

(21:47):
parts that come in, and we needto grind those. We need to
surface finish those thatsomebody with a hand grinder.
You could picture something youmight buy at Home Depot or Ace
Hardware in your grinding partswith, with a with a rotary
grinder and just just grindingthe surface of a part. Well,
what our hands company did, andI think this is really, really
fascinating, is they basicallytook a 3d scanner and put it at

(22:09):
the end of a robot, and thenpresented that scanner to a
production part in a lot size,as small as one, and they scan
that part using 3d scanning, andthen they use that 3d scan, and
they create what we call an STLfile, or a CAD file, basically a
digital file of the design ofthat part. So they create the

(22:29):
digital file of the design ofthe part using 3d scanning,
using CAD software, and thenthey program the robot using AI
to finish the part. So now,rather than having human beings
holding a sand, a sander or agrinder, not a glamorous job,
not a not a particularlyfascinating job, and not always
a safe job, because there's alot of potential for soft tissue

(22:50):
injuries and repetitive motioninjuries, and now we've
automated that and made that jobsafer, taking the technology
closer to the robot, the peoplethat are now programming and
operating those robots makingmore money than they would if
they were just sanding thoseparts. So just an example of the
advancing technology, but anexample of how we can use these
AI enabled robots to improveperformance. The same thing is

(23:11):
happening in the space ofmeasurement engaging. So if you
think about how we used tomeasure parts, and in many
cases, still do in amanufacturing environment, a lot
of folks are still using what wecall micrometers, or dial dial
calipers, where you're basicallydialing a measurement system and
then measuring parts based upon,you know, based upon their
physical measurement. So they'llbe like using a tape measure,

(23:32):
something a little bit more,quite a bit more advanced and
more precise than that, butmeasuring parts in a in a
manufacturing environment tomake sure, you know, form, fit
and function, make sure thatwe're within, within our
tolerances for how big that partcan be. Well, that's manual, and
it's open to error, and it'sit's not particularly efficient.

(23:53):
Well now with next gen AImetrology. So Metrology is kind
of a fancy word for how do wemeasure stuff in manufacturing
or in other in otherdisciplines. So using next gen
artificial intelligencemetrology, what I learned from
my friend Fannie trishaun, whois also a guest here on the
podcast, her she's the presidentof a company called Creative

(24:13):
form, linking that one up too.
So when Fanny joined me on thepodcast, we talked about her
technology and her company'stechnology, what she has done is
taken a 3d scanner, similarsomewhat to what gray matter
robotics did, and put it at theend of a robot. We put it at the
end of a collaborative robot. Wecan put it at the end of a a six
axis traditional industrialrobot. We can also use a

(24:35):
handheld scanner as well, buteither either putting it at the
end of a robot and automatingit, or scanning a part by hand
using 3d scanning technology.
Now we can scan that part. Sothink about running, you know, a
unit about the size of the, youknow, maybe three or four times
the size of your of your phone.
Or, you know, in the case of a,what they call their Metro scan,

(24:55):
it's about the size of abasketball. But run that across
a part, over the top of a part.
Right, and you get all thedimensions of the part. So you
can, if there's key dimensionsthat you're measuring, you can
measure all of that. You can getall of your dimensions. You can
test you can test all yourtolerances. You can make sure
the part is fitting thespecification using technology,
as opposed to using some ofthose hand driven systems that
we used in the past. Also agreat way, by the way, to create

(25:18):
a CAD file of a part. So if youwant to reverse engineer
something, yeah, an examplewould be somebody out on a on a
Navy ship in the middle of theocean, and they have a failure
on a part they use to to runthat naval vessel. They can't
just order that on Amazon,right, have it delivered out in
the middle of the ocean. You canbuild it on the ship. So scan,
scan apart, create a CAD file,and then create the physical

(25:40):
manifestation of that part usingnext gen AI metrology and 3d
scanning. So another examplewhere we can use artificial
intelligence in a manufacturingopportunity or environment
without building our own ACPserver, MCP server, I should
say, without building our ownagent, without building our own
knowledge graph. It's somebodyelse that has engineered this
for us and and we bring it intoour facility. A lot of these

(26:02):
technologies, by the way, whenwe think about capex or capital
expenses, spending money andinvestment in manufacturing,
Aryan Kabir is product or graymatter. Robotics is a is
basically a paper pay forservice. You're basically paying
a monthly fee to use thetechnology. So it allows it to
become an operating cost in manycases, and and also helps you
avoid that huge upfrontinvestment, in some cases, in

(26:25):
technology, because you'repaying it over the course of
time. Not to say that his, hishis product, is prohibitively
expensive, and that wasn't theimplication. But rather than, if
a company, for whatever reason,wants to make investments in
other parts of his business withits cash, they can turn this
into an ongoing expense, asopposed to having to pay for it
up front. All right. Keepinggoing here, having some great

(26:46):
conversations about advancedmanufacturing technology. Now we
are on a topic. Back to myfriend Leo Reddy, my 92 year old
friend, who is learning allabout artificial intelligence.
He did work for the StateDepartment. He worked there from
for a number of years in the 70sand 80s. During part of that
time, he reported to the personwho was secretary of state from,
I believe, 1974to 77 give or take a year, I

(27:07):
might have it perfect if youwant to think about who was the
Secretary of State of the UnitedStates of America from 1974 to
1977 and if you guessed HenryKissinger, you would be right.
Kissinger, as many know, passedaway toward the end of 2024 he
was close to 100 might have been100 but Kissinger wrote a great
book, and it's one of the it's,I would just say, if you're

(27:31):
starting your AI journey, or ifyou haven't read the book yet,
it's a must read. And it's abook called Genesis. He wrote it
with two other people, CraigMonday, who is the Senior Vice
President of Strategy forMicrosoft. And Eric Schmidt, who
is the CEO of Google, and thethree of them, wrote this book
called Genesis. It's calledartificial intelligence, hope
and the human spirit. How do younot get excited about that

(27:51):
particular subtitle? So checkthe book out. But here's one of
the things they talk about, allkinds of things in AI. One of
the things they go deep in isthis whole idea of AI generated
smart materials. I'm going toquote from the book. It says,
quote, AI will be put to useresearching and developing
increasingly cheap and abundantsources of raw materials for its
own inputs as AI issimultaneously deployed in

(28:15):
manufacturing, that's our topic.
It could reduce the capitalneeded for any given good. So
think about that. AI is going tobe developing its own cheap and
abundant sources of rawmaterials for its own inputs. In
other words, as AI tries toexpand itself, it's going to be
developing really, reallyinnovative materials. And so to

(28:35):
quote from another part of thebook, quote using new,
sustainable synthetic materials,AI could AIS could build cities
around the world to provideshelter, regulate temperature,
ensure access to power anddigital connections and provide
clean water, food, medicine andsanitation. So those are the
kind of things that we are goingto be using AI for in the

(28:56):
future. And again, that is aquote from the book Genesis,
written by Henry Kissinger andtwo of his co writers. So we
start thinking about what wecall AI, accelerated discovery
and design. That's where allthis is going, that the
materials we use inmanufacturing are going to look
different, and we're going tohave all kinds of different
materials. We're not just goingto be talking about, is it mild
steel? Is it stainless steel?
You know, in all the othervariables that we're tracking,

(29:18):
we're going to have all of theseother variables in
manufacturing. We're going to beable to create materials and
machine materials and formmaterials and in mold materials
that we have never thought of inthe past. They're going to have
all kinds of really, really coolproperties that we've also never
thought of. Some of the greatexamples that I love. They're
talking about having, literallyprogrammable matter, shape

(29:40):
shifting materials with embeddedintelligence that can transform
as needed. It can transform ondemand for adaptive
applications. Another examplewould be self healing
composites. So we saw this inthe metal finishing space for a
long time, where if we had apart that got damaged, we could
self heal it an example.
Would be a chromate convertedpart, where we put zinc plating

(30:02):
on a part, and then we put aPromate chromate conversion over
the top of it. And if thatchromate got nicked, it would
actually grow back over theNick. Over time, we're going to
see all kinds of applicationsthere materials that can detect
their own damage, repair theirown damage automatically, and
that, of course, will expand theand extend the lifespan of a
product, in many cases,dramatically. But other, you

(30:23):
know, other examples are quantumengineered materials, materials
that are AI designed to stand upto extreme environments,
biometric nano structures. Wetalked about carbon negative
materials that could be could beaccretive to sustainability. So
all kinds of great thingshappening in the world of
materials here in the age of AI.
All right, so I saw this quotenot too long ago. It's from a

(30:46):
guy named Jensen Wong. If youdon't know Jensen Wong's name,
he is known to many, known tomany of us, and is the CEO of
Nvidia. And here's what he said,I'm gonna, I'm gonna put a blank
in here and let you answer thequestion in your head. He said,
cars, drones and blank are theonly three types of robots that
can scale to extremely highvolumes because they can be

(31:08):
deployed to the world as it is.
I'm going to repeat that quoteagain. It's one from Jensen Wong
of Nvidia. Quote, cars, dronesand blank are the only three
types of robots that can scaleto extremely high volumes
because they can be deployed tothe world as it is. What's the
answer? If you guessed humanoidrobots, cars, drones and

(31:28):
humanoids are the only threetypes of robots that can scale
to extremely high volumes,because they can be deployed to
the world as it is. If that wasyour guess, you are absolutely
correct. Let's talk about number12 humanoid robots. When I was
in China, I saw no less than, Ibet it was eight, at least six
humanoid robot companies. So youthink about a humanoid robot, I
think most of our audienceprobably knows what that is,

(31:49):
but, man, they look just likepeople, or at least like robotic
people. They've got arms,they've got legs, they can
march, they can walk, they canbalance. They can see. They can
stand up. They can sit down. Youcan knock them down. They can
stand back up. They can sit in achair. They can, you know, load
inventory. They can rack andunrack parts, not perfectly yet,
so there's still some innovationthat has to happen before we see

(32:10):
these widely deployed in in theworld of manufacturing. But it's
coming. Elon Musk is oneexample. Predicts a day where we
will have 10 billion humanoidrobots on the planet, more so
than the than the number ofhumans doing all kinds of
things, doing our laundry,making our dinner, washing our
cars, driving our cars. If weeven need a humanoid, obviously
those will be self driving, butall kinds of tasks, and then,

(32:31):
certainly in manufacturing,think about any application
where we could benefit fromhumanoids, racking unwrapping,
performing dangerous ormonotonous jobs as the cost of
labor goes probably to about $1an hour over time, through
through humanoid robots andmanufacturing, but lots of
really crazy stuff happening.
What I love about humanoidrobots, when we think about I do
a whole separate discussion.

(32:53):
Maybe we'll do this as aseparate podcast sometime this,
this year here, before the endof 2025 on the different
materials and the differenttechnologies that are advancing
advanced manufacturing, many ofwhich we've talked about today,
advanced materials, autonomoussystems, battery technology,
biomimicry, so mimicking what wesee in nature as we do our
innovation, the cost and thepower of compute, the global

(33:17):
positioning systems,electrification, vision systems
like Lidar and standard vision,smart sensors, telemetry. All
these technologies areinnovating different aspects of
manufacturing. They are allinnovating humanoid robots. So
we have humanoid robots becauseof all that, that entire list,
which I won't go through again,you can, you can hit your 15
second rewind if you want tohear it again. But those

(33:37):
technologies that are totallytransforming manufacturing, we
see every single one of them onhumanoid robots. And so as we
talked about a moment ago, weare seeing huge improvements in
efficiency relative to thenumber of people we need in
manufacturing. If you look backin the 1980s
in order to produce a milliondollars of revenue in real
dollars in the 1980s so in realdollars, what we mean by that,

(34:01):
of course, is inflationadjusted. So taking the effect
of inflation out of the numberswe're sharing with you, seven
and a half people per milliondollars of revenue in 19 in the
early 1980s that's how manypeople it took to produce a
million dollars of revenue inmanufacturing. And the s, p5,
100, by the way, is the is thesample source, or the source of
data for that number, seven anda half people down to in the

(34:23):
year 2022 the last number lastyear. We have the data for this,
by the way, according to bothBank of America and a study that
they did, down to two people. Sowe went from seven and a half
people to produce a milliondollars of revenue down to two
people to produce a milliondollars of revenue over the
course of that period of time,1980s to 2022 and if you think

(34:43):
about that path that we're on,that trajectory, the downward
project trajectory, it's goingto continue to shrink. We're
going to produce more and moremore and more revenue with fewer
and fewer people. One of myfavorite quotes from Sam Altman
open AI, he expects that thatwe're going to see.
One person, billion dollarcompany. He said that at
TechCrunch, actually. And soit's super, super fascinating

(35:06):
that if you think about we'regoing to get to a point where we
will see one individual, oneperson, leveraging AI,
leveraging technology, creatinga billion with a B dollars of
revenue. So the amount of laborthat we're going to need in
manufacturing relative to thethe amount of revenue that we
can produce is going to continueto shrink in with the advent of
humanoid robots, that's going tobecome even more acute, which

(35:29):
leaves us with all kinds ofquestions, as we kind of get
near the end of this episode ofthe podcast, all kinds of
questions about the future ofwork. So let's talk about the
breakthroughs of and the impactof AI in this age and talk about
the future of work. Anothergreat study, this one was done
by three economists, EdwardFelton from Princeton, Manav Raj

(35:49):
from the University ofPennsylvania, and then Robert
Siemens from New YorkUniversity. Those three
individuals did a fascinatingdid some fascinating research,
and they looked at how languagemodelers like chatgpt and other
generative pre trainedtransformers are going to affect
both occupations and industries.
They looked at 1016 differentjob categories. And those job
categories the study, by theway, I read about in the book co

(36:09):
intelligence, by Malek, which isanother great book from 20,
24,016job categories. And they looked
at the degree to which those jobcategories were going to be
disrupted in the age ofartificial intelligence, what
they found was 36 of them. Afull 36 of those 1016 jobs were
unlikely to overlap with AI. Andlook, I know some of us are
afraid of artificialintelligence. Some of us don't

(36:30):
like change. Some of us areworried about the world of work
changing like crazy. Well, ifyou're one of those people, and
you are a pile driver operator,or you are a roofer, or you are
a professional dancer, the newsis actually really good for you,
because your world is unlikelyto change significantly. For the
rest of us that aren't in those36 jobs, those jobs, or the
other 33 jobs like them, theworld is going to change like

(36:54):
crazy in the age of artificialintelligence. It really is, and
we need to be ready. We have alot of educators that listen to
this podcast, I can tell youthat same study said that of
those of the top 20, of the top20 jobs that were most likely to
be disrupted by artificialintelligence, of the top 20 jobs
most likely to be disrupted byAI, 10 of those are in the
education space. Now, most ofthose are university professors.

(37:17):
So depending upon what level ofeducation you're working in,
maybe a little bit lessinnovation and disruption in the
short term in spaces like K 12or Technical and Community
Colleges, although, trust me,it's coming in the space of
higher education at theuniversity level, I think we're
in for some really, reallysignificant disruption in the

(37:38):
next three to five years.
Anyway, a lot of us are worriedabout losing parts of our job to
artificial intelligence. I knowa little bit about what that is
like. We talked earlier abouthow manufacturing people love
spreadsheets, right? There areweeks where, if I could live my
whole life in a spreadsheet, Iwould just kind of grew up with,
you know, first it was first, itwas lotus, 123, and there were a

(37:58):
couple other Apple basedplatforms. And then, of course,
Microsoft Excel for the last 30years. And love Excel, love
data. I was always the guy inthe office they would come to
with questions, right? Ifsomebody wanted to know about a
macro, if they wanted to knowabout a formula, if they wanted
to format a spreadsheet, whowould they come to? They come
into my office and they say, I'mjust, can you just tell me to
show me how to do this?
Struggling with this, I wasalways happy to do it. And then

(38:20):
I noticed about a year and ahalf ago that it had been a
while before or since anybodyhad walked in and asked an Excel
question. And I actually praisedmyself a little bit, right? All
these years of trying to trainmy teammates on how to use
Excel, we finally reached theend of my knowledge. There
wasn't anything I could helpthem with. They were off on
their own. I was so proud. Andthen I thought for another
moment and realized that thathad nothing to do with it. That

(38:40):
really the reason that thesefolks were no longer coming to
me with questions. And by now,I've dwelled enough on the
subject that you've all figuredit out they weren't coming to me
because they were going tochatgpt, right? They were going
to perplexity, they were goingto Claude asking a question,
asking for a formula, asking fora macro, asking how to do
something, and it was givingthem the answer at the snap of
their fingers. So I know alittle bit about what it's like

(39:01):
to lose part of your job to AI,because I did now some 18 months
ago. It's okay, though, ithappened with my son too. I have
this son that lives inWashington, DC. He works in
consulting, and he TechEd me nottoo long ago, and he said, Have
you ever used chatgpt to writean Excel macro? He said, I did
it yesterday, and it was lifechanging. So, so in that case,

(39:22):
generative AI certainly tookpart of my job away. What was
life changing for my son? Isokay, though, by the way, that
is the same son who told mequote I always say, Please and
thank you to chat GPT, becausewhen the robots take over, I
want them to like me. All right.
Well, I'm not worried about therobots taking over, but I will

(39:42):
tell you that our world is goingto change here in manufacturing.
I'm going to close this extendedepisode of the podcast with a
little bit of a story. It's anoldie, but a goodie. If you
heard it before. Bear with me,but I love this one. It's about
a woman who lives in a home, andin her house she has.
A floor that creaks. And everytime somebody walks across the

(40:03):
floor, it makes this creakingnoise. It's in the kitchen, if
somebody is like relaxing oreven sleeping in the next room,
and somebody walks through thefloor, the creak is so loud that
it can wake them up. And shegets sick of hearing this creak
day after day, week after week,month after month, year after
year, and so she finally breaksdown and says, I'm going to call
the carpenter and have him comeand fix this. And so she she

(40:24):
calls him, and the carpentercomes, and he stands in her
kitchen, and she shows himexactly where the squeak is. And
he he patiently walks one wayacross the floor, and he listens
to the squeak. Then he turnsaround, and he patiently walks
the other way across the floor,and he listens to the squeak.
And then he looks at the floorfor a bit, and he kneels down on
the floor, and he and he pullsone nail, a single nail, out of

(40:47):
his out of his tool belt. Thenhe in with his other hand, he
pulls a hammer out of his toolbelt, and he places the nail
carefully in one specific spoton the floor, and he drives the
nail into the floor using thehammer, and then he stands up
and he walks across the floor,and there's absolutely no
squeak. And he looks at thewoman and invites her to do the
same thing. And she does, andshe walks across the floor, and

(41:10):
she's in and there's absolutelyno squeak. And she said, Oh,
thank you. She said, I can'ttell you that was just driving
me crazy. It was so loud, youknow, sometimes people literally
couldn't sleep in the next room.
I am so, grateful to you forwhat you did here. And he pauses
for a moment, and then he takesa piece of paper out of his tool
belt. Then he writes on thepiece of paper. He writes
invoice across the top of thepiece of paper, and then he

(41:33):
writes on the invoice, fixsqueaky floor, $75 and he hands
it to the woman, and the womanlooks at it, and she says, Well,
$75 she's like, thank you somuch. I mean, I really
appreciate the work that youdid, but $75 that seems like a
lot of money. All you did waspound one nail. And he takes the
invoice back from the woman, andhe crumples it up, and he puts

(41:56):
it in his pocket, and he takesout another piece of paper, and
across the top, he writesinvoice, and he writes pound one
nail. And next to that, hewrites $1
and then below that, he writes,knowing where to pound the nail,
$74 total, $75 and he hands itback to the woman. You see, the
truth of the matter is, myfriends, that many of us, and

(42:18):
especially our parents, grew upin an era of physical labor,
right? They grew up in an erawhere you could get paid for
what you did, right? If youworked on a production line, if
you were a carpenter and you andyou built houses, I mean, what
you did with your physical body,what you did with your time,
that was a great way, in manycases, to earn a living. And I
have people in my family who I'msuper, super close to that

(42:40):
earned great living. Earn agreat living working in
manufacturing, in, you know, outon production lines, in
manufacturing environments, andsupported their families and
created great futures forthemselves and so on. So the
truth is that individuals inthose spaces were living in a
physical labor era. They wereliving in a physical economy,
and they got paid for what theydo and what they did. Well, then

(43:01):
we evolved into a knowledgeeconomy where it was no longer
an economy where we can get paidfor what we do. The world moved
on. We automated a lot ofprocesses, and it became all
about what we know. And so nolonger were we paid for pounding
the nail. We were paid and weare paid for knowing where to
pound the nail, and we've beenin a knowledge economy for quite

(43:23):
some time, some might argue, 50or 60 years, where knowledge is
power, and the more we know, themore earning power we have.
Well, we are quickly evolvinginto a time where knowledge
isn't going to be good enough.
We are moving into what we callthe intelligence economy, and
that is going to be an economyin which we are paid for how we
use artificial intelligence, andthe better more adept you are at

(43:45):
implementing turnkey AI systems,the more adept you are at
understanding artificialintelligence, whether that's
generative, pre trainedtransformers, whether it's
embedded technology like that wetalked about, whether it's
predictive analytics, the Folksthat understand how to deploy
artificial intelligence and howto leverage it in the work world
own the day in the intelligenceeconomy. So as we move toward

(44:07):
that period of time, it isreally, really important that we
stay on top of this stuff. Thereason I wanted to do this
podcast, and the reason that wededicated such a long episode to
advancing technologies in theworld of manufacturing, and
particularly artificialintelligence in the world of
manufacturing, is that this iscoming quick. I started in my
trip to China. They're using AIin all kinds of ways. We're

(44:29):
starting to do it in many wayshere in the United States, in
some ways, we're leading here inthe US. The race is still in
process, and who ends up winningthe day is yet to be determined.
I'm betting on the United Statesof America. I never will bet
against our country, but we'vegot to get this right, and
especially for our small to midsize companies, we've got to be
investing and understanding howto deploy artificial

(44:52):
intelligence to improve ourprocesses, because other
companies in your space aredoing it. The ones that do are
going to be the ones.
End up way ahead and five yearsfrom now, the ones that don't, I
worry, and I'm quite certain,are the ones that are going to
fall behind, many of them, tothe point where they won't even
be in existence five years fromnow. So as we close our time

(45:13):
together, I will just reiteratethe fact that back in early in
the knowledge economy, and evenpartly in the physical labor
economy. I worked inmanufacturing 30 years, running,
leading, owning, investing inindustrial companies, advanced
manufacturing companies, and Ilived and benefited from that
knowledge economy. And while wewere still in that knowledge

(45:34):
economy, and after selling ourmanufacturing company, my
partners and I did at the end of2014 I dedicated the rest of my
career to securing the AmericanDream for the next generation of
STEM and workforce talent, andjust as important, literally,
securing it for the nextgeneration and manufacturing
here in the United States, sucha huge part of our economy is
going to play a huge part insecuring that American dream.

(45:57):
Adopt artificial intelligence.
Understand how it's going tochange your market space. How
it's going to disrupt yourmarket space? Upskill yourself,
equip your team with the skillsthey'll need to be successful in
the age of AI. Let's get goingon our AI journey. And when we
input and when we implement thistechnology in advanced
manufacturing, together, we willsecure the American Dream for
the next generation. Thanks somuch for tuning in to this

(46:20):
episode of The TechEd podcast.
We are going to put this episodeat TechEd podcast.com/applied
AI, that is TechEd podcast.com/aP, P, L, I, E, D, A, I, so check
it out there. And when you'redone, come see us on social
media. We've got all kinds ofgreat stuff that we're doing on
social whether it's Facebook orTiktok or LinkedIn or YouTube, I

(46:43):
follow our podcast on every oneof my feeds. It's amazing. The
kind of cool stuff that our teamputs up all the time. I learn
from the things that end up onsocial media from the TechEd
podcast. So be sure to tune inand follow us. And the other
thing we should make sure you dois tune in next week to the
TechEd podcast, where this weekand every week, we impart all
kinds of great knowledge to theincredible people doing

(47:06):
incredible work in education andmanufacturing and across the
entire US economy. My name isMatt Kirkner. I'm your host for
the TechEd podcast, and thanksso much for joining us.

All Episodes

How to Get Started with AI in Your Business (Practical Tips & Real Technologies) - Part 2

Episode Transcript

Popular Podcasts

On Purpose with Jay Shetty

Ruthie's Table 4

The Joe Rogan Experience

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}How to Get Started with AI in Your Business (Practical Tips & Real Technologies) - Part 2