January 9, 2025 • 35 mins
Ever wondered how empathy might be the missing link between cutting-edge technology and real-world business needs? Join us as we chat with Vic Miles, the General Manager for Industry Strategies in the Americas at Microsoft, who brings a wealth of experience from his 15-year journey at the tech giant, including his impactful work with Walmart. Vic offers a fresh perspective on how understanding the end user's experience, even through simple tasks like running a cash register, can lead to groundbreaking tech solutions that resonate with the human side of industries. His insights promise to deepen your understanding of how empathy can drive innovation in the tech landscape.
Unlock the secrets of transforming the workplace through AI as Vic explains how automation is changing the game for businesses, freeing up employees for more meaningful roles. Discover how AI is challenging traditional business processes by taking over mundane tasks and allowing seasoned professionals to focus on high-impact activities. As we transition from chatbots to fully autonomous systems, Vic shares how AI is not just about efficiency but also about redefining productivity and what it means for the future workforce.
In the ever-evolving field of AI, prompt engineering emerges as a critical skill. Learn about the art of crafting precise questions to guide AI systems, ensuring they understand context and produce accurate responses. With human oversight still playing a vital role, Vic discusses the importance of balancing technology with human judgment to avoid errors. Plus, hear about the exciting ways AI is enhancing communication in collaborative environments, bridging gaps between businesses, and even smoothing workflows for small companies with innovations like AI-enhanced QuickBooks. This episode is a treasure trove of insights into the ongoing AI revolution and its transformative impact on industries today.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Welcome to the 4 Bars podcast.
I'm Ken Leith and I'm PattiLeith.
We're your hosts for somecompelling dialogue, encouraging
our listeners to strengthentheir connections and build
strong communities, lifting eachother up and connecting in ways
that matter.
We named the podcast 4 Bars asa reference to how hard we work
to find a 4 Bars connection onour devices.
Speaker 2 (00:19):
And we wondered what could happen with relationships
if we worked as hard atconnecting.
Speaker 1 (00:24):
Let's find out.
Welcome back to Four Bars, thepodcast.
We're excited to be here.
I have my co-host with me, kenLeith, and we are excited to
host a guest who will bespeaking to us today about AI,
vic Miles.
Vic is the general manager forindustry strategies in America,
so welcome, vic, it's great tosee you.
Thank you for having me Tell usa little bit about yourself.
(00:46):
What got you to this point?
Speaker 3 (00:56):
Yeah, so I've been 15 years with Microsoft.
As you said, I'm the generalmanager of our industry business
in the Americas.
We look after our moststrategic customers in the
Americas North and South and wehelp them sort of unpack how to
use technology to meet theirbusiness needs.
So we sit at the intersectionbetween the business imperatives
and the right technologysolution.
(01:16):
I have a team of folks about 20people or so who engage in
marketing and we at Microsoftlike to hire out of the industry
right.
So my background I came fromindustry.
I think of myself as a retailer.
I've worked for Walmart storesfor 10 years and did many things
(01:36):
in that business.
I was on the IT side, so Istarted out as a software
developer and engineer.
Worked in the operation side ofthe business, so point of sale,
cash registers, time clocks andpayroll and all that stuff in
the store.
So I really have a heart forwhat we call them as frontline
associates who work in the storeand have just built a career
(01:59):
around this intersection betweenthe needs of consumers.
I find that I'm really tunedinto what we call the market.
What do consumers want and howdo retailers and service
providers, whether they'rehealthcare or anything how do we
serve that need with the rightbalance of profit and all of the
(02:23):
rest.
So it's a great position.
I I've enjoyed it now as, as Imentioned, for 15 years, and so
now it feels like a constantevolution, right.
We probably know two years havebeen the same right, or
certainly no more than two yearshave been the same, and it's
just constant change.
And so again, excited to behere.
Speaker 1 (02:44):
Well, it's great to have you.
Great to have you.
Thanks for sharing that.
As you think about yourbackground and how you've put
together your experiences, whatdid you learn early that really
helps drive you today?
Speaker 3 (03:05):
I learned that and you and I met years ago.
Yes, we did, and we all havethese sort of moments of growth,
these spurts of growth in ourcareer, and you and I were
meeting and I think you helpedme grow into a more leadership
position right from management.
(03:25):
What I, what I, learned earlyon was you have to, you have to
get on the ground right, andit's a growing up in the culture
of Sam Walton, it was about you.
You really couldn't have a goodunderstanding of the business
unless you were in the field.
Yes, right, and you gotta go tothe field.
So you know, in the field, yes,right, and you've got to go to
the field.
So you know, in my tenure I usedto run a cash register on
(03:47):
Saturday because my team and Ideveloped the solution that you
know, 300,000 cashiers would useevery day.
And just to gain empathy, right, I used to go on the Saturday
it wasn't my idea to be fair inthe beginning and run a register
and you haven't lived until youcan identify a uh, it's true,
(04:07):
uh, produce item yeah, right,and a mother with with three
children is there and you'retrying to get through and you
start to have empathy for so Iwould tell you I learned empathy
for, uh, how the systems thatwe develop, how they impact the
end user, yes, and so I bringthat forward into, uh, my, my
current career.
And so a lot of times we'll sayshow me, I'm from Missouri as
(04:29):
well, so it's the show me state.
We'll say, show me this thingthat you're talking about, can
we see it?
Can we do a plant tour?
Can we get on the floor of astore?
That kind of thing.
Speaker 1 (04:37):
There is no replacement for that kind of
experience and, similar to you,I started with Food Lion and
learned the business from theground up and just spent time on
the floor before going intomanagement.
That was a requirement for themand I still draw back on that
experience on a regular basis.
I agree Well, I discovered yourpassion for this topic as we
(05:01):
reconnected, but also in yourown podcast.
Speaker 3 (05:04):
Can you?
Speaker 1 (05:05):
tell us a little bit about your podcast so, uh,
conversations on retail.
Speaker 3 (05:08):
It's a.
It's a youtube-based podcast.
Um, a gentleman by the name ofmatt pfeiffer had run sort of
the program and he has hoststhere, and so I do a show called
beyond the tech, and so I'mvery interested in so the tech
side.
I live and breathe that daily.
But I also get in my role anopportunity to meet fascinating
(05:29):
people Right, who are, who aredoing things that
technologically make the press,but then they themselves are
fascinating, yeah, right, it'slike oh, you're you do marathon
running, yes, or you like tohike, or you're a world-class.
What is it the in the Olympics?
We're the Curly.
Speaker 2 (05:48):
Yes, Curly.
Speaker 3 (05:49):
How did you know that kind of thing, and so I invite
guests onto the show to talkabout their background and how
again that intersection withtechnology so what makes them a
whole person?
Speaker 1 (06:01):
outside of this Exactly Love it.
We've watched a few episodesand I highly recommend listeners
that you check it out.
We're here to talk about AI, ofcourse, so I'm interested in
learning about the journey thatMicrosoft has been on with AI.
Speaker 3 (06:19):
Yes, and it's interesting.
I didn't sort of start out evenin my own podcast being an AI
sort of evangelist or aneducator, but it is the topic du
jour, everybody wants to talkabout it, and so I've had my own
journey in understanding it,and we're on the cusp of yet
(06:39):
another shift in AI and sohopefully we will unpack that
today as we talk.
But artificial intelligence,really, I like to start with a
primer by saying it's not new.
So artificial intelligence hasbeen around really since the
late 50s and it didn't look thesame.
(07:00):
It was just, you know, itstarted as machine learning
trying to find patterns, nowwhat you do with the pattern
back to you, human right, thatkind of thing.
And now we're we're enteringinto this space where um
companies are very excited aboutbringing expertise to their
employees and the folks thatwork in the business, um, being
able to take a uh, a four yearassociate and give them the same
(07:23):
expertise as your 16-yearassociate, so that they can both
have the same level of whetherit's service or performance in
the business, and that's supercompelling.
There's dollars involved as well.
Depending on which source youplug into, whether it's Goldman
Sachs or McKinsey or IDC, it'sbetween $2 and $19 trillion of
(07:48):
GDP impact.
Wow, $19 trillion, I mean, it'sa pretty wide range.
Let's see.
I think IDC has it at $14 andMcKinsey had it at $2 to $4
trillion, and so that is what'sdriving companies to say I want
in.
And there's cautious optimism,right, as we think about the
(08:11):
impact to jobs, right.
So I always say I don't want tolive in a community and neither
do you or the audience eitherthat has 40% unemployment.
That's not what we're here for,right.
So you know when the machinescan do everything better, and so
it's about finding.
You know, as the audience andyourselves think about AI, it's
about finding where are thefriction points.
(08:32):
What are the things thatirritate you about engaging with
brands?
You know, in my space, theretail and consumer goods that
irritate you about engaging withbrands or that keep you from
having the best experience, andcan we use AI to smooth that out
?
Right, I'm trying to replace abunch of people, but if there
are some areas, some seams thatwe can sew up, yes, yes,
(08:55):
interesting.
Speaker 2 (08:56):
Yeah, I mean and to your point, it's a topic
everywhere.
We do not go into a client'sfacility over the last several
years and that's not coming upin a discussion, and everyone's
strategic roadmap has an AIcomponent to it now, whereas
three years ago that wasn't thecase.
It's kind of just being talkedabout on the peripheral.
But if you think about it now,where companies are starting to
(09:17):
embrace it and where they'restill hesitant, where do you see
those junctures as far aswhere's that embracing and where
is their hesitancy stillthere's?
Speaker 3 (09:26):
again being in the retail space when you're working
with a business withsingle-digit margins right, the
hesitancy comes from there's alot of cost here, right, so so
we talk about AI being areasoning engine on the back end
(09:48):
, you know, and then a largeamount of data that it reasons
over, and then an interfaceRight, and so we know that
interface largely today as aconversational interface.
For Microsoft, we branded itco-pilot Right, so it's a
co-pilot, you're the pilot andthis is here to help you.
The hesitancy is is with cost,because that reasoning engine
really does consume resources,compute resources, and it costs
(10:09):
money.
And so we see people say holdon, where's the value?
And so now we're seeing peopleshift into I like the idea of it
.
What's the roi?
Right, the?
The uptake that we've seen is in, uh, generally, content
creation.
Um, so, companies, who, who,perhaps, uh, who sell cars,
vehicles, especially usedvehicles, where each one is
(10:31):
different, right, and you haveto create copy material that
says here is a Ford truck that'sblue, it's got one den on the
side, it's got 60,000 miles, orwhatever.
That can take days to create.
Right, you have to go see thevehicle, that sort of thing.
What all is relevant to theconsumer and having AI
understand the relevancy of theconsumer.
(10:52):
So AI is able to take who isyour target audience right by
locale.
So my target audience is SanDiego, california and the 150
mile radius.
Okay, I now know somethingabout this, this, this
demographic and then so on.
Right, am I going for theprofessional person to use it as
work and it will then createcopy that is relevant to that
(11:15):
audience right In seconds.
Right, and so we're seeinglarge uptake there.
Call centers you know being atthat front end and we've
experienced IVRs right whenthey're sort of trying to route
you through.
But having a conversation in acall center that says why are
you calling?
Oh, have you tried thisMeanwhile.
Having a conversation in the ina call center that says why are
you calling?
Oh, have you tried this?
Meanwhile, you're in the queuebut I'm going to converse with
(11:36):
you, that sort of thing.
So we've seen uptake on thecontent side, but hesitancy on
the cost side yeah, understand,understand, absolutely for many
people they're they're afraid ofit.
Speaker 1 (11:46):
They're afraid of what it will do to the future.
Can you speak to that a littlebit?
Speaker 3 (11:50):
Yeah, you know, and it's if I can do some small
measure of assuaging people'sconcerns around AI.
I think that in the in the US,about 75 percent of our GDP is
in the service industry, right,that's, that's health care,
retail, hospitality, that sortof thing, and so those hands-on
(12:13):
businesses.
I don't see a lot changing,right?
So, again, ai that knows you inthe hotel room and presents a
particular menu on the TV basedon your preferences, you still
need, you know, housekeeping,you still need front desk, you
still need all of the folks thatcreate the rest of the
experience, and so a lot offolks won't.
(12:34):
I would think, for folks likemyself and the people that you
work with the professionals onthe back end, right, that says,
hey, am I really adding value oram I just part of the value
chain?
I'm in the process, right?
I don't know that.
I have the same message for them, right, and I think the answer
is get to value, right, get towhere you're creating value,
(12:56):
adding value, because if you'reworking a, a let's call it an
inbound queue of invoices thatneed to be cleared, right, right
, and and the, your goal isclear the invoices every day,
right, and and, with a, withsome threshold, at least 100
plus empty, you know, within 10by the time you leave, kind of
thing.
Um, ai will do a lot of thatbetter than people.
(13:19):
The good news is, though, is itfrees you up.
So start to understand why am Iclearing invoices, right?
What is the value that thisprocess has?
And then I can go add valuesomewhere else in the value
chain, but if I only know this,the rope process of clearing
invoices, that that's going tochange, yeah, yeah.
Speaker 1 (13:38):
One of the fundamentals we work with
executives on building theirstrategic capacity is to let
someone or something else helpyou do that, Because then it
frees you up to really do thethings that your intellect and
your training and background canbenefit.
The organization you know Italked about it's into that, I
(13:59):
think.
Speaker 3 (14:00):
I talked about the tenured associate, the 16-year
associate and the four-yearassociate, right, I think, one
of our findings?
We looked at some studies.
We found that the 16 yearassociate has aspiration to do
more, but they are the glue thatholds this process together.
Yes, right, it takes that levelof tenure to keep this process
(14:20):
moving.
Very right, and so, if we canremove that, you're now free to
go explore your own passions.
Yeah, right, I didn't reallyintend to be here clearing
invoices for 16 years, but Iunderstand it front and back.
There's nothing I haven't seen.
I am critical to the business,but my real passion is I want to
lead X department, and so nowthis frees you up to go do that.
(14:44):
So we do see school teachers.
Do they really want to gradeall the papers at night?
Speaker 1 (14:53):
teachers right.
Speaker 3 (14:53):
Do they really want to grade all the papers at night
?
We think not right, and sobeing able to sort of reduce
that, I think that the qualityof life will will raise.
There's a lot of folks that say, um, uh, the most uh famous
person that says bill gates, uh,founder, says that you know,
people work will look verydifferent in the future and for
many people they won't have towork.
I don't know what that lookslike yet.
Yes, but that's the future thatwe're going to, is there'll be
(15:14):
enough value created that thatthat 19 trillion right.
As long as we don't sort of,you know, in a social construct,
let that consolidate at the top.
Speaker 2 (15:25):
I think all of our lives keep it, yeah yeah, yeah,
yeah, and it's exciting andscary at the same time when it
comes out.
But, as you were talking there,one of the things that hit me
part of it really can be aboutit's delegating those tasks that
really are not tasks that arerelational.
So it's taking those thingsthat just need to get done at
(15:47):
someone's desk and primarilyalone, potentially where they do
those things.
Ai today, as we see it, andwhat you're defining is more of
I can take that off your plate,delegate it over here.
Speaker 3 (16:04):
You go, do something more relational something that
has higher value to theorganization.
Correct, indeed?
Yes, exactly, and maybe now's agood time to sort of talk about
the fork that is occurring inAI market right now.
We're about two years in, let'ssay, to the big bang of AI.
We began with conversational AIyes, Right.
So chatbots yes, right, very,very intelligent chatbots.
(16:27):
I don't know if you've used it.
Speaker 2 (16:28):
I have yeah.
Speaker 3 (16:30):
You know, when you no longer have to search, right.
So I wanted to search.
What is it?
What is the procedure forgetting a new driver's license
in Arkansas, right and so, and Icould go to the Department of
Finance Administration, the DFAside, and then elite, read and
get the PDF, and and that, but.
But an AI agent told me, oh,you need these four documents,
(16:51):
you have to come here.
Most offices closed by four.
Told me exactly what I wantedto know.
Yeah, in in seconds.
So that's the conversational ai.
Okay, and ken, what you talkedabout is this this, what I like
to call units of work?
Right, these tasks?
Yes, um, we're moving into this, this notion of, uh, agentbased
AI.
They call it agentic.
(17:11):
So agentic AI is agent-based AI, and agents will autonomously,
so without prompting, do unitsof work.
And so when you think about theinterconnection between
departments in a business, right, so I'll go back to my invoice
thing.
(17:32):
So the, the invoices come in,they first get before they go to
finance to be paid.
They have to be clear did,where were all the numbers right
?
Was the inventory actuallymoved?
That kind of thing?
And that inventory movementcomes from a different
department.
Yes, and and and and then wemove it over into finance and
they pay it.
And that's how we used to doprocesses right and build
corporations.
(17:52):
And now we're saying well, thedata of the inbound invoice and
what was received is availableto the agent, the data about
what was ordered is available tothe agent and the data about
the compliance aspects of payingand finance is available to the
agent.
Compliance aspects of payingand finance is available to the
agent.
So, bringing those threetogether, the agent can just do
(18:14):
the process without prompting orintervention.
So we're moving from human inthe loop, right, human in the
loop to no human in the loop,and so again.
But if you're thinking aboutfriction, if your job is
clearing the friction meaning Iwork on invoices all day your
job's going to change.
But there are those two thingsand I would invite the audience
(18:36):
to really think about in theirown businesses.
Right, what are those units ofwork?
And you call them tasks, butit's actually even bigger
because we had three tasks there.
Yes, right.
Speaker 1 (19:02):
And it's really a unit of work trying to just pay
the invoice, and so, yeah, it'svery interesting.
You can almost get more out ofthat than you can out of any
other type of searching thesedays, but there are some
limitations that you all havefound, which has prompted your
journey into it.
Speaker 3 (19:19):
Jente Can you speak to that a little bit.
Yes, exactly so.
The first limitation and Kenmentioned it is the hesitancy
around the cost.
Yes, right.
So executives, boards aresaying what does it mean to make
my employee 20% more productive?
Yeah, right, is that a longerbreak?
Or do we get 20% more work done?
(19:40):
Right, are we changing thebusiness process?
So that queue of invoices isthere going to be 20% more?
Do I get rid of people?
They want to know where is thesavings?
I see the cost side.
Yes, percent more.
Do I get rid of people?
They want to know where is thesavings?
See the cost side.
Yes, and so the ROI has beenone of the challenges with that
is we'll we'll?
Yes, we understand people canseek information and find
information.
(20:01):
There is this notion ofprompting.
There is there's a new sort ofengineering role in the tech
space called a prompt engineer.
So before we used to, welearned how to talk to the
computers, what I call machinespeak.
Right, so, as a developer,they're talking to a computer
all day, right through alanguage.
(20:22):
And now this prompt engineersays knowing the data I want, I
want to actually be able to askthe question in a well-formed
way that gives me the answer Iwant.
So you you ask what were thechallenges we found?
Depending on how you ask thequestion, the machine may take a
different context yes, rightand give you a different answer
(20:44):
Right, and so therein lies thechallenge.
Hold on.
We don't want to just unleashthis powerful tool and someone
says, you know, should I ordermore?
And the machine says, well,absolutely, sure, order up.
Speaker 1 (20:57):
Yeah, right, right, right, yeah, never mind Never.
Speaker 3 (20:58):
Yeah, right, right, right, yeah, Never mind that
there's only two weeks left inthe season, and you know, and
the cost changed and we don'thave, yeah, the supply chain
isn't there, and so that's been,the hesitancy is hold on, we
don't know how to create prompts, broadly speaking.
Right, we don't know how tocreate prompts, and is the model
(21:20):
we call the, the model answersthe language model, does it have
the proper context?
And so that's a that's a bigchallenge.
We're still working on that yeah, yeah so agentic is is sort of
the new uh way that says givingit, given a process that happens
every day and has to happenevery day, that's, we can do
that without thinking yeah, Ilove that element.
Speaker 2 (21:42):
Yeah, I think the other thing about that.
So I mean we're really talkingabout shifting and almost
becoming bilingual with that, Imean with the technology and
then the natural language thatpeople utilize.
Yes, when we think about that,even knowing language, and then
there's syntax, there's theability to understand
behaviorally uh, how should Ianswer a question?
(22:04):
Does ai, do you see thatevolving to another level,
potentially with agents to beable to kind of pick up on some
other nuances and languages?
Speaker 3 (22:12):
yes, I do see it evolving and it's it's already
there to some degree.
It's about harnessing it.
Okay, right.
So, not knowing that, by askinga question a certain way,
you've implied a, a bias, not ina bad way, but maybe your bias
is I only have 20 minutes, whatshould I cook for dinner Right?
Versus I'm trying to impress myboss and 15 guests.
(22:33):
What should I cook for dinner,right?
And so, um, I think,behaviorally, in in, when we
tune as we the, when we tune aswe, when the engineers, when
they tune the AI, you canactually have it be more
professional.
Yeah, right.
And we've even given the usersthe ability to say do you want
the response in a moreprofessional tone, more playful?
(22:55):
Do you want me to sort of dreambig Right.
Do you want the wide open RightParameters?
Here's the possibilities, yes,possibilities, and so.
So, right now it's still aparameter based, uh, sentiment,
um, and so we're not there yetto say when.
When patty asked a question, Ican tell by the inflection in
her voice, like you can, perhapsworking with her closely.
Oh, that's right.
(23:16):
Which?
What's the context of thequestion?
You know who moved this here.
It's different than you know.
Oh, how did this get here right.
It's very true and context ofthe question.
You know who moved this here.
It's different than you know.
Oh, how did this get here Right?
Very, true.
Speaker 1 (23:24):
One of the things I've noticed, too is
occasionally it'll misssomething key, right, so there
is still the need to bring itback in and read through and
look at it, but occasionallyit'll miss something that really
is pretty key to the solution,or whatever it is you're trying
to create.
Speaker 3 (23:41):
Yes, there is also a notion of doing your own
research, right, saying don'trely on this as the definitive
source, right, and I think thatgives people pause as well.
Well, hold on, what do you mean?
So we're back to the personneeding to know the right answer
, right?
No, I ask you for the answer.
(24:02):
There's.
It's called hallucination.
Yeah, when the, when themachine sort of fills in the
ranks, your brain does it Right.
And that's what's sort ofexciting and a bit scary, if you
will, about AI, is it's sort ofacting like our brain, like our
brain, right?
You read a word that doesn'thave a letter and you fill in
(24:23):
the letter.
Speaker 1 (24:24):
Yes.
Speaker 3 (24:25):
Right, just naturally , you fill it in, and so it's
filling in things, and we're notexactly sure how it does that,
what context it's using to dothat, and so there is a bit of
caution involved and so there isa bit of of caution involved.
Speaker 1 (24:41):
So we are printing our Christmas cards and the
printer contacted me about aword that I had left a letter
out of and we probably bothedited that thing 10 times and
it's funny because it's it said.
The last bullet point in it waswe started a podcast and I left
out the T.
So it's like we stared apodcast and I'm like the T.
So it's like we stared apodcast and I'm like, oh, we're
(25:04):
staring at that podcast, so justyeah.
Yeah, so exactly, but she caughtit, so it didn't go to print
and I was very thankful.
Speaker 3 (25:13):
And yeah, so and it takes a human to do that.
Speaker 1 (25:15):
It takes a human to do that.
Speaker 3 (25:17):
Not misspelled.
Speaker 1 (25:18):
Yeah.
Speaker 2 (25:19):
Yes and yeah, and we try to actually have everything
within our organization thatgoes out to a customer, in this
case, or people we know umedited so and you don't do your
own editing, because that sameissue so somewhere along the
lines is how does that crosscheck come in for the future of
(25:39):
agents and is there a secondaryagent that is almost the editor
and checking for that stuff?
Speaker 3 (25:47):
that's not their role if you, if that's part of your
work process and you're, you're,uh, tapping into a key sort of
aspect of agents.
Um, agents, so I mentioned thatai is a set of data right.
There's information, areasoning engine, that that
reasons over that that data, andthen an interface right.
(26:08):
And then I said that the agentwill doesn't need a human right,
but it still has an interface.
And you're exactly right,agents can talk to each other,
right?
Um, I don't know for folks whoare in the tech space.
We had a big tech show back inNovember called Ignite, and
while one of our engineeringleaders was on stage doing a
(26:31):
demo, the demo ran long.
It was supposed to just comeback in about you know 10
seconds or something.
So he's talking, talking and hesays, hold on, it's not
supposed to run this long.
And the demo was of an agentbuilding a program, like doing a
programming language.
The agent was doing what aperson would do building a
(26:52):
program and sending it to avalidation agent to validate,
and they were having aconversation about what was
right and that's what wasrunning, so long as they were
still discussing.
That's fascinating, and when Isay discussing, they can
actually use natural language todiscuss.
They don't have to use machinelanguage, they don't have to use
(27:14):
ones and zeros to discuss.
They can say that's not quiteright.
They say, well, myspecification said it should be
this way and that's why I did itthat way.
Well, realistically, I don'thave that information until
later in the process.
Well, and then you're likewithin, so you ended up stopping
the, the demo, yeah, so, yes,they can talk to each other.
So, and you're they collaborate,they can collaborate that's
amazing.
Speaker 2 (27:34):
That's amazing.
What's really interesting aboutthat is we see that play out
every single day with ourclients in their business worlds
.
Is that back and forth that, inthis case, can slow down the
overall goal of achieving theoverall goal?
So some of the same things willhave to be over addressed and
overcome to be able to have itto be as efficient as possible.
(27:56):
Let me pull you.
Speaker 3 (27:57):
I agree.
I think one of the one of theuse cases I think about is, as
you mentioned, between twocompanies who are trying to get
work done, but they havedifferent taxonomy and language.
We call it an order number, youcall it a customer number or
something like that, and so Ihave to translate and I say our
(28:18):
order number is this oh, let mefind our equivalent of customer
number and so it's not reallyhigh value interaction.
We're still trying tocommunicate Now.
We've been doing business for15 years together, but we still
every day have to remember ourown language, the translation,
and leaving the people to say ah, how is the business going?
(28:41):
What's the sell-through?
How can we improve this?
And now doing the higher ordermore value add capabilities, I
think, is the real promise.
Speaker 2 (28:52):
Absolutely, and those things again.
We see that vernacular to apoint.
We probably three times withinthe last week and a half have
had conversations with companieswhere same words mean have
different meanings within thesame organization.
You're in the department a andpatty's in department b.
I'm in department c and we'llbe collaborating using the same
(29:17):
terminology, but walk away withdifferent, uh, ideation of what
it is I am to go do based uponthe fact that my definition is
different than both of yours.
Speaker 3 (29:26):
Yeah, yes, yes, and and getting that uh correct, I
think is has been a challengefor, uh, the tech industry, for
developers, for years.
Yes, right, and just liketranslation tables, yes, uh,
that we keep, and you have tokeep them current, and that sort
of thing.
And so I'll tell you that'sgood news for those of us who
work for a living is they stillneed us.
Speaker 1 (29:50):
Yes.
Which is the message I thinkeveryone wants to hear.
Speaker 3 (29:53):
Yes, ai will help, it will be an aid.
I would say, embrace it.
I would say, embrace it.
There is some level of and youmentioned it, patty sort of
resistance to change and that'swhy, for that reason, we say
that for every good AIimplementation, you need an
(30:14):
equal change managementImplementation.
You need to have someone whosays, ok, so how do we get the
workers right, the people whoare going to have the bottom 30
percent of their role done forthem?
How do we get them tounderstand that and to adopt it
Right?
And where's their new musclethey're going to build?
(30:36):
And so bring that in and that'sa whole.
Nother discipline inside thebusiness.
Yeah, because, just like youand in the coaching space and
the readiness and performancespace, you come at it
differently yes, absolutelyabsolutely.
Speaker 2 (30:49):
One thing I do want to ask about, too, is what you
mentioned.
Uh, development is going on atthe enterprise level.
It is large corporations wholive in that space and or
startups that are coming in Atsome point in time.
Typically, things filter downto impact small-sized businesses
, which are the backbone of theworkforce.
(31:09):
Does AI and the new phase of AIwith agents come down to that
point and really start to impactthem in a way that helps to
accelerate them?
Speaker 3 (31:18):
I think it does right .
So you point out a good pointthat our enterprise, my
enterprise customers, who buildtheir own right, they're out
there, we see press releasesevery day and they're building
their own and pushing theenvelope.
The smaller companies aresaying, well, they're just not
able to do that and don't reallyhave a need to do that.
You'll start to see AI creepinto the products that they buy.
(31:42):
So things like QuickBooks yes,right, so the QuickBooks will
take away.
They'll say, oh, no, that youcan press here and have the AI
agent evaluate and tell youwhere your problems are.
You know, if you file taxeswith your own taxes, they have
the little checker at the end.
Yes, it's that kind of thing.
So I think, um, financesoftware, uh, crm software,
(32:04):
order management software itwill start to get built in and
so they will know it as just afaster, easier, more seamless
workflow for themselves.
And so, uh, there's a breakneckspeed right now in terms of the
companies who build thosesolutions wanting to push it
down.
Right, that that 19 trillion isgoing to come from somewhere,
(32:25):
and so there might be a highercost because we're going to
reason over your, you know,seven years of of, uh, financial
data.
Uh, so that now we can we knowyour business yeah, we, we know
it as well as your financeperson who happens to be out
sick today.
I got this, yeah good.
Speaker 1 (32:45):
Well, we are excited to be thinking about a second
podcast with you where we'regoing to really dig deeper into
this topic and change managementassociated with it.
But as we seek to wrap up today, I just have one last question
for you.
This is a lot of change.
How can people keep up with itrather than resist it?
(33:06):
What are some tips you have?
Speaker 3 (33:08):
I think um use it in your daily life.
You know it's um.
Enterprises saw that theiremployees, once everybody got an
iphone right, they didn't likethe green screens and the f12
and all of the crazy keystrokesyou had to do to get stuff done.
They're like no, I want it,just like here.
You want it right, just like,just like, right here.
Yeah.
And so, um, we're now seeingthat consumers are sort of
(33:32):
demanding a differentinteraction.
They don't mind the chatbots onthe front side of the phone
call.
Now, right, if they'reintelligent and they can
converse.
So I would say, use it, youknow, if you're.
If you're thinking about achristmas gift, ask ai describe
the recipient.
Um, if you're writing thatemail, that needs to be
impactful.
Um, ask ai, what's a better wayto do this?
(33:56):
Yeah, right, and then start tojust get familiar with it and
say, oh, it's actually helpingme, I don't have to sort of
resist it.
Speaker 1 (34:03):
I've put things in and asked for it to make it more
succinct or more professionalor more playful, and it really
does a great job there.
Speaker 3 (34:12):
That's a learning you could actually learn by doing
you really can.
Speaker 2 (34:16):
You really can.
What I have to say reallyquickly is I wish we would have
had this conversation earlier,because that idea of putting in
information about the Christmaspresent was so helpful.
Speaker 1 (34:25):
Would have been helpful before the holidays.
Yes, indeed, Vic.
Thanks so much for being here.
We look forward to our nextconversation with you, and
that'll be in two weeks.
Listeners, thanks for joining.
Have a great day.
Speaker 2 (34:39):
Thank you, thank you.
Speaker 1 (34:40):
The 4 Bars Podcast has been brought to you by Edges
Inc.
A growth advisory firm based inBentonville, arkansas.
I founded the company in 2001.
Speaker 2 (34:51):
Edges promotes growth , people, companies and ideas.
Our team collaboration tool,called Interface Methods, is a
basis for teams to work togethermore collaboratively,
understand each other and acceptdifferences and address
challenges together.
Speaker 1 (35:01):
We also started a nonprofit called Unform your
Bias.
We teach kids and their adultinfluencers how to utilize
storytelling as a means toreduce bias in the world.
We hope you'll check us out,subscribe to our podcast and
look at our website.
Welcome to the 4 Bars podcast.
I'm Ken Leith and I'm PattiLeith.
We're your hosts for somecompelling dialogue, encouraging
our listeners to strengthentheir connections and build
strong communities, lifting eachother up and connecting in ways
that matter.
We named the podcast 4 Bars asa reference to how hard we work
to find a 4 Bars connection onour devices.
Speaker 2 (00:19):
And we wondered what could happen with relationships
if we worked as hard atconnecting.
Speaker 1 (00:24):
Let's find out.
Welcome back to Four Bars, thepodcast.
We're excited to be here.
I have my co-host with me, kenLeith, and we are excited to
host a guest who will bespeaking to us today about AI,
vic Miles.
Vic is the general manager forindustry strategies in America,
so welcome, vic, it's great tosee you.
Thank you for having me Tell usa little bit about yourself.
(00:46):
What got you to this point?
Speaker 3 (00:56):
Yeah, so I've been 15 years with Microsoft.
As you said, I'm the generalmanager of our industry business
in the Americas.
We look after our moststrategic customers in the
Americas North and South and wehelp them sort of unpack how to
use technology to meet theirbusiness needs.
So we sit at the intersectionbetween the business imperatives
and the right technologysolution.
(01:16):
I have a team of folks about 20people or so who engage in
marketing and we at Microsoftlike to hire out of the industry
right.
So my background I came fromindustry.
I think of myself as a retailer.
I've worked for Walmart storesfor 10 years and did many things
(01:36):
in that business.
I was on the IT side, so Istarted out as a software
developer and engineer.
Worked in the operation side ofthe business, so point of sale,
cash registers, time clocks andpayroll and all that stuff in
the store.
So I really have a heart forwhat we call them as frontline
associates who work in the storeand have just built a career
(01:59):
around this intersection betweenthe needs of consumers.
I find that I'm really tunedinto what we call the market.
What do consumers want and howdo retailers and service
providers, whether they'rehealthcare or anything how do we
serve that need with the rightbalance of profit and all of the
(02:23):
rest.
So it's a great position.
I I've enjoyed it now as, as Imentioned, for 15 years, and so
now it feels like a constantevolution, right.
We probably know two years havebeen the same right, or
certainly no more than two yearshave been the same, and it's
just constant change.
And so again, excited to behere.
Speaker 1 (02:44):
Well, it's great to have you.
Great to have you.
Thanks for sharing that.
As you think about yourbackground and how you've put
together your experiences, whatdid you learn early that really
helps drive you today?
Speaker 3 (03:05):
I learned that and you and I met years ago.
Yes, we did, and we all havethese sort of moments of growth,
these spurts of growth in ourcareer, and you and I were
meeting and I think you helpedme grow into a more leadership
position right from management.
(03:25):
What I, what I, learned earlyon was you have to, you have to
get on the ground right, andit's a growing up in the culture
of Sam Walton, it was about you.
You really couldn't have a goodunderstanding of the business
unless you were in the field.
Yes, right, and you gotta go tothe field.
So you know, in the field, yes,right, and you've got to go to
the field.
So you know, in my tenure I usedto run a cash register on
(03:47):
Saturday because my team and Ideveloped the solution that you
know, 300,000 cashiers would useevery day.
And just to gain empathy, right, I used to go on the Saturday
it wasn't my idea to be fair inthe beginning and run a register
and you haven't lived until youcan identify a uh, it's true,
(04:07):
uh, produce item yeah, right,and a mother with with three
children is there and you'retrying to get through and you
start to have empathy for so Iwould tell you I learned empathy
for, uh, how the systems thatwe develop, how they impact the
end user, yes, and so I bringthat forward into, uh, my, my
current career.
And so a lot of times we'll sayshow me, I'm from Missouri as
(04:29):
well, so it's the show me state.
We'll say, show me this thingthat you're talking about, can
we see it?
Can we do a plant tour?
Can we get on the floor of astore?
That kind of thing.
Speaker 1 (04:37):
There is no replacement for that kind of
experience and, similar to you,I started with Food Lion and
learned the business from theground up and just spent time on
the floor before going intomanagement.
That was a requirement for themand I still draw back on that
experience on a regular basis.
I agree Well, I discovered yourpassion for this topic as we
(05:01):
reconnected, but also in yourown podcast.
Speaker 3 (05:04):
Can you?
Speaker 1 (05:05):
tell us a little bit about your podcast so, uh,
conversations on retail.
Speaker 3 (05:08):
It's a.
It's a youtube-based podcast.
Um, a gentleman by the name ofmatt pfeiffer had run sort of
the program and he has hoststhere, and so I do a show called
beyond the tech, and so I'mvery interested in so the tech
side.
I live and breathe that daily.
But I also get in my role anopportunity to meet fascinating
(05:29):
people Right, who are, who aredoing things that
technologically make the press,but then they themselves are
fascinating, yeah, right, it'slike oh, you're you do marathon
running, yes, or you like tohike, or you're a world-class.
What is it the in the Olympics?
We're the Curly.
Speaker 2 (05:48):
Yes, Curly.
Speaker 3 (05:49):
How did you know that kind of thing, and so I invite
guests onto the show to talkabout their background and how
again that intersection withtechnology so what makes them a
whole person?
Speaker 1 (06:01):
outside of this Exactly Love it.
We've watched a few episodesand I highly recommend listeners
that you check it out.
We're here to talk about AI, ofcourse, so I'm interested in
learning about the journey thatMicrosoft has been on with AI.
Speaker 3 (06:19):
Yes, and it's interesting.
I didn't sort of start out evenin my own podcast being an AI
sort of evangelist or aneducator, but it is the topic du
jour, everybody wants to talkabout it, and so I've had my own
journey in understanding it,and we're on the cusp of yet
(06:39):
another shift in AI and sohopefully we will unpack that
today as we talk.
But artificial intelligence,really, I like to start with a
primer by saying it's not new.
So artificial intelligence hasbeen around really since the
late 50s and it didn't look thesame.
(07:00):
It was just, you know, itstarted as machine learning
trying to find patterns, nowwhat you do with the pattern
back to you, human right, thatkind of thing.
And now we're we're enteringinto this space where um
companies are very excited aboutbringing expertise to their
employees and the folks thatwork in the business, um, being
able to take a uh, a four yearassociate and give them the same
(07:23):
expertise as your 16-yearassociate, so that they can both
have the same level of whetherit's service or performance in
the business, and that's supercompelling.
There's dollars involved as well.
Depending on which source youplug into, whether it's Goldman
Sachs or McKinsey or IDC, it'sbetween $2 and $19 trillion of
(07:48):
GDP impact.
Wow, $19 trillion, I mean, it'sa pretty wide range.
Let's see.
I think IDC has it at $14 andMcKinsey had it at $2 to $4
trillion, and so that is what'sdriving companies to say I want
in.
And there's cautious optimism,right, as we think about the
(08:11):
impact to jobs, right.
So I always say I don't want tolive in a community and neither
do you or the audience eitherthat has 40% unemployment.
That's not what we're here for,right.
So you know when the machinescan do everything better, and so
it's about finding.
You know, as the audience andyourselves think about AI, it's
about finding where are thefriction points.
(08:32):
What are the things thatirritate you about engaging with
brands?
You know, in my space, theretail and consumer goods that
irritate you about engaging withbrands or that keep you from
having the best experience, andcan we use AI to smooth that out
?
Right, I'm trying to replace abunch of people, but if there
are some areas, some seams thatwe can sew up, yes, yes,
(08:55):
interesting.
Speaker 2 (08:56):
Yeah, I mean and to your point, it's a topic
everywhere.
We do not go into a client'sfacility over the last several
years and that's not coming upin a discussion, and everyone's
strategic roadmap has an AIcomponent to it now, whereas
three years ago that wasn't thecase.
It's kind of just being talkedabout on the peripheral.
But if you think about it now,where companies are starting to
(09:17):
embrace it and where they'restill hesitant, where do you see
those junctures as far aswhere's that embracing and where
is their hesitancy stillthere's?
Speaker 3 (09:26):
again being in the retail space when you're working
with a business withsingle-digit margins right, the
hesitancy comes from there's alot of cost here, right, so so
we talk about AI being areasoning engine on the back end
(09:48):
, you know, and then a largeamount of data that it reasons
over, and then an interfaceRight, and so we know that
interface largely today as aconversational interface.
For Microsoft, we branded itco-pilot Right, so it's a
co-pilot, you're the pilot andthis is here to help you.
The hesitancy is is with cost,because that reasoning engine
really does consume resources,compute resources, and it costs
(10:09):
money.
And so we see people say holdon, where's the value?
And so now we're seeing peopleshift into I like the idea of it
.
What's the roi?
Right, the?
The uptake that we've seen is in, uh, generally, content
creation.
Um, so, companies, who, who,perhaps, uh, who sell cars,
vehicles, especially usedvehicles, where each one is
(10:31):
different, right, and you haveto create copy material that
says here is a Ford truck that'sblue, it's got one den on the
side, it's got 60,000 miles, orwhatever.
That can take days to create.
Right, you have to go see thevehicle, that sort of thing.
What all is relevant to theconsumer and having AI
understand the relevancy of theconsumer.
(10:52):
So AI is able to take who isyour target audience right by
locale.
So my target audience is SanDiego, california and the 150
mile radius.
Okay, I now know somethingabout this, this, this
demographic and then so on.
Right, am I going for theprofessional person to use it as
work and it will then createcopy that is relevant to that
(11:15):
audience right In seconds.
Right, and so we're seeinglarge uptake there.
Call centers you know being atthat front end and we've
experienced IVRs right whenthey're sort of trying to route
you through.
But having a conversation in acall center that says why are
you calling?
Oh, have you tried thisMeanwhile.
Having a conversation in the ina call center that says why are
you calling?
Oh, have you tried this?
Meanwhile, you're in the queuebut I'm going to converse with
(11:36):
you, that sort of thing.
So we've seen uptake on thecontent side, but hesitancy on
the cost side yeah, understand,understand, absolutely for many
people they're they're afraid ofit.
Speaker 1 (11:46):
They're afraid of what it will do to the future.
Can you speak to that a littlebit?
Speaker 3 (11:50):
Yeah, you know, and it's if I can do some small
measure of assuaging people'sconcerns around AI.
I think that in the in the US,about 75 percent of our GDP is
in the service industry, right,that's, that's health care,
retail, hospitality, that sortof thing, and so those hands-on
(12:13):
businesses.
I don't see a lot changing,right?
So, again, ai that knows you inthe hotel room and presents a
particular menu on the TV basedon your preferences, you still
need, you know, housekeeping,you still need front desk, you
still need all of the folks thatcreate the rest of the
experience, and so a lot offolks won't.
(12:34):
I would think, for folks likemyself and the people that you
work with the professionals onthe back end, right, that says,
hey, am I really adding value oram I just part of the value
chain?
I'm in the process, right?
I don't know that.
I have the same message for them, right, and I think the answer
is get to value, right, get towhere you're creating value,
(12:56):
adding value, because if you'reworking a, a let's call it an
inbound queue of invoices thatneed to be cleared, right, right
, and and the, your goal isclear the invoices every day,
right, and and, with a, withsome threshold, at least 100
plus empty, you know, within 10by the time you leave, kind of
thing.
Um, ai will do a lot of thatbetter than people.
(13:19):
The good news is, though, is itfrees you up.
So start to understand why am Iclearing invoices, right?
What is the value that thisprocess has?
And then I can go add valuesomewhere else in the value
chain, but if I only know this,the rope process of clearing
invoices, that that's going tochange, yeah, yeah.
Speaker 1 (13:38):
One of the fundamentals we work with
executives on building theirstrategic capacity is to let
someone or something else helpyou do that, Because then it
frees you up to really do thethings that your intellect and
your training and background canbenefit.
The organization you know Italked about it's into that, I
(13:59):
think.
Speaker 3 (14:00):
I talked about the tenured associate, the 16-year
associate and the four-yearassociate, right, I think, one
of our findings?
We looked at some studies.
We found that the 16 yearassociate has aspiration to do
more, but they are the glue thatholds this process together.
Yes, right, it takes that levelof tenure to keep this process
(14:20):
moving.
Very right, and so, if we canremove that, you're now free to
go explore your own passions.
Yeah, right, I didn't reallyintend to be here clearing
invoices for 16 years, but Iunderstand it front and back.
There's nothing I haven't seen.
I am critical to the business,but my real passion is I want to
lead X department, and so nowthis frees you up to go do that.
(14:44):
So we do see school teachers.
Do they really want to gradeall the papers at night?
Speaker 1 (14:53):
teachers right.
Speaker 3 (14:53):
Do they really want to grade all the papers at night
?
We think not right, and sobeing able to sort of reduce
that, I think that the qualityof life will will raise.
There's a lot of folks that say, um, uh, the most uh famous
person that says bill gates, uh,founder, says that you know,
people work will look verydifferent in the future and for
many people they won't have towork.
I don't know what that lookslike yet.
Yes, but that's the future thatwe're going to, is there'll be
(15:14):
enough value created that thatthat 19 trillion right.
As long as we don't sort of,you know, in a social construct,
let that consolidate at the top.
Speaker 2 (15:25):
I think all of our lives keep it, yeah yeah, yeah,
yeah, and it's exciting andscary at the same time when it
comes out.
But, as you were talking there,one of the things that hit me
part of it really can be aboutit's delegating those tasks that
really are not tasks that arerelational.
So it's taking those thingsthat just need to get done at
(15:47):
someone's desk and primarilyalone, potentially where they do
those things.
Ai today, as we see it, andwhat you're defining is more of
I can take that off your plate,delegate it over here.
Speaker 3 (16:04):
You go, do something more relational something that
has higher value to theorganization.
Correct, indeed?
Yes, exactly, and maybe now's agood time to sort of talk about
the fork that is occurring inAI market right now.
We're about two years in, let'ssay, to the big bang of AI.
We began with conversational AIyes, Right.
So chatbots yes, right, very,very intelligent chatbots.
(16:27):
I don't know if you've used it.
Speaker 2 (16:28):
I have yeah.
Speaker 3 (16:30):
You know, when you no longer have to search, right.
So I wanted to search.
What is it?
What is the procedure forgetting a new driver's license
in Arkansas, right and so, and Icould go to the Department of
Finance Administration, the DFAside, and then elite, read and
get the PDF, and and that, but.
But an AI agent told me, oh,you need these four documents,
(16:51):
you have to come here.
Most offices closed by four.
Told me exactly what I wantedto know.
Yeah, in in seconds.
So that's the conversational ai.
Okay, and ken, what you talkedabout is this this, what I like
to call units of work?
Right, these tasks?
Yes, um, we're moving into this, this notion of, uh, agentbased
AI.
They call it agentic.
(17:11):
So agentic AI is agent-based AI, and agents will autonomously,
so without prompting, do unitsof work.
And so when you think about theinterconnection between
departments in a business, right, so I'll go back to my invoice
thing.
(17:32):
So the, the invoices come in,they first get before they go to
finance to be paid.
They have to be clear did,where were all the numbers right
?
Was the inventory actuallymoved?
That kind of thing?
And that inventory movementcomes from a different
department.
Yes, and and and and then wemove it over into finance and
they pay it.
And that's how we used to doprocesses right and build
corporations.
(17:52):
And now we're saying well, thedata of the inbound invoice and
what was received is availableto the agent, the data about
what was ordered is available tothe agent and the data about
the compliance aspects of payingand finance is available to the
agent.
Compliance aspects of payingand finance is available to the
agent.
So, bringing those threetogether, the agent can just do
(18:14):
the process without prompting orintervention.
So we're moving from human inthe loop, right, human in the
loop to no human in the loop,and so again.
But if you're thinking aboutfriction, if your job is
clearing the friction meaning Iwork on invoices all day your
job's going to change.
But there are those two thingsand I would invite the audience
(18:36):
to really think about in theirown businesses.
Right, what are those units ofwork?
And you call them tasks, butit's actually even bigger
because we had three tasks there.
Yes, right.
Speaker 1 (19:02):
And it's really a unit of work trying to just pay
the invoice, and so, yeah, it'svery interesting.
You can almost get more out ofthat than you can out of any
other type of searching thesedays, but there are some
limitations that you all havefound, which has prompted your
journey into it.
Speaker 3 (19:19):
Jente Can you speak to that a little bit.
Yes, exactly so.
The first limitation and Kenmentioned it is the hesitancy
around the cost.
Yes, right.
So executives, boards aresaying what does it mean to make
my employee 20% more productive?
Yeah, right, is that a longerbreak?
Or do we get 20% more work done?
(19:40):
Right, are we changing thebusiness process?
So that queue of invoices isthere going to be 20% more?
Do I get rid of people?
They want to know where is thesavings?
I see the cost side.
Yes, percent more.
Do I get rid of people?
They want to know where is thesavings?
See the cost side.
Yes, and so the ROI has beenone of the challenges with that
is we'll we'll?
Yes, we understand people canseek information and find
information.
(20:01):
There is this notion ofprompting.
There is there's a new sort ofengineering role in the tech
space called a prompt engineer.
So before we used to, welearned how to talk to the
computers, what I call machinespeak.
Right, so, as a developer,they're talking to a computer
all day, right through alanguage.
(20:22):
And now this prompt engineersays knowing the data I want, I
want to actually be able to askthe question in a well-formed
way that gives me the answer Iwant.
So you you ask what were thechallenges we found?
Depending on how you ask thequestion, the machine may take a
different context yes, rightand give you a different answer
(20:44):
Right, and so therein lies thechallenge.
Hold on.
We don't want to just unleashthis powerful tool and someone
says, you know, should I ordermore?
And the machine says, well,absolutely, sure, order up.
Speaker 1 (20:57):
Yeah, right, right, right, yeah, never mind Never.
Speaker 3 (20:58):
Yeah, right, right, right, yeah, Never mind that
there's only two weeks left inthe season, and you know, and
the cost changed and we don'thave, yeah, the supply chain
isn't there, and so that's been,the hesitancy is hold on, we
don't know how to create prompts, broadly speaking.
Right, we don't know how tocreate prompts, and is the model
(21:20):
we call the, the model answersthe language model, does it have
the proper context?
And so that's a that's a bigchallenge.
We're still working on that yeah, yeah so agentic is is sort of
the new uh way that says givingit, given a process that happens
every day and has to happenevery day, that's, we can do
that without thinking yeah, Ilove that element.
Speaker 2 (21:42):
Yeah, I think the other thing about that.
So I mean we're really talkingabout shifting and almost
becoming bilingual with that, Imean with the technology and
then the natural language thatpeople utilize.
Yes, when we think about that,even knowing language, and then
there's syntax, there's theability to understand
behaviorally uh, how should Ianswer a question?
(22:04):
Does ai, do you see thatevolving to another level,
potentially with agents to beable to kind of pick up on some
other nuances and languages?
Speaker 3 (22:12):
yes, I do see it evolving and it's it's already
there to some degree.
It's about harnessing it.
Okay, right.
So, not knowing that, by askinga question a certain way,
you've implied a, a bias, not ina bad way, but maybe your bias
is I only have 20 minutes, whatshould I cook for dinner Right?
Versus I'm trying to impress myboss and 15 guests.
(22:33):
What should I cook for dinner,right?
And so, um, I think,behaviorally, in in, when we
tune as we the, when we tune aswe, when the engineers, when
they tune the AI, you canactually have it be more
professional.
Yeah, right.
And we've even given the usersthe ability to say do you want
the response in a moreprofessional tone, more playful?
(22:55):
Do you want me to sort of dreambig Right.
Do you want the wide open RightParameters?
Here's the possibilities, yes,possibilities, and so.
So, right now it's still aparameter based, uh, sentiment,
um, and so we're not there yetto say when.
When patty asked a question, Ican tell by the inflection in
her voice, like you can, perhapsworking with her closely.
Oh, that's right.
(23:16):
Which?
What's the context of thequestion?
You know who moved this here.
It's different than you know.
Oh, how did this get here right.
It's very true and context ofthe question.
You know who moved this here.
It's different than you know.
Oh, how did this get here Right?
Very, true.
Speaker 1 (23:24):
One of the things I've noticed, too is
occasionally it'll misssomething key, right, so there
is still the need to bring itback in and read through and
look at it, but occasionallyit'll miss something that really
is pretty key to the solution,or whatever it is you're trying
to create.
Speaker 3 (23:41):
Yes, there is also a notion of doing your own
research, right, saying don'trely on this as the definitive
source, right, and I think thatgives people pause as well.
Well, hold on, what do you mean?
So we're back to the personneeding to know the right answer
, right?
No, I ask you for the answer.
(24:02):
There's.
It's called hallucination.
Yeah, when the, when themachine sort of fills in the
ranks, your brain does it Right.
And that's what's sort ofexciting and a bit scary, if you
will, about AI, is it's sort ofacting like our brain, like our
brain, right?
You read a word that doesn'thave a letter and you fill in
(24:23):
the letter.
Speaker 1 (24:24):
Yes.
Speaker 3 (24:25):
Right, just naturally , you fill it in, and so it's
filling in things, and we're notexactly sure how it does that,
what context it's using to dothat, and so there is a bit of
caution involved and so there isa bit of of caution involved.
Speaker 1 (24:41):
So we are printing our Christmas cards and the
printer contacted me about aword that I had left a letter
out of and we probably bothedited that thing 10 times and
it's funny because it's it said.
The last bullet point in it waswe started a podcast and I left
out the T.
So it's like we stared apodcast and I'm like the T.
So it's like we stared apodcast and I'm like, oh, we're
(25:04):
staring at that podcast, so justyeah.
Yeah, so exactly, but she caughtit, so it didn't go to print
and I was very thankful.
Speaker 3 (25:13):
And yeah, so and it takes a human to do that.
Speaker 1 (25:15):
It takes a human to do that.
Speaker 3 (25:17):
Not misspelled.
Speaker 1 (25:18):
Yeah.
Speaker 2 (25:19):
Yes and yeah, and we try to actually have everything
within our organization thatgoes out to a customer, in this
case, or people we know umedited so and you don't do your
own editing, because that sameissue so somewhere along the
lines is how does that crosscheck come in for the future of
(25:39):
agents and is there a secondaryagent that is almost the editor
and checking for that stuff?
Speaker 3 (25:47):
that's not their role if you, if that's part of your
work process and you're, you're,uh, tapping into a key sort of
aspect of agents.
Um, agents, so I mentioned thatai is a set of data right.
There's information, areasoning engine, that that
reasons over that that data, andthen an interface right.
(26:08):
And then I said that the agentwill doesn't need a human right,
but it still has an interface.
And you're exactly right,agents can talk to each other,
right?
Um, I don't know for folks whoare in the tech space.
We had a big tech show back inNovember called Ignite, and
while one of our engineeringleaders was on stage doing a
(26:31):
demo, the demo ran long.
It was supposed to just comeback in about you know 10
seconds or something.
So he's talking, talking and hesays, hold on, it's not
supposed to run this long.
And the demo was of an agentbuilding a program, like doing a
programming language.
The agent was doing what aperson would do building a
(26:52):
program and sending it to avalidation agent to validate,
and they were having aconversation about what was
right and that's what wasrunning, so long as they were
still discussing.
That's fascinating, and when Isay discussing, they can
actually use natural language todiscuss.
They don't have to use machinelanguage, they don't have to use
(27:14):
ones and zeros to discuss.
They can say that's not quiteright.
They say, well, myspecification said it should be
this way and that's why I did itthat way.
Well, realistically, I don'thave that information until
later in the process.
Well, and then you're likewithin, so you ended up stopping
the, the demo, yeah, so, yes,they can talk to each other.
So, and you're they collaborate,they can collaborate that's
amazing.
Speaker 2 (27:34):
That's amazing.
What's really interesting aboutthat is we see that play out
every single day with ourclients in their business worlds
.
Is that back and forth that, inthis case, can slow down the
overall goal of achieving theoverall goal?
So some of the same things willhave to be over addressed and
overcome to be able to have itto be as efficient as possible.
(27:56):
Let me pull you.
Speaker 3 (27:57):
I agree.
I think one of the one of theuse cases I think about is, as
you mentioned, between twocompanies who are trying to get
work done, but they havedifferent taxonomy and language.
We call it an order number, youcall it a customer number or
something like that, and so Ihave to translate and I say our
(28:18):
order number is this oh, let mefind our equivalent of customer
number and so it's not reallyhigh value interaction.
We're still trying tocommunicate Now.
We've been doing business for15 years together, but we still
every day have to remember ourown language, the translation,
and leaving the people to say ah, how is the business going?
(28:41):
What's the sell-through?
How can we improve this?
And now doing the higher ordermore value add capabilities, I
think, is the real promise.
Speaker 2 (28:52):
Absolutely, and those things again.
We see that vernacular to apoint.
We probably three times withinthe last week and a half have
had conversations with companieswhere same words mean have
different meanings within thesame organization.
You're in the department a andpatty's in department b.
I'm in department c and we'llbe collaborating using the same
(29:17):
terminology, but walk away withdifferent, uh, ideation of what
it is I am to go do based uponthe fact that my definition is
different than both of yours.
Speaker 3 (29:26):
Yeah, yes, yes, and and getting that uh correct, I
think is has been a challengefor, uh, the tech industry, for
developers, for years.
Yes, right, and just liketranslation tables, yes, uh,
that we keep, and you have tokeep them current, and that sort
of thing.
And so I'll tell you that'sgood news for those of us who
work for a living is they stillneed us.
Speaker 1 (29:50):
Yes.
Which is the message I thinkeveryone wants to hear.
Speaker 3 (29:53):
Yes, ai will help, it will be an aid.
I would say, embrace it.
I would say, embrace it.
There is some level of and youmentioned it, patty sort of
resistance to change and that'swhy, for that reason, we say
that for every good AIimplementation, you need an
(30:14):
equal change managementImplementation.
You need to have someone whosays, ok, so how do we get the
workers right, the people whoare going to have the bottom 30
percent of their role done forthem?
How do we get them tounderstand that and to adopt it
Right?
And where's their new musclethey're going to build?
(30:36):
And so bring that in and that'sa whole.
Nother discipline inside thebusiness.
Yeah, because, just like youand in the coaching space and
the readiness and performancespace, you come at it
differently yes, absolutelyabsolutely.
Speaker 2 (30:49):
One thing I do want to ask about, too, is what you
mentioned.
Uh, development is going on atthe enterprise level.
It is large corporations wholive in that space and or
startups that are coming in Atsome point in time.
Typically, things filter downto impact small-sized businesses
, which are the backbone of theworkforce.
(31:09):
Does AI and the new phase of AIwith agents come down to that
point and really start to impactthem in a way that helps to
accelerate them?
Speaker 3 (31:18):
I think it does right .
So you point out a good pointthat our enterprise, my
enterprise customers, who buildtheir own right, they're out
there, we see press releasesevery day and they're building
their own and pushing theenvelope.
The smaller companies aresaying, well, they're just not
able to do that and don't reallyhave a need to do that.
You'll start to see AI creepinto the products that they buy.
(31:42):
So things like QuickBooks yes,right, so the QuickBooks will
take away.
They'll say, oh, no, that youcan press here and have the AI
agent evaluate and tell youwhere your problems are.
You know, if you file taxeswith your own taxes, they have
the little checker at the end.
Yes, it's that kind of thing.
So I think, um, financesoftware, uh, crm software,
(32:04):
order management software itwill start to get built in and
so they will know it as just afaster, easier, more seamless
workflow for themselves.
And so, uh, there's a breakneckspeed right now in terms of the
companies who build thosesolutions wanting to push it
down.
Right, that that 19 trillion isgoing to come from somewhere,
(32:25):
and so there might be a highercost because we're going to
reason over your, you know,seven years of of, uh, financial
data.
Uh, so that now we can we knowyour business yeah, we, we know
it as well as your financeperson who happens to be out
sick today.
I got this, yeah good.
Speaker 1 (32:45):
Well, we are excited to be thinking about a second
podcast with you where we'regoing to really dig deeper into
this topic and change managementassociated with it.
But as we seek to wrap up today, I just have one last question
for you.
This is a lot of change.
How can people keep up with itrather than resist it?
(33:06):
What are some tips you have?
Speaker 3 (33:08):
I think um use it in your daily life.
You know it's um.
Enterprises saw that theiremployees, once everybody got an
iphone right, they didn't likethe green screens and the f12
and all of the crazy keystrokesyou had to do to get stuff done.
They're like no, I want it,just like here.
You want it right, just like,just like, right here.
Yeah.
And so, um, we're now seeingthat consumers are sort of
(33:32):
demanding a differentinteraction.
They don't mind the chatbots onthe front side of the phone
call.
Now, right, if they'reintelligent and they can
converse.
So I would say, use it, youknow, if you're.
If you're thinking about achristmas gift, ask ai describe
the recipient.
Um, if you're writing thatemail, that needs to be
impactful.
Um, ask ai, what's a better wayto do this?
(33:56):
Yeah, right, and then start tojust get familiar with it and
say, oh, it's actually helpingme, I don't have to sort of
resist it.
Speaker 1 (34:03):
I've put things in and asked for it to make it more
succinct or more professionalor more playful, and it really
does a great job there.
Speaker 3 (34:12):
That's a learning you could actually learn by doing
you really can.
Speaker 2 (34:16):
You really can.
What I have to say reallyquickly is I wish we would have
had this conversation earlier,because that idea of putting in
information about the Christmaspresent was so helpful.
Speaker 1 (34:25):
Would have been helpful before the holidays.
Yes, indeed, Vic.
Thanks so much for being here.
We look forward to our nextconversation with you, and
that'll be in two weeks.
Listeners, thanks for joining.
Have a great day.
Speaker 2 (34:39):
Thank you, thank you.
Speaker 1 (34:40):
The 4 Bars Podcast has been brought to you by Edges
Inc.
A growth advisory firm based inBentonville, arkansas.
I founded the company in 2001.
Speaker 2 (34:51):
Edges promotes growth , people, companies and ideas.
Our team collaboration tool,called Interface Methods, is a
basis for teams to work togethermore collaboratively,
understand each other and acceptdifferences and address
challenges together.
Speaker 1 (35:01):
We also started a nonprofit called Unform your
Bias.
We teach kids and their adultinfluencers how to utilize
storytelling as a means toreduce bias in the world.
We hope you'll check us out,subscribe to our podcast and
look at our website.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.
24/7 News: The Latest
Today’s Latest News In 4 Minutes. Updated Hourly.