March 18, 2026 • 35 mins
A 20-year veteran retires, and suddenly the “way we do it” disappears with them. That is the reality across manufacturing, maintenance, and field service right now, and it shows up everywhere: longer onboarding, inconsistent work, safety gaps, and teams stuck relearning the same fixes under pressure.
We talk with Siva Kumar Lakshmanan (Siva), CEO of DeepHow, about a practical way to capture tribal knowledge before it walks out the door. We dig into how generative AI and large language models can transform real shop floor work, including video of experienced technicians, into training that new hires can actually use. Along the way, we get specific about adoption, because the hardest part is rarely the software. It is change management, trust, and proving value in a way that makes sense to skilled workers who would rather be on their feet than at a desk.
You will also hear a clear framework for when to move fast on AI and when to wait, how to run pilots that give the technology a fair shot, and how to use KPI scorecards to make fact-based decisions without burning out the team. We close with a candid conversation about job-loss fear, why business-case transparency matters, and how to position AI as a tool for safety, standardization, and faster time-to-competency in supply chain operations.
Subscribe for more Supply Chain - Unfiltered, share this with a colleague in manufacturing or procurement, and leave a review if the conversation helps you rethink training and technology adoption.
What is the one process in your operation you would capture first?
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_00 (00:07):
This is Supply Chain Unfiltered, presented by the
Institute for Supply Management.
SPEAKER_02 (00:17):
Good to have you here today.
I'm Melanie Stern for ISM, anduh this episode of Supply Chain
Unfiltered is gonna be a funone.
Um it's actually uh talking witha company, um Deepal, that's
their name, first had um them onthe show probably about 18
months ago.
And um was actually, I think,probably one of my favorite
(00:39):
episodes.
It was just so fun to hear whatthis company was doing.
And um they had introduced alanguage learning model, and uh
it was changing the way thatglobal companies can scale and
and train their employeesbecause there's lots of
intricacies in communicationthat can go really well or go
(01:04):
haywire if you're working withpeople from different cultures,
different um places in theworld, and which can make
training really difficultsometimes.
So they had a great solution,and um if you wanted to check
out that episode where we firstchatted with them, I think it's
episode 38.
Nonetheless, fast forward totoday.
(01:26):
A lot has changed since my firstconversation with them.
So we're gonna get all caught uptoday.
Um, please welcome CEO ofDeepow, Steva Kumar Lakshmanan.
How are you, Steva?
SPEAKER_01 (01:40):
Very good, very good.
Thank you.
Thanks for having me.
SPEAKER_02 (01:42):
Awesome.
Yeah, um, I love what you guysdo.
So I think, in all fairness, thebest place to start would be if
you could just tell us a littlebit about yourself and Deepow's
mission for anyone that's um youknow isn't aware of of all the
great things that you do.
SPEAKER_01 (02:01):
Perfect.
Appreciate the opportunity totalk to you and to your
audience.
I am uh Sivakumar Lakshmanan.
I go by Siva.
I live in uh Dallas, Texas.
SPEAKER_02 (02:13):
Yeah.
SPEAKER_01 (02:14):
Um and Deepau, let me quickly give you a high-level
picture of what mission is andwhat we do.
Deepau is uh Deepau's mission isvery simple.
It is to empower skilledworkers.
These are people who buildthings that we use every day,
(02:35):
who manage, maintain things.
We often refer to that as coreindustries or skilled work,
things that need physical work.
And globally, and particularlyin the United States, there's a
shortage in availability ofskilled workers, particularly in
(02:57):
manufacturing.
If you go to a manufacturingindustry, you are typically
doing three shifts, runningthree shifts, people go in at
six o'clock, work eight to tenhours, and then the second
shift, and then the third shifthappens, and that's a very
strenuous work.
As the demographics change,people preferring to work in a
(03:20):
manufacturing setting is lowerand lower, and as a result, this
industry is going through alabor shortage.
And while that is happening,there is also an aging workforce
that is retiring, and thatknowledge, the tribal knowledge
of skilled work, is leaving withthese very experienced people
(03:43):
who have done this for 35, 40years, who knows this, who has
developed muscle memory of howto do these things.
They are never digitized, theyare never captured in a
consumable way.
And that is where default comesin, is to capture this knowledge
in a way that is consumable,that is immersive and useful for
(04:07):
the next generations of workerswalking into these factories or
these field services, takingover an absolutely economically
critical work that we need tomake everything that we use
today.
So Deep How makes this happen.
Deepow takes captures thisknowledge, converts this into a
(04:31):
know-how training, leveraginggenerative AI, also known as
large language models.
So this generative AI is the onethat captures this knowledge,
converts it into something very,very useful for people to
acquire, onboard themselves, useit, troubleshoot whatever they
(04:52):
want to do with.
SPEAKER_02 (04:55):
Okay.
Um, who wouldn't want that,right?
Um so so in this quest um thatyou have to basically help um
improve employer to employeerelationships, um, improved
interdepartmentalcommunications, you've got the
training for the labor force,and enhance the productivity
(05:18):
through through your languagelearning model.
I I would imagine that thecompany has hit some of the
goals that you meant to over thelast few years, but there were
probably, I'm guessing, somesurprises, some good and maybe
some not so good.
Um but would you mind sharing afew of those wins and and
(05:42):
learnings with us?
And the reason I ask is I knowum there's so many conversations
that take place about um AI andwho's doing what and what the
latest um in the advancementsare.
But I guess um I'll say, youknow, in comparison, we hear
(06:02):
about those stories, but wedon't tend to hear a lot about
what we find out once we rollout these new uh programs, these
new innovations, and um what howdoes testing go and what what
went right and what what needsimprovement.
So I'm wondering, can you sharesome of those with us?
SPEAKER_01 (06:22):
Absolutely.
Any any journey uh involves youknow wins and lose losses,
right?
And building a company um takesmore of those in a very shorter
period of time, and particularlyif you're dealing with AI, and
if you add a layer ofblue-collar work to it, then you
(06:44):
have a roller coaster ride.
And and which is which is whichis the exciting part about uh
this is that's the exciting partfor me, uh being the CEO at DPA.
What what are the what are thesome of the challenges you face?
You why do you why do you wantto go and work in a
manufacturing facility or dophysical work?
(07:05):
It's because you generally don'tenjoy disk chart, right?
You're not a fan of sitting infront of a computer writing
emails, etc.
I want to be out there, I wantto be on my feet, I want to use
my physical skills to achievesomething.
That's generally the careerinterest that draws people
towards this.
SPEAKER_03 (07:21):
Okay.
SPEAKER_01 (07:21):
Now you are suddenly introducing a technology to them
to say, hey, here is atechnology that's going to make
you be really good at what youdo.
Uh, what do you think is goingto be the response?
The response is um yeah, thankyou, but I don't think so.
Exactly.
I like the pillars and themachines in my hands.
Right there is there's a changemanagement, there's an adoption
(07:44):
part of it that's extremelychallenging.
So if you think AI ischallenging on how kids adopt AI
for education, or AI ischallenging, how are doctors
going to adopt?
This is not very different,right?
Uh it doesn't matter which isthe industry, some more complex
to adopt than the others.
So similarly, we see that inskilled work and manufacturing
as well.
(08:04):
How do people recognize where itcan add value and where it
doesn't?
And that is the mission that wehave undertaken at DePOW is to
distill the noise around this.
We always tell our customersthat it we think it is our
responsibility to identify whatis snake oil and what is real
(08:24):
value, and we want to distillout all the snake oil part of
AI.
Focus on the things that matter,things that are meaningful,
things that are going to addvalue.
Let me give an example.
Oh, please.
So if you if you take a if youtake a factory setting and
(08:44):
you're going to go in and workin a factory, there is a
procedure called lockout tagout.
SPEAKER_03 (08:49):
Right.
SPEAKER_01 (08:50):
Lockout tag out procedure is if I'm going to go
and work on maintaining amachine or cleaning a machine or
doing something with it, youneed to make sure that the
machine is switched off, right?
You don't want someone toaccidentally switch on the
machine.
It is like you going andswitching off the main switch at
your home and you're working onreplacing a bulb and your wife
is like, hey, why is there I'mnot getting a bar?
(09:11):
No, some idiot switched it offand then switched it on and
you're working on it.
That's fatal, right?
That's dangerous.
So the lockout tagout procedureactually makes sure that whoever
is doing this locking out isactually making sure it is
sealed, it is documented, and noone can accidentally switch it
back when you're working onthat.
Now, this was okay when everyonein your factory had a
(09:37):
20-year-old tenure because theyknew, right?
They knew what needs to be done,they knew the process, they knew
everything back off their hand.
But suddenly you have arevolving door in manufacturing
and in skilled work.
New people are entering, olderguard is leaving, and you don't
(09:57):
stay there for 30 years anymore,maybe three years, and then you
want to do something else.
Now the time you take to onboardand then get this critical
knowledge is very short.
And if you don't have therequisite knowledge, this can be
a serious safety issue that canhappen.
So it can be someone's life, orit can be productivity, a
(10:20):
machine goes down, or you aregenerating a lot of scrap
because you have set upsomething incorrectly, or you
know, it could be a variety ofthese real impact.
So it is always about how thistechnology solves a particular
problem rather than this is theart of possible AI, and you can
(10:43):
go and solve global hunger, thatis when it creates the
confusion, that is when itcreates the fear.
When you are very specific aboutthe value and you can
demonstrate, then the adoptionhappens.
And that is what we find acrossour customers.
We focus on specific use casesthat are top of the mind for
them, measurable value, and thistechnology and AI is just a
(11:07):
consequence, which is going tohelp.
But it starts with the value,use case, and the people.
SPEAKER_02 (11:13):
So I'm I'm glad you bring the um the use cases up
because uh it's I I I'mwondering, like, okay, let me
just see if I can articulatethis.
So a lot it it's so competitiveright now.
Um supply chain management, verycompetitive space for lots of
(11:35):
different reasons.
But there's this overwhelming umsense of urgency to uh get it
done, whatever it whatever it isthat we need to get done,
including um developinginnovation, uh adopting it into
our practices.
But I'm wondering if you know,in this you know, quest to get
(11:56):
something done as soon aspossible, do uh do things like
do do you think we ever areforgetting the value that we're
creating on a global scale bymaking work environments just
more efficient and less tedious?
Are we are we kind ofoverarching from that initial um
(12:19):
goal with it all?
Are we making things morecomplicated?
I mean it sounds like you'vedistilled everything down into,
I don't want to say simpleterms, but you look at your
client, what's in front of you,and what their needs are, and
give them the specifics theyneed to make their staff
comfortable and so that they'reall more productive.
(12:41):
But are are the is the initialgoal you think are just as a
whole, do you think maybe someof us are kind of getting away
from that and we're just sofocused on developing the
innovation that we kind offorget what the initial goal was
in the first place?
SPEAKER_01 (12:58):
This is not new, right?
I've been in technology for twodecades and this has been uh
every technology transition,this has been a challenge.
Uh, do you go all in ontechnology and then assume that
use cases and the value willwork itself out?
Case in point, I go andimplement a large ERP and
everything will solve, or I'lldo a large cloud migration
(13:20):
project and then I'll deal withthe value later, or are you
going to start with the valueand work backwards?
And different situations demanddifferent approaches.
SPEAKER_03 (13:29):
Okay.
SPEAKER_01 (13:30):
Whether we are going too fast, um, is I always say
this when you adopt a technologythat's particularly a technology
that is bleeding edge, case inpoint generative AI.
Right?
There's values to be proven,there's a lot of excitement
about it, there's a lot of usecases, everyone comes in, you
know, be an evangelist trying totell you what are all the things
(13:52):
possible, etc.
Um, do you go all in onsomething like this or do you
wait?
What do you do?
The answer actually lies in twodimensions.
One dimension being is thisparticular use case
competitively differentiatingfor you?
Is it going to give you thecompetitive edge?
(14:15):
Then go and take the risk.
If not, wait for the technologyto play it out.
SPEAKER_02 (14:21):
Ah, okay.
SPEAKER_01 (14:22):
You don't, you you yeah, this is something that's
super critical for me to staycompetitive or to keep my people
safe, or it needs me to drivethe margin, then it's a
different story.
Then your risk appetite for afailed technology implementation
is different.
You will approach the problemdifferently, the change
management focus you will put onthat problem is different.
(14:47):
Versus, hey, you know what?
I need to find an applicationfor this modern technology, and
then you work backwards, thenyou are just playing the
technology for the technology'ssake, and you're not putting the
executive mind, the changemanagement focus, the process
changes that needs to happenwith it, and then naturally the
technology is going to fizzleout.
So if you are adopting if you'readapting anything new, it should
(15:11):
always be am I convinced thatthis is mission critical for me?
Then because technology isalways less than the 50% mark.
It's the change management, itis the adoption, it is the
executive focus, it is justmaking that happen.
It's in fact the technology partis the easiest part of all,
because there are vendors whomake it work for you.
(15:31):
It is the remaining 70%, theonly way you are going to put
your heart and soul into theremaining 70% is one leaf
everyone in the organization,particularly at the top,
convinced that this is a problemI want to solve today and right
now.
SPEAKER_02 (15:45):
And what I'm wondering though, that is that
if we're looking at, you know,okay, I want to uh adopt this
technology into my operationsbecause I want to deal with this
uh problem that's in front ofme.
I want to deal with this issue.
But because things change sofast, how do you assess whether
(16:10):
what you uh initially want totackle is still going to be
relevant in the short term?
Like because things change soquickly.
SPEAKER_01 (16:22):
It does, and and it is less problematic for us
because we are very specificfocused technology targeting a
problem rather than afoundational technology that can
solve that has a promise tosolve a lot of problems.
SPEAKER_02 (16:36):
Okay.
SPEAKER_01 (16:37):
Um so it is the the example that you are giving is
very, very relevant.
If I am like, hey, I'm going tobuild it, they will come.
So in in a technology that cansolve a lot of problems, yes,
this foundation is needed, I'mgoing to spend a year and a half
putting this largetransformation initiative and
build a platform and then seehow it goes, it's a very
(16:59):
high-risk play, right?
Yeah, for us, for us, it is alittle different.
In fact, one of the things thatyou know, um, when we when we
initially, when I joined thepitch deck that we showed in
front of our customers talkingabout deep how we talk about the
industry problem, etc.
One, two, three, four, five, tencustomers, they all said, We
know this, you're preaching thecore.
SPEAKER_03 (17:21):
Okay.
SPEAKER_01 (17:22):
So for us, the problem is so profound that
people are like, Yes, of courseI have labor shortage, of course
I have a skills gap.
Yeah, of course I have aging umworker population, of course,
people are leaving with tribalknowledge.
I have a larger onboard, longeronboarding, safety, blah blah
blah.
Easy, easy, easy.
Tell me the solution, right?
(17:43):
And that's not always the case.
You need to teach people on theproblem and then the solution
after, right?
So I'm lucky to be in anindustry where you know the
problems are well recognized andthe need for technology to solve
the problem is also wellrecognized.
Now it's the question of what isthe right technology, which is a
little different from generalpurpose platform technologies
(18:05):
where the risk you are callingout is particularly relevant.
SPEAKER_02 (18:08):
Okay.
So so now kind of um I'm lookmaybe talking about this in kind
of an opposite way.
So instead of focusing on umadoption and progression of AI
within a company, what aboutlooking at what happens when it
(18:31):
gets to a point where, hey, youknow what?
Maybe we need to put a pause onthis and give a look at uh how
things are going, assess thefeedback that we have, and then
rework the expectations and thedates for the AI integration.
(18:52):
Like, like, is it okay to put apause on things?
And how often have you foundthat you've had to do that
sooner than you expected, ormaybe not?
SPEAKER_01 (19:03):
You know, oftentimes the real life is more complex.
Uh right, there are alwayspeople who have made a decision
who is part of that, and it is aharder thing to stop rather than
to just go in and see throughit.
Yeah, so it it, you know, inreality it might be to a point,
and it does happen.
There are a lot of technologyinitiatives where you reach a
(19:25):
point where you need to say,hmm, it feels like we are not
ready for it.
And that is why a lot ofcustomers uh who are adopting
this technology take a more uhdeliberate view.
You know, I haven't been in a ina situation where an enterprise
customer has bought a technologywhere the rollout has been big
bang.
(19:46):
It is always pilot first.
SPEAKER_02 (19:49):
That's right.
SPEAKER_01 (19:50):
You start with a pilot, I know we could, you
know, the classic crawl, walk,and run, uh, or you controlled
scaling, there are differentterminologies that are used in
rolling out these technologies,but it is always the pilot, and
and that there have been a lotof cases where what you are
saying is true, which is youknow, when the pilot is you you
understand that this is toocomplex than what we thought to
(20:12):
be, and it dies at the pilotstage.
Any technology for that matter.
And to me, that's the rightapproach.
You start with a quick pilot,and you always, you know, it is,
you know, previously people usedto take their most complex use
case or a factory or whateverthat is, as the pilot use case,
which is actually a wrongstrategy, right?
(20:33):
You want to give the bestpossible chance for the
technology to succeed.
You want to actually give theyou know the most vanilla
situation for it to succeed sothat you can use that to
demonstrate value because halfof the success lies in people's
willing to change how they arethinking about the problem, how
they adapt the process.
(20:54):
So once you see the initialsuccess, it is, but not all
pilots are successful.
So that is when you do a pilot,you make sure that you you made
the right choice.
If yes, just keep going.
If not, cut your losses at thepilot stage.
SPEAKER_02 (21:10):
Okay, so let's talk about the the pilot where you
think you might have to cut yourlosses.
Because I I want to bringsomething up as far as how do
you have that conversation withuh the team that you're working
with in the pilot and you've putall this time and energy, not to
(21:31):
mention some finances in, andyou've been working collectively
towards something really greatand it doesn't work out.
How do you help the people thatyou've worked with on this
endeavor go from a like afailure mindset to a continuous
(21:54):
improvement mindset so you cankind of dust it off quickly and
move forward?
And stay positive?
SPEAKER_01 (22:02):
That's a good question, actually.
You know, you one of the one ofthe things with the reality is
when something is failing, thereis actually a collective
knowledge, people generallyassume that it is failing.
Yes, there are a few people whoare more personally invested in
it than others.
Um, right?
Yeah.
So uh which which makes itharder, but there's a general
sense.
(22:22):
There are two two actions onecan take, right?
You understand what is thereason why this is failing, and
because you are not ready, inwhich case you decide that these
are the things we want to bedoing to be ready before we
adopt this, or you say that thisis not for us, this is this this
(22:42):
solution is not for us, and inboth cases, the paramount thing
is communication, and it is notthe news that you are giving it
in one go.
SPEAKER_03 (22:54):
Okay, right?
SPEAKER_01 (22:55):
So it is it it is about constant communication,
constant way of managing andtracking the success KPIs, and
that is something that I youknow it is worth mentioning
here.
More many successful rollouts ortechnology, and particularly
evaluations, have had verythought-through scorecard on
(23:18):
what you're going to measure andhow you are going to measure on
what does success look like.
Once you have that, then you arenot communicating to someone
this is not successful.
Everyone sees it, right?
You see what I see, right?
We are expecting the KPA A, B,C, D to go in a particular
direction, and we are doing thispilot for a month, we extended
(23:40):
it for two months, now sixmonths.
It is not picking up.
What is the reason?
Oh, that's because it's not theright technology for you, or
maybe it is uh that we as anorganization is not ready, or we
don't have internal championswho are willing to change.
It could be any of this and allof those reasons.
Some reasons are morecorrectable than the others, and
some are not.
(24:01):
So it is the constantcommunication and facts,
fact-based decision making,those are critical there.
SPEAKER_02 (24:06):
And when you have these kind of conversations with
your clients, because it's aprocess, right?
Um how are there specificactionable steps that DPAW takes
in your response to the feedbackthat you receive as as the LLM
(24:29):
rolls out?
SPEAKER_01 (24:32):
The the large language models, particularly
the generative AI solutions areparticularly tricky because you
are talking about something thatis inherently extremely smart,
but at the same time it is not.
What do I mean by that?
(24:53):
If at the core of it, generativeAI or the large language models,
which has taken this world bystorm in the last year, is a
very eloquent speaker who knowsexactly what is the next word
and next sentence to use.
People who are very eloquentwith their language need not be
(25:15):
the smartest one in the room.
Right?
It is a general purposephilosopher who has a really
good opinion on a lot of things,and many times it may or may not
be true, that opinion.
Right now, the moment you thinkabout it that way, the nature of
generative AI solution, theapplication of the generative AI
(25:37):
solution changes.
I was talking to someone who'sbeen using this technology to
synthesize all the HRdocumentation and tells you what
the policy I mean, not what DeepHub does, but another situation.
(25:58):
Great, uh nice use case, right?
But I go in and ask a question,nine out of ten times it gives
you an accurate answer, that onetime it gives you a wrong
answer.
Because it is generating thenext word reading these
documents, it is not stating thefact, it's not expecting the
(26:19):
exact fact.
That could be problematic.
SPEAKER_02 (26:22):
I would say so.
SPEAKER_01 (26:24):
If you make your life decision based on that, or
if you think that this is whatyour insurance is going to
cover, then you go procedure,whatever that is, right?
Then there is a problem.
So there is this inherentunderstanding of the technology
that comes to play.
Uh, and there are there are alot of technology providers who
sell it for more than what itis, etc.
Right?
That's why the use cases, buttake another example.
(26:47):
Take an example of where you area customer service agent and you
are getting a call from acustomer who has called you 10
times before.
Okay.
Right?
Or you have hundred customerswho have called you on exactly
the same problem, which issolved by different people.
When the call happens andsomeone talks about the problem,
(27:08):
how nice will it be for an AIsolution to synthesize all this
information and put it in frontof your screen on what is
important for this customer andwhat is the solution that
actually previously worked forall the other customers,
auto-summarized in front of you.
How powerful it is becausethere's still a human involved
who is going to read it and thenget to the problem pretty
(27:30):
quickly.
And that is the kind of use casethat Depot has, right?
You are taking your real peopleworking on the shop floor.
We are capturing a video of whatyou are doing, and you are
converting your video into atraining content that is useful
using generative AI.
We are not creating trainingcontent out of nowhere, it is
(27:51):
your video that is synthesizedinto meaningful guidance,
snippets, question and answers,quizzes for you, which is
reliable, which is factual.
Right?
So then understanding thistechnology and then using it for
the use case where it is usefuland where it is dangerous is
(28:12):
super important.
And that is where I think theorganization needs to be careful
in terms of where to apply thetechnology and where not to
apply the technology.
And when there is so much noiseabout the large language models,
right, in the market, yeah, itis hard to distill those facts
(28:32):
because you are hearing onething, you know, you are hearing
about uh artificial generalintelligence, one day these
models are becoming conscious,becoming conscientious, and all
of those stuff.
Yes, great, cute, but you know,let us be honest.
SPEAKER_02 (28:50):
So I I I want to touch on something that um I
know when when generative AIfirst came out, there was a um
kind of a wave of panic thatwent through a lot of um
professionals and multipleindustries, you know, the the
fear of it, right?
(29:11):
And so I'm wondering um in yourmind, you know, when we're
working with suppliers andmanufacturers that are either in
the midst of an LLM process forthemselves or considering
integrating one into theiroperations, what do you think is
the best way to communicate thattransition of use to employees
(29:33):
without stoking that fear of jobloss?
How do you how do you how do youapproach that?
What what is the tone that needsto take place?
SPEAKER_01 (29:43):
I think it starts with the business case.
Um be it the generative AI or beit the previous generation of AI
company I was the U attentionwas predictive AI, right?
Where you are predicting stuffwith the AI model, the
traditional machine learningbased on data.
So be it doesn't matter which itis, it starts with the business
(30:03):
case.
If your business case is aboutheadcount reduction as one of
the payback, you can just assumethat it is going to die on the
line, right?
It's not it's not going to seethe daylight.
People are going to sense it,people are going to resist it,
and the technology is not goingto go anywhere, right?
So it is it starts with what isyour foundational business case,
(30:27):
number one.
And number two is being verytransparent and honest about it.
Which is yes, generative AI isonce in a generation technology,
right?
It is, you know, the AI has beenthe predictive AI has been
around for 20 years, it nevergot that kind of attraction and
attention.
But generative AI suddenly, whenChat GPT was introduced,
(30:51):
everyone started logging in, andthen what million users in an R
kind of adoption, right?
It's nothing that we have seenbefore.
Why?
Because it captured theimagination of everyone.
It said, yeah, this is somethingthat I see why this is magical.
So when when that happens,people start sensing things.
(31:11):
You know, people are smarterthan most of us speakers.
We are all smarter than mostpeople want to give credit to
us.
So it is important to betransparent, it is important to
discuss it, and it is importantto also acknowledge that there
is a displacement of skill setthat's going to happen with this
(31:31):
technology.
Whether this is going to makethe pie larger for everyone and
generate employment, I thinkthat is the truth.
I think that is what is going tohappen.
There are economic theoriesduring industrialization and
automation that proves that.
Right?
There are there are economictheories that are well
documented that says that whenfirst automation came during the
(31:55):
industrialization age, whichpeople thought they were going
to lose jobs, it actuallyincreased the job market
significantly.
Right?
Yeah.
So it is going to raise the pie,but it is going to have
displacements of the skill setthat's won't happen.
The particular area on supplychain manufacturing or the
physical work, that displacementis going to be lesser and lesser
(32:18):
and lesser.
Because this technology doesn'tdo physical work.
This technology is not meant todo physical work.
This technology, even what Depotprovides, is going to make the
life of people who are doing thephysical work better, more
efficient, and effective.
(32:41):
It's standardization.
Exactly.
It is going to make sure thatwhen you do something, when I do
something, when someone else whoturned up tomorrow for the job
does something, they all do itthe same thing.
SPEAKER_02 (32:54):
I can't tell you how many times I've been in
situations, you know, whetherI'm a contractor for a company
or I'm on staff, and especiallywhen I've just I'm onboarding
and you know I'm looking for SOPso I know how to do different
things, and they're nowhere tobe found, and nobody's available
(33:15):
to train you.
So um yeah, I I can definitelysee the um I'll say the peace of
mind, to say the least, that itprovides.
SPEAKER_01 (33:27):
And then you know, think about that.
Just take a moment to justextrapolate what you faced.
Yeah.
You're standing in front of amachine trying to make
something.
SPEAKER_02 (33:36):
Oh my that that is daunting.
SPEAKER_01 (33:40):
Exactly.
It is like me standing in mykitchen, don't know what to do.
I'm like, okay, where do Istart?
SPEAKER_02 (33:46):
Yeah, wow.
Wow.
That that definitely doesn'tmake, you know, I mean, we all
get, you know, have a little bitof nervousness when we start a
new job, but that would justmake it so much worse.
Um But um this has beenabsolutely delightful.
You um you've just made such anice positive take on um how we
(34:08):
can look at um integrating allthese new technologies and feel
good about it and you know takethe fear out of it and um you
know make it a game changer foreverybody.
Uh I really appreciate your timewith us today, Siva Lakshmanan,
CEO of Deep How.
If anyone wants to continuechatting with you, is there a
good way to connect with you?
SPEAKER_01 (34:30):
Yeah, and I'm available uh on LinkedIn.
Um Siva Kumar Lakshmanan is myname.
And if you want to drop an emailto me, CEO at uh deep how.com,
um, you know, I would be happyto engage in a conversation.
SPEAKER_02 (34:42):
Thank you so much.
Um, this has been great.
Really appreciate hanging outwith you today, and um I'll keep
in touch.
For um any other you know,tidbits of information that you
think you might want to know andhow to progress further in the
supply management profession,you will find what you're
looking for.
Just go to ismworld.org.
(35:04):
It's all there for you.
Thanks for tuning in.
I'm Melanie Stern for ISM.
This is Supply Chain Unfiltered, presented by the
Institute for Supply Management.
SPEAKER_02 (00:17):
Good to have you here today.
I'm Melanie Stern for ISM, anduh this episode of Supply Chain
Unfiltered is gonna be a funone.
Um it's actually uh talking witha company, um Deepal, that's
their name, first had um them onthe show probably about 18
months ago.
And um was actually, I think,probably one of my favorite
(00:39):
episodes.
It was just so fun to hear whatthis company was doing.
And um they had introduced alanguage learning model, and uh
it was changing the way thatglobal companies can scale and
and train their employeesbecause there's lots of
intricacies in communicationthat can go really well or go
(01:04):
haywire if you're working withpeople from different cultures,
different um places in theworld, and which can make
training really difficultsometimes.
So they had a great solution,and um if you wanted to check
out that episode where we firstchatted with them, I think it's
episode 38.
Nonetheless, fast forward totoday.
(01:26):
A lot has changed since my firstconversation with them.
So we're gonna get all caught uptoday.
Um, please welcome CEO ofDeepow, Steva Kumar Lakshmanan.
How are you, Steva?
SPEAKER_01 (01:40):
Very good, very good.
Thank you.
Thanks for having me.
SPEAKER_02 (01:42):
Awesome.
Yeah, um, I love what you guysdo.
So I think, in all fairness, thebest place to start would be if
you could just tell us a littlebit about yourself and Deepow's
mission for anyone that's um youknow isn't aware of of all the
great things that you do.
SPEAKER_01 (02:01):
Perfect.
Appreciate the opportunity totalk to you and to your
audience.
I am uh Sivakumar Lakshmanan.
I go by Siva.
I live in uh Dallas, Texas.
SPEAKER_02 (02:13):
Yeah.
SPEAKER_01 (02:14):
Um and Deepau, let me quickly give you a high-level
picture of what mission is andwhat we do.
Deepau is uh Deepau's mission isvery simple.
It is to empower skilledworkers.
These are people who buildthings that we use every day,
(02:35):
who manage, maintain things.
We often refer to that as coreindustries or skilled work,
things that need physical work.
And globally, and particularlyin the United States, there's a
shortage in availability ofskilled workers, particularly in
(02:57):
manufacturing.
If you go to a manufacturingindustry, you are typically
doing three shifts, runningthree shifts, people go in at
six o'clock, work eight to tenhours, and then the second
shift, and then the third shifthappens, and that's a very
strenuous work.
As the demographics change,people preferring to work in a
(03:20):
manufacturing setting is lowerand lower, and as a result, this
industry is going through alabor shortage.
And while that is happening,there is also an aging workforce
that is retiring, and thatknowledge, the tribal knowledge
of skilled work, is leaving withthese very experienced people
(03:43):
who have done this for 35, 40years, who knows this, who has
developed muscle memory of howto do these things.
They are never digitized, theyare never captured in a
consumable way.
And that is where default comesin, is to capture this knowledge
in a way that is consumable,that is immersive and useful for
(04:07):
the next generations of workerswalking into these factories or
these field services, takingover an absolutely economically
critical work that we need tomake everything that we use
today.
So Deep How makes this happen.
Deepow takes captures thisknowledge, converts this into a
(04:31):
know-how training, leveraginggenerative AI, also known as
large language models.
So this generative AI is the onethat captures this knowledge,
converts it into something very,very useful for people to
acquire, onboard themselves, useit, troubleshoot whatever they
(04:52):
want to do with.
SPEAKER_02 (04:55):
Okay.
Um, who wouldn't want that,right?
Um so so in this quest um thatyou have to basically help um
improve employer to employeerelationships, um, improved
interdepartmentalcommunications, you've got the
training for the labor force,and enhance the productivity
(05:18):
through through your languagelearning model.
I I would imagine that thecompany has hit some of the
goals that you meant to over thelast few years, but there were
probably, I'm guessing, somesurprises, some good and maybe
some not so good.
Um but would you mind sharing afew of those wins and and
(05:42):
learnings with us?
And the reason I ask is I knowum there's so many conversations
that take place about um AI andwho's doing what and what the
latest um in the advancementsare.
But I guess um I'll say, youknow, in comparison, we hear
(06:02):
about those stories, but wedon't tend to hear a lot about
what we find out once we rollout these new uh programs, these
new innovations, and um what howdoes testing go and what what
went right and what what needsimprovement.
So I'm wondering, can you sharesome of those with us?
SPEAKER_01 (06:22):
Absolutely.
Any any journey uh involves youknow wins and lose losses,
right?
And building a company um takesmore of those in a very shorter
period of time, and particularlyif you're dealing with AI, and
if you add a layer ofblue-collar work to it, then you
(06:44):
have a roller coaster ride.
And and which is which is whichis the exciting part about uh
this is that's the exciting partfor me, uh being the CEO at DPA.
What what are the what are thesome of the challenges you face?
You why do you why do you wantto go and work in a
manufacturing facility or dophysical work?
(07:05):
It's because you generally don'tenjoy disk chart, right?
You're not a fan of sitting infront of a computer writing
emails, etc.
I want to be out there, I wantto be on my feet, I want to use
my physical skills to achievesomething.
That's generally the careerinterest that draws people
towards this.
SPEAKER_03 (07:21):
Okay.
SPEAKER_01 (07:21):
Now you are suddenly introducing a technology to them
to say, hey, here is atechnology that's going to make
you be really good at what youdo.
Uh, what do you think is goingto be the response?
The response is um yeah, thankyou, but I don't think so.
Exactly.
I like the pillars and themachines in my hands.
Right there is there's a changemanagement, there's an adoption
(07:44):
part of it that's extremelychallenging.
So if you think AI ischallenging on how kids adopt AI
for education, or AI ischallenging, how are doctors
going to adopt?
This is not very different,right?
Uh it doesn't matter which isthe industry, some more complex
to adopt than the others.
So similarly, we see that inskilled work and manufacturing
as well.
(08:04):
How do people recognize where itcan add value and where it
doesn't?
And that is the mission that wehave undertaken at DePOW is to
distill the noise around this.
We always tell our customersthat it we think it is our
responsibility to identify whatis snake oil and what is real
(08:24):
value, and we want to distillout all the snake oil part of
AI.
Focus on the things that matter,things that are meaningful,
things that are going to addvalue.
Let me give an example.
Oh, please.
So if you if you take a if youtake a factory setting and
(08:44):
you're going to go in and workin a factory, there is a
procedure called lockout tagout.
SPEAKER_03 (08:49):
Right.
SPEAKER_01 (08:50):
Lockout tag out procedure is if I'm going to go
and work on maintaining amachine or cleaning a machine or
doing something with it, youneed to make sure that the
machine is switched off, right?
You don't want someone toaccidentally switch on the
machine.
It is like you going andswitching off the main switch at
your home and you're working onreplacing a bulb and your wife
is like, hey, why is there I'mnot getting a bar?
(09:11):
No, some idiot switched it offand then switched it on and
you're working on it.
That's fatal, right?
That's dangerous.
So the lockout tagout procedureactually makes sure that whoever
is doing this locking out isactually making sure it is
sealed, it is documented, and noone can accidentally switch it
back when you're working onthat.
Now, this was okay when everyonein your factory had a
(09:37):
20-year-old tenure because theyknew, right?
They knew what needs to be done,they knew the process, they knew
everything back off their hand.
But suddenly you have arevolving door in manufacturing
and in skilled work.
New people are entering, olderguard is leaving, and you don't
(09:57):
stay there for 30 years anymore,maybe three years, and then you
want to do something else.
Now the time you take to onboardand then get this critical
knowledge is very short.
And if you don't have therequisite knowledge, this can be
a serious safety issue that canhappen.
So it can be someone's life, orit can be productivity, a
(10:20):
machine goes down, or you aregenerating a lot of scrap
because you have set upsomething incorrectly, or you
know, it could be a variety ofthese real impact.
So it is always about how thistechnology solves a particular
problem rather than this is theart of possible AI, and you can
(10:43):
go and solve global hunger, thatis when it creates the
confusion, that is when itcreates the fear.
When you are very specific aboutthe value and you can
demonstrate, then the adoptionhappens.
And that is what we find acrossour customers.
We focus on specific use casesthat are top of the mind for
them, measurable value, and thistechnology and AI is just a
(11:07):
consequence, which is going tohelp.
But it starts with the value,use case, and the people.
SPEAKER_02 (11:13):
So I'm I'm glad you bring the um the use cases up
because uh it's I I I'mwondering, like, okay, let me
just see if I can articulatethis.
So a lot it it's so competitiveright now.
Um supply chain management, verycompetitive space for lots of
(11:35):
different reasons.
But there's this overwhelming umsense of urgency to uh get it
done, whatever it whatever it isthat we need to get done,
including um developinginnovation, uh adopting it into
our practices.
But I'm wondering if you know,in this you know, quest to get
(11:56):
something done as soon aspossible, do uh do things like
do do you think we ever areforgetting the value that we're
creating on a global scale bymaking work environments just
more efficient and less tedious?
Are we are we kind ofoverarching from that initial um
(12:19):
goal with it all?
Are we making things morecomplicated?
I mean it sounds like you'vedistilled everything down into,
I don't want to say simpleterms, but you look at your
client, what's in front of you,and what their needs are, and
give them the specifics theyneed to make their staff
comfortable and so that they'reall more productive.
(12:41):
But are are the is the initialgoal you think are just as a
whole, do you think maybe someof us are kind of getting away
from that and we're just sofocused on developing the
innovation that we kind offorget what the initial goal was
in the first place?
SPEAKER_01 (12:58):
This is not new, right?
I've been in technology for twodecades and this has been uh
every technology transition,this has been a challenge.
Uh, do you go all in ontechnology and then assume that
use cases and the value willwork itself out?
Case in point, I go andimplement a large ERP and
everything will solve, or I'lldo a large cloud migration
(13:20):
project and then I'll deal withthe value later, or are you
going to start with the valueand work backwards?
And different situations demanddifferent approaches.
SPEAKER_03 (13:29):
Okay.
SPEAKER_01 (13:30):
Whether we are going too fast, um, is I always say
this when you adopt a technologythat's particularly a technology
that is bleeding edge, case inpoint generative AI.
Right?
There's values to be proven,there's a lot of excitement
about it, there's a lot of usecases, everyone comes in, you
know, be an evangelist trying totell you what are all the things
(13:52):
possible, etc.
Um, do you go all in onsomething like this or do you
wait?
What do you do?
The answer actually lies in twodimensions.
One dimension being is thisparticular use case
competitively differentiatingfor you?
Is it going to give you thecompetitive edge?
(14:15):
Then go and take the risk.
If not, wait for the technologyto play it out.
SPEAKER_02 (14:21):
Ah, okay.
SPEAKER_01 (14:22):
You don't, you you yeah, this is something that's
super critical for me to staycompetitive or to keep my people
safe, or it needs me to drivethe margin, then it's a
different story.
Then your risk appetite for afailed technology implementation
is different.
You will approach the problemdifferently, the change
management focus you will put onthat problem is different.
(14:47):
Versus, hey, you know what?
I need to find an applicationfor this modern technology, and
then you work backwards, thenyou are just playing the
technology for the technology'ssake, and you're not putting the
executive mind, the changemanagement focus, the process
changes that needs to happenwith it, and then naturally the
technology is going to fizzleout.
So if you are adopting if you'readapting anything new, it should
(15:11):
always be am I convinced thatthis is mission critical for me?
Then because technology isalways less than the 50% mark.
It's the change management, itis the adoption, it is the
executive focus, it is justmaking that happen.
It's in fact the technology partis the easiest part of all,
because there are vendors whomake it work for you.
(15:31):
It is the remaining 70%, theonly way you are going to put
your heart and soul into theremaining 70% is one leaf
everyone in the organization,particularly at the top,
convinced that this is a problemI want to solve today and right
now.
SPEAKER_02 (15:45):
And what I'm wondering though, that is that
if we're looking at, you know,okay, I want to uh adopt this
technology into my operationsbecause I want to deal with this
uh problem that's in front ofme.
I want to deal with this issue.
But because things change sofast, how do you assess whether
(16:10):
what you uh initially want totackle is still going to be
relevant in the short term?
Like because things change soquickly.
SPEAKER_01 (16:22):
It does, and and it is less problematic for us
because we are very specificfocused technology targeting a
problem rather than afoundational technology that can
solve that has a promise tosolve a lot of problems.
SPEAKER_02 (16:36):
Okay.
SPEAKER_01 (16:37):
Um so it is the the example that you are giving is
very, very relevant.
If I am like, hey, I'm going tobuild it, they will come.
So in in a technology that cansolve a lot of problems, yes,
this foundation is needed, I'mgoing to spend a year and a half
putting this largetransformation initiative and
build a platform and then seehow it goes, it's a very
(16:59):
high-risk play, right?
Yeah, for us, for us, it is alittle different.
In fact, one of the things thatyou know, um, when we when we
initially, when I joined thepitch deck that we showed in
front of our customers talkingabout deep how we talk about the
industry problem, etc.
One, two, three, four, five, tencustomers, they all said, We
know this, you're preaching thecore.
SPEAKER_03 (17:21):
Okay.
SPEAKER_01 (17:22):
So for us, the problem is so profound that
people are like, Yes, of courseI have labor shortage, of course
I have a skills gap.
Yeah, of course I have aging umworker population, of course,
people are leaving with tribalknowledge.
I have a larger onboard, longeronboarding, safety, blah blah
blah.
Easy, easy, easy.
Tell me the solution, right?
(17:43):
And that's not always the case.
You need to teach people on theproblem and then the solution
after, right?
So I'm lucky to be in anindustry where you know the
problems are well recognized andthe need for technology to solve
the problem is also wellrecognized.
Now it's the question of what isthe right technology, which is a
little different from generalpurpose platform technologies
(18:05):
where the risk you are callingout is particularly relevant.
SPEAKER_02 (18:08):
Okay.
So so now kind of um I'm lookmaybe talking about this in kind
of an opposite way.
So instead of focusing on umadoption and progression of AI
within a company, what aboutlooking at what happens when it
(18:31):
gets to a point where, hey, youknow what?
Maybe we need to put a pause onthis and give a look at uh how
things are going, assess thefeedback that we have, and then
rework the expectations and thedates for the AI integration.
(18:52):
Like, like, is it okay to put apause on things?
And how often have you foundthat you've had to do that
sooner than you expected, ormaybe not?
SPEAKER_01 (19:03):
You know, oftentimes the real life is more complex.
Uh right, there are alwayspeople who have made a decision
who is part of that, and it is aharder thing to stop rather than
to just go in and see throughit.
Yeah, so it it, you know, inreality it might be to a point,
and it does happen.
There are a lot of technologyinitiatives where you reach a
(19:25):
point where you need to say,hmm, it feels like we are not
ready for it.
And that is why a lot ofcustomers uh who are adopting
this technology take a more uhdeliberate view.
You know, I haven't been in a ina situation where an enterprise
customer has bought a technologywhere the rollout has been big
bang.
(19:46):
It is always pilot first.
SPEAKER_02 (19:49):
That's right.
SPEAKER_01 (19:50):
You start with a pilot, I know we could, you
know, the classic crawl, walk,and run, uh, or you controlled
scaling, there are differentterminologies that are used in
rolling out these technologies,but it is always the pilot, and
and that there have been a lotof cases where what you are
saying is true, which is youknow, when the pilot is you you
understand that this is toocomplex than what we thought to
(20:12):
be, and it dies at the pilotstage.
Any technology for that matter.
And to me, that's the rightapproach.
You start with a quick pilot,and you always, you know, it is,
you know, previously people usedto take their most complex use
case or a factory or whateverthat is, as the pilot use case,
which is actually a wrongstrategy, right?
(20:33):
You want to give the bestpossible chance for the
technology to succeed.
You want to actually give theyou know the most vanilla
situation for it to succeed sothat you can use that to
demonstrate value because halfof the success lies in people's
willing to change how they arethinking about the problem, how
they adapt the process.
(20:54):
So once you see the initialsuccess, it is, but not all
pilots are successful.
So that is when you do a pilot,you make sure that you you made
the right choice.
If yes, just keep going.
If not, cut your losses at thepilot stage.
SPEAKER_02 (21:10):
Okay, so let's talk about the the pilot where you
think you might have to cut yourlosses.
Because I I want to bringsomething up as far as how do
you have that conversation withuh the team that you're working
with in the pilot and you've putall this time and energy, not to
(21:31):
mention some finances in, andyou've been working collectively
towards something really greatand it doesn't work out.
How do you help the people thatyou've worked with on this
endeavor go from a like afailure mindset to a continuous
(21:54):
improvement mindset so you cankind of dust it off quickly and
move forward?
And stay positive?
SPEAKER_01 (22:02):
That's a good question, actually.
You know, you one of the one ofthe things with the reality is
when something is failing, thereis actually a collective
knowledge, people generallyassume that it is failing.
Yes, there are a few people whoare more personally invested in
it than others.
Um, right?
Yeah.
So uh which which makes itharder, but there's a general
sense.
(22:22):
There are two two actions onecan take, right?
You understand what is thereason why this is failing, and
because you are not ready, inwhich case you decide that these
are the things we want to bedoing to be ready before we
adopt this, or you say that thisis not for us, this is this this
(22:42):
solution is not for us, and inboth cases, the paramount thing
is communication, and it is notthe news that you are giving it
in one go.
SPEAKER_03 (22:54):
Okay, right?
SPEAKER_01 (22:55):
So it is it it is about constant communication,
constant way of managing andtracking the success KPIs, and
that is something that I youknow it is worth mentioning
here.
More many successful rollouts ortechnology, and particularly
evaluations, have had verythought-through scorecard on
(23:18):
what you're going to measure andhow you are going to measure on
what does success look like.
Once you have that, then you arenot communicating to someone
this is not successful.
Everyone sees it, right?
You see what I see, right?
We are expecting the KPA A, B,C, D to go in a particular
direction, and we are doing thispilot for a month, we extended
(23:40):
it for two months, now sixmonths.
It is not picking up.
What is the reason?
Oh, that's because it's not theright technology for you, or
maybe it is uh that we as anorganization is not ready, or we
don't have internal championswho are willing to change.
It could be any of this and allof those reasons.
Some reasons are morecorrectable than the others, and
some are not.
(24:01):
So it is the constantcommunication and facts,
fact-based decision making,those are critical there.
SPEAKER_02 (24:06):
And when you have these kind of conversations with
your clients, because it's aprocess, right?
Um how are there specificactionable steps that DPAW takes
in your response to the feedbackthat you receive as as the LLM
(24:29):
rolls out?
SPEAKER_01 (24:32):
The the large language models, particularly
the generative AI solutions areparticularly tricky because you
are talking about something thatis inherently extremely smart,
but at the same time it is not.
What do I mean by that?
(24:53):
If at the core of it, generativeAI or the large language models,
which has taken this world bystorm in the last year, is a
very eloquent speaker who knowsexactly what is the next word
and next sentence to use.
People who are very eloquentwith their language need not be
(25:15):
the smartest one in the room.
Right?
It is a general purposephilosopher who has a really
good opinion on a lot of things,and many times it may or may not
be true, that opinion.
Right now, the moment you thinkabout it that way, the nature of
generative AI solution, theapplication of the generative AI
(25:37):
solution changes.
I was talking to someone who'sbeen using this technology to
synthesize all the HRdocumentation and tells you what
the policy I mean, not what DeepHub does, but another situation.
(25:58):
Great, uh nice use case, right?
But I go in and ask a question,nine out of ten times it gives
you an accurate answer, that onetime it gives you a wrong
answer.
Because it is generating thenext word reading these
documents, it is not stating thefact, it's not expecting the
(26:19):
exact fact.
That could be problematic.
SPEAKER_02 (26:22):
I would say so.
SPEAKER_01 (26:24):
If you make your life decision based on that, or
if you think that this is whatyour insurance is going to
cover, then you go procedure,whatever that is, right?
Then there is a problem.
So there is this inherentunderstanding of the technology
that comes to play.
Uh, and there are there are alot of technology providers who
sell it for more than what itis, etc.
Right?
That's why the use cases, buttake another example.
(26:47):
Take an example of where you area customer service agent and you
are getting a call from acustomer who has called you 10
times before.
Okay.
Right?
Or you have hundred customerswho have called you on exactly
the same problem, which issolved by different people.
When the call happens andsomeone talks about the problem,
(27:08):
how nice will it be for an AIsolution to synthesize all this
information and put it in frontof your screen on what is
important for this customer andwhat is the solution that
actually previously worked forall the other customers,
auto-summarized in front of you.
How powerful it is becausethere's still a human involved
who is going to read it and thenget to the problem pretty
(27:30):
quickly.
And that is the kind of use casethat Depot has, right?
You are taking your real peopleworking on the shop floor.
We are capturing a video of whatyou are doing, and you are
converting your video into atraining content that is useful
using generative AI.
We are not creating trainingcontent out of nowhere, it is
(27:51):
your video that is synthesizedinto meaningful guidance,
snippets, question and answers,quizzes for you, which is
reliable, which is factual.
Right?
So then understanding thistechnology and then using it for
the use case where it is usefuland where it is dangerous is
(28:12):
super important.
And that is where I think theorganization needs to be careful
in terms of where to apply thetechnology and where not to
apply the technology.
And when there is so much noiseabout the large language models,
right, in the market, yeah, itis hard to distill those facts
(28:32):
because you are hearing onething, you know, you are hearing
about uh artificial generalintelligence, one day these
models are becoming conscious,becoming conscientious, and all
of those stuff.
Yes, great, cute, but you know,let us be honest.
SPEAKER_02 (28:50):
So I I I want to touch on something that um I
know when when generative AIfirst came out, there was a um
kind of a wave of panic thatwent through a lot of um
professionals and multipleindustries, you know, the the
fear of it, right?
(29:11):
And so I'm wondering um in yourmind, you know, when we're
working with suppliers andmanufacturers that are either in
the midst of an LLM process forthemselves or considering
integrating one into theiroperations, what do you think is
the best way to communicate thattransition of use to employees
(29:33):
without stoking that fear of jobloss?
How do you how do you how do youapproach that?
What what is the tone that needsto take place?
SPEAKER_01 (29:43):
I think it starts with the business case.
Um be it the generative AI or beit the previous generation of AI
company I was the U attentionwas predictive AI, right?
Where you are predicting stuffwith the AI model, the
traditional machine learningbased on data.
So be it doesn't matter which itis, it starts with the business
(30:03):
case.
If your business case is aboutheadcount reduction as one of
the payback, you can just assumethat it is going to die on the
line, right?
It's not it's not going to seethe daylight.
People are going to sense it,people are going to resist it,
and the technology is not goingto go anywhere, right?
So it is it starts with what isyour foundational business case,
(30:27):
number one.
And number two is being verytransparent and honest about it.
Which is yes, generative AI isonce in a generation technology,
right?
It is, you know, the AI has beenthe predictive AI has been
around for 20 years, it nevergot that kind of attraction and
attention.
But generative AI suddenly, whenChat GPT was introduced,
(30:51):
everyone started logging in, andthen what million users in an R
kind of adoption, right?
It's nothing that we have seenbefore.
Why?
Because it captured theimagination of everyone.
It said, yeah, this is somethingthat I see why this is magical.
So when when that happens,people start sensing things.
(31:11):
You know, people are smarterthan most of us speakers.
We are all smarter than mostpeople want to give credit to
us.
So it is important to betransparent, it is important to
discuss it, and it is importantto also acknowledge that there
is a displacement of skill setthat's going to happen with this
(31:31):
technology.
Whether this is going to makethe pie larger for everyone and
generate employment, I thinkthat is the truth.
I think that is what is going tohappen.
There are economic theoriesduring industrialization and
automation that proves that.
Right?
There are there are economictheories that are well
documented that says that whenfirst automation came during the
(31:55):
industrialization age, whichpeople thought they were going
to lose jobs, it actuallyincreased the job market
significantly.
Right?
Yeah.
So it is going to raise the pie,but it is going to have
displacements of the skill setthat's won't happen.
The particular area on supplychain manufacturing or the
physical work, that displacementis going to be lesser and lesser
(32:18):
and lesser.
Because this technology doesn'tdo physical work.
This technology is not meant todo physical work.
This technology, even what Depotprovides, is going to make the
life of people who are doing thephysical work better, more
efficient, and effective.
(32:41):
It's standardization.
Exactly.
It is going to make sure thatwhen you do something, when I do
something, when someone else whoturned up tomorrow for the job
does something, they all do itthe same thing.
SPEAKER_02 (32:54):
I can't tell you how many times I've been in
situations, you know, whetherI'm a contractor for a company
or I'm on staff, and especiallywhen I've just I'm onboarding
and you know I'm looking for SOPso I know how to do different
things, and they're nowhere tobe found, and nobody's available
(33:15):
to train you.
So um yeah, I I can definitelysee the um I'll say the peace of
mind, to say the least, that itprovides.
SPEAKER_01 (33:27):
And then you know, think about that.
Just take a moment to justextrapolate what you faced.
Yeah.
You're standing in front of amachine trying to make
something.
SPEAKER_02 (33:36):
Oh my that that is daunting.
SPEAKER_01 (33:40):
Exactly.
It is like me standing in mykitchen, don't know what to do.
I'm like, okay, where do Istart?
SPEAKER_02 (33:46):
Yeah, wow.
Wow.
That that definitely doesn'tmake, you know, I mean, we all
get, you know, have a little bitof nervousness when we start a
new job, but that would justmake it so much worse.
Um But um this has beenabsolutely delightful.
You um you've just made such anice positive take on um how we
(34:08):
can look at um integrating allthese new technologies and feel
good about it and you know takethe fear out of it and um you
know make it a game changer foreverybody.
Uh I really appreciate your timewith us today, Siva Lakshmanan,
CEO of Deep How.
If anyone wants to continuechatting with you, is there a
good way to connect with you?
SPEAKER_01 (34:30):
Yeah, and I'm available uh on LinkedIn.
Um Siva Kumar Lakshmanan is myname.
And if you want to drop an emailto me, CEO at uh deep how.com,
um, you know, I would be happyto engage in a conversation.
SPEAKER_02 (34:42):
Thank you so much.
Um, this has been great.
Really appreciate hanging outwith you today, and um I'll keep
in touch.
For um any other you know,tidbits of information that you
think you might want to know andhow to progress further in the
supply management profession,you will find what you're
looking for.
Just go to ismworld.org.
(35:04):
It's all there for you.
Thanks for tuning in.
I'm Melanie Stern for ISM.
Popular Podcasts
Hey Jonas!
Hey Jonas! The official Jonas Brothers podcast. Hosted by Kevin, Joe, and Nick Jonas. It’s the Jonas Brothers you know... musicians, actors, and well, yes, brothers. Now, they’re sharing another side of themselves in the playful, intimate, and irreverent way only they can. Spend time with the Jonas Brothers here and stay a little bit longer for deep conversations like never before.
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