Subscribe
Share
Share
Embed
Authentic, Authoritative, Unapologetic ServiceNow commentary by Cory "CJ" Wesley and Robert "The Duke" Fedoruk
Duke: right, CoRey, what
are we talking about today?
CJ: Today, Duke, we got
something special for him.
We've got, you want, is
it Forrest Green, Ben,
Ben: It is for screen.
Yeah.
As a matter of
CJ: Okay, cool.
Awesome.
All right.
Ben: In between last time I was on
the show and now I've , got married,
as you guys know, , at knowledge,
, this year, which was pretty cool.
So.
Duke: Yes.
Ben: But, , I'm glad you guys let me,
, hang out and chase you down , and talk
at you and everybody else, , on the show.
Because I was like yelling at my
car, , radio, like listening to
the podcast last week and there
were lots of like, Oh my God, yes.
And, and what about it like this?
And , don't forget to
tell them about that.
So, , I guess, , a CTA mini boss, , yells
at car stereo while listening to podcast.
So
CJ: I love it.
Mini boss,
Duke: yeah, it's one of those things
like when Ben asked to come on the
episode, we're like, yeah, like,
CJ: right?
Ben: good times.
It's good times.
, you guys are always on the cusp or
on the pulse of whatever's happening,
, whether intentional or not, although
I suspect it's very intentional.
So, one thing in particular , last
weekend you guys were talking about.
So where does gen AI fit, , and
then different levels of
complexity in the platform and,
what architects need to know.
And, , I feel like the table stakes
to, to call oneself an architect these
days, , continue to get higher , as more
and more people start to understand.
, the kind of standard data abstractions
that we've got in the platform, , the
traditional people do stuff to things.
Everybody kind of knows that now, right?
Everybody understands
how task tables work.
And, , it's a different world than the
one we started with, 10 or 15 years ago
when there was one handful of tables
and, , none of the different, , UI
pages , or methods or other, uh, data
abstractions that we have access to now.
So I feel like everybody's
got that started.
Duke: I just did a thread on
that on LinkedIn the other day.
, Joshua Saxton said, Hey, I'd love
to hear a story about service
now as it was back in the day.
And so I'll drop the LinkedIn, , link
in this description below, but it
was a lot of fun going back and.
Just realizing just how far
the platform would come.
CJ: There's so much that's different
now than it was when we got started,
but so much that's also still the same.
And, I think that's 1 of the,
that's some of the magic of
service now for me is that.
You can still pick it up, like having
started 10 years ago and still if you
understand the platform, like every new
thing that's built on top of the platform,
you still start off like, , what a head
start on it, on anybody, on anyone else.
, and that's what I like about it.
Ben: Well, so it not only has it
evolved, , from what you can see just
from , the interface layer, right?
15 years ago, it was
still just , one stack.
It was, , a Linux box that was running,
, Apache Tomcat with an application node and
a database node at the time it was, MySQL.
These days, , if you're in the data
center as a production instance, you're
still all the way down to the bare metal.
They manage those at , the
using Linux C groups, right?
So you own your stack all the way
down, but now instead of just a handful
of app nodes in a database node.
Depending on what you've got
installed, , you have another, , like
things like metric base, right?
Running , as a note in that same stack,
you've got, , potentially, , call outs
using integrations that use the Microsoft
Azure, , collaboration proxy that
goes directly to an Azure data center.
If you've got.
Depending on what kind of data you've
got coming in and out, you've got, , our
implementation of Kafka and Flink, , doing
data streaming out to other sources
or consuming Kafka streams, right?
Like your data , is not in the same kind
of, Simplified context that it used to be.
There's messaging buses and
systems and services and triggers
and all this and that, right.
It's a little bit more
complicated and robust now.
, and now you can even go and just buy
yourself a tenancy in Azure or in AWS,
if you want to just spin up service
now on a virtual stack somewhere else.
CJ: Wait, what?
Ben: Yeah, you guys hadn't seen that?
CJ: I think I missed
Ben: Oh, yeah.
So earlier this year, it was a
race to see who could announce
first Microsoft announced it first.
And so in Azure, you can go and
buy a service now tenancy and in
your Azure data center, you can
do the same thing in AWS now.
So , if you wanted to adjust your
spend, or, spin up additional resources
on your own, you can go do that now
in those other two data centers.
Now, it's not the same level of
service and system and excellence.
Microsoft . You're not
buying the Ferrari now.
, you're maybe buying the Peugeot
version of service now, uh, over
in somebody else's data center.
And you're also managing that risk
and that performance and your spend.
But yeah, we announced
those, , just earlier this year.
CJ: Wow.
Ben: It's kind of bonkers, right?
CJ: yeah, that's pretty awesome.
Right?
I can see that opening up a whole lot.
, we're going to have to come back
on, on that one, Ben and do a deep
dive on that one on another show.
That's that one took me, that's,
that took me away by surprise.
I got so many questions, what I wanted to
say though, is that I think it's amazing.
The tour you just get a virtual
tour of the data stack, right?
Of service down that you just gave us,
, from, talking about the under the hoods
aspects, because I don't know that I've
ever , heard it presented in that way,
or at least with that level of detail
and specificity around, , the Kafka and
Linux and all those sorts of things.
And, , typically, I think, , this
is a platform in the sky, right?
, and you don't really get below
that level , of service now.
So it's really cool to hear about, how
service now is running on actual hardware
and software in the data center somewhere.
Cause that's like my
form of life as an ops
Ben: Right?
Right?
And the, the Ops piece has continued to
evolve and grow, under the hood, , well,
the rest of , the App that we see the
interface layer that we get to, to mess
around with and develop, , in order to
support all of those cool new features.
And, , it's interesting to, think
about the way the new stack, has evolved
in the last five years or so, right?
With, a change in mindset and
an understanding that data has,
well, . Even 10 years ago, we
understood that data float, right?
And we talk about how people do, their
traditional like system import stuff,
and they bring data into the platform.
You do your extract transform and load
and whatever, but large scale enterprises
data is either consuming or moving between
huge federated environments sometimes.
And so you have to think about, , not
just the service now end of the spectrum,
but , you've probably heard people talk
about data warehouses or data lakes.
, and we, I think a lot of big
organization service now included
are starting to understand and
think more in terms of data meshes.
So you remember Frank Slootman, who was
our second CEO , , after Fred, right?
If you look at what Snowflake's business
is, it's about creating data streams
and data messaging and basically
your portability of cloud data across
a whole host of services, right?
So it's data as a service.
, I put a link , in the chat here in
StreamYard , for you all to check out.
, but it's one of those things
that really changed my
thinking about data governance.
Cause , , ultimately we can't just have
the data flowing in different places
and you can't just stick it in a data
lake, which eventually becomes the
data swamp because it gets stale and
yucky and you can't always rely on it.
And you're having to basically train up
and maintain a whole army of, , engineers
just to do and maintain all of those.
, ETLs, , or extract,
transform, and load jobs.
Instead, you got to think about it from
a messaging bus service system, right?
You got to think about
data as it comes in.
And, uh, it's a mosque Dhani, , who
I've linked to on the data mesh book,
, over on O'Reilly good old O'Reilly.
Right.
CJ: around forever.
Ben: But, I would check out
some of her thinking , on this.
, you'll see that it's influenced a
lot of, , systems , in the new era,
the new stack of just thinking about.
Everything that runs in our world these
days is all just compute, , and containers
and, , data , and storage, right?
That's it.
And managing that across multiple
systems, from an enterprise architecture
standpoint and from performance
standpoint, , we think, or at least I
think, we need to be thinking about all
of this stuff as we design solutions.
That's at the high end of the
architect spectrum, right?
But I think even table stakes,
like thinking about holistic
solutions is really important.
, Duke: I want to double click on that.
, holistic solutions.
Ben: So if you've got, a customer
service management, . Kind of thing.
If you're making choices about what to
customize, at what point do you decide
to make that make or break decision
to, to do some major customization
versus trying to maintain, , baseline
deployment of service now, right.
And it's about looking at the
value proposition, , what do you
get for doing a customization?
What does that buy you?
Are you buying a net new
capability that ServiceNow
hasn't figured out how to do yet?
Or is it more about, buying goodwill
in the organization to get people
to do adoption more quickly?
And do you do that knowing that
you might eventually have to move?
pay back some of that tech debt in a long
term fashion and go back baseline after
you've already bought the goodwill and
the adoption within the organization.
It's about trying to just decide, like I
said, , at a big picture level, using all
of those elements, a holistic view, right?
It's not just about treating the symptom.
It's not just about creating a
solution right then and there.
It's how is this going to
affect the life of this organism
that we call the enterprise?
You know?
CJ: Yeah, I love how you framed it about
around buying goodwill and adoption.
. Because for me, that highlights
and really elevates that.
A lot of your technical solutions
have non technical drivers, right?
And so sometimes the choice between
configuration isn't, always like just
detect that aspect of it, , can I maintain
this or should I maintain this sometimes
it's, if I can get this very influential
group inside of the organization on board
by doing something that's really custom
for them, then they will throw their
weight behind this thing internally.
And now we can get it moving.
. Ben: Sometimes you buy goodwill.
Sometimes the drug of
customization , is sometimes used
to, lessen the pain of adoption.
Right.
CJ: Yeah, but knowing
that up front, right?
Having that, thought in your head when
you're approaching these, , conversations
, and knowing those things up front, I
think are what allow you to make the
decisions effectively so that you are
managing that customization, right?
Without it, without
allowing it to manage you.
Ben: Totally.
And I think that more generally, I think
we have a better understanding in the
ecosystem of what those trade offs are.
, we know what we didn't know before.
And I think a lot of that , was
hard one learning, , over time.
Duke: I'm going to go ahead
and subtly disagree on that.
Ben: Oh,
CJ: Oh, yeah.
Duke: a certain layer of the most
senior have certainly learned those
lessons, but I feel like , everything
from the upper middle down is
still learning those lessons.
It's been the dark side
of the explosive growth.
, every couple of years, a new swath , of
service providers come on the scene.
They hire newer, cheaper talent who
learn these lessons, get so excited
about the platform and then it's like,
oh, let's take all these cool things.
We're learning, roll them
up into an accelerator.
And that's how we'll get our services to
go faster and more profitable per unit.
And, , it just feels to me like there's
a, Two to three year rhythm of people
who decide to sell all the stuff
they've learned as a single package.
you know what I mean?
One gigantic update set,
Ben: Right.
Duke: and, and completely power
screw somebody's instance.
And, if you think about this
year, next year, there will be
more new people in the ServiceNow
ecosystem than there are veterans
Ben: Mm hmm.
Duke: by a mile.
I'm not blaming this on anybody.
I'm not taking anybody down, but the
reality is the aggregate ServiceNow
platform is going to trend downwards
sharply downwards because we will
have more people that are very fresh.
Zero best practice.
Ben: I think that's
Duke: Not even like I know you said
don't do best practices, right?
But zero like hard lessons.
CJ: yeah, you know, a lot of what you
Duke: sorry, go ahead,
CJ: know, go ahead and
Duke: I was gonna say, the dark side
of the excitement about the platform.
We love this platform.
It can do anything is
to take a posture of.
Yes.
Let me show you how, like, can we do this?
Yes, absolutely.
Let me get started today.
And this is part of what you learn
in the, architecture frame it.
And what was the thing you said,
Corey, like controlling the build
before it controls you is that.
Yes.
You don't say yes to everything.
Okay, not in terms of like,
will we do, will we not do, but
also how we intend to do it.
I'll break out the Amish metaphor
again, if I'm not careful.
Ben: You know what?
No.
Well, so there's, a lot of validity in
what you're saying, and I think that
there's, there has been an intentional
shift, at least internally, , at the
Mothership to create programs and
systems and try to do the thing like
where we build the barn or raise
the barn, or like they do in, Japan,
.
Every generation, they basically knock
the temple over and then have all of the
craftsmen and artisans come in and you've
got a multi generational set, right?
The master gets to oversee the process,
having been a journeyman the last time
the temple , was rebuilt and so on, right?
To continue to share that
operational kind of knowledge
that's only gained from doing.
And I, think that you're
starting to see more and more
programs try to address that.
And.
I've always seen the accelerator.
I know that it's kind of been the bane of
a lot of, , our existence having to come
in and clean up after an accelerator,
but in some ways the accelerator
is like the starter house, right?
It's like the cookie cutter, small house
that somebody builds and the contractors
build it that way, cause it's really easy
and repeatable, but the person framing
up or putting up sheet rock in that.
, starter house doesn't have the
kind of visibility of, of how that
house is going to live over time.
Right.
And that's why they're kind of a
little bit on the junkier side and
people have to come in and fix it
or, or do a renovation or somebody
just moves up to the next house up.
Duke: that were true, I would
say It's a tiny house where
the template was a hoarder.
Okay, so it's not only you don't you
not only you not only get the frame and
the utilities and all the electrical
and all the driver, you get all of that.
But you also get 30 years of newspapers
Ben: Mm hmm.
Duke: and fast food, take
out bags and aluminum cans
Ben: And, and to your point.
Yep.
Duke: And that's what you get.
Ben: And somebody thought
it was a really good idea.
Like, Hey, let's put an
electrical outlet in the shower.
Cause maybe that'll be helpful for folks.
Right?
Duke: yeah, exactly.
CJ: I love it.
I've been thinking a lot , around,
, consulting and service now and
how all this stuff comes together
, and, , do some of this stuff, right?
For me, like this deals down to this
really, really simple 3 words and
that's that consulting doesn't scale.
Right?
And I think a lot of this is
really because after a certain
point, as a consulting firm,
you really got too many clients.
? And the solution that folks invent
to that are like those accelerators
and things of that nature, right?
Because now they're trying to get
in and out as quickly as possible
and trying to deliver You know, a,
at least somewhat uniform experience
between all their different clients so
they can try to maintain a standard,
.
, but that doesn't work, right.
Because the very definition of
consultant is ad hoc, right.
Nobody's the same.
And so when you go in and you try
to deliver the same for people who
aren't the same, like you just end
up , in this situation where you're
delivering great to anyone, you might
be delivering good, but often you're
delivering average, sometimes poor.
Right.
And so I just think , Where the
industry is going and how do you get
there and bring in and like when you
said like new service providers show
up and then they spin up a bunch of
new resources and clients and go in.
The first few are always great.
? Because they're still
below that scale selling.
Right?
Like you can go in and have, I don't
know, whatever , the magic number of
clients is where you can deliver basically
white glove service to all of them.
But then what?
But then that makes you popular.
, and then you ultimately like
your reputation outgrows
your ability to deliver it,
Ben: Well, and I think at that
point you have to decide what
lane you want to be in, right?
You can't be all things to all people,
just like the instance can't be all
things to every department and everybody
that wants it at all at the same time.
And I've found over the years, if I want
to have a reputation for excellence,
then I need to be more than willing
to say no to more people as that
reputation is gained, , and therefore
I need to charge more for my time.
CJ: Yes.
Ben: I think that , there is a place
for , all of the stratification, right?
If you, if good enough is just meh, right?
You get the , average
brown bat experience.
There is a place for consultancies
to fill that lane, right?
They, get that through volume and
through massive scale, but you
know that you get what you pay for.
You're taking the least expensive clothes
off the rack and you're just wearing them.
You're getting the 7 haircut.
You're getting the gas
station coffee, right?
If you wanted to have , the higher
tiers of service, I think that
there's a set of expectations
that, that have to come with that.
And I think there's room in
the ecosystem for all of those.
CJ: No, totally agree.
, it's 1 of the great things about the
ecosystem is that it's so vast is
that there are a lot of different
lanes that you can slide into.
Right?
And really carve out
a niche for yourself.
You know, I just think I don't think
often that, , folks in this and who are.
Planning out where that niche
is going to be or thinking about
it from this perspective, right?
Like I think about , I need to be
high end and they charge more , and,
I'm going to offset that by having
fewer clients because this is the
level of service that I can provide.
Right.
And versus , I'm really good at this
and I've run into this , trap, right?
Like I'm really good at this.
Let me try to give it to
as many people as possible.
And you ultimately
commoditize something that's.
Better than the average, right?
And you burn yourself out and
ultimately deliver a poorer product.
Ben: Well, and I think the hope is, and
I didn't learn this until much later in
life, and , I still regularly relearn
this lesson about no, and about what my
time is worth, And getting to a point
where I understand that, okay, I've
already leveraged that sweat equity, and
delivered more than the customer paid for.
But the hope is, is that buys the
reputational excellence that I can
then start to charge more and be, and
have a more reliable clientele or have
a more reliable, stream where people
are going to come for the reputation.
Cause they know that you
are par excellence, right?
, you do have to kind of.
Sell that equity or burn up that
equity earlier in your career.
And then you have to find
out , when to make the transition.
And I think that's the
part I still struggle with.
I don't know about you fellas.
CJ: Yeah, I've got a buddy and
so we've said charge more enough
down that I'm going to actually
go ahead and give him a shout out.
, so Thomas is a good friend of
mine and him and his buddy.
Buddy Patrick McKenzie, AKA patio
11 are all, huge in, , hacker News.
And one of the things that
they're huge about is talking
about charging more, right?
, once you understand the worth of
your time, like charge more for it
and deliver better to less, right?
Instead of delivering, worse to more.
I learned that over and over once a
year, at least, Ben, honestly, right?
Because all projects look.
Some of them look really exciting.
. And you want to get in there
and you want to do the thing.
Right.
, and it's like, well, no, you can't
pay me 15 an hour to do it though.
But it looks so cool.
Maybe I want to.
Ben: Right.
Right.
Duke: for me is when you can, when
you can understand what an excellent
solution is and know that it's not
necessarily what the customer asked for.
That's going to sound completely wrong.
But you know how, like, when people ask
for stuff, they can't really help but
inject their idea of how it should be
Ben: Yes, yes.
Duke: and if you can , basically
extract the outcomes out of whatever
it is they're asking for, extract the
outcomes, and get them to the outcomes.
You know what I mean?
Well, navigate while navigating that
hole, but that's not how I asked for it.
CJ: Yeah.
Yeah.
Duke: Once you get to that state,
then you have a consultative
mindset because , you can command
more trust in the instance,
Ben: Mm hmm.
Duke: hey, I know it's not what you
asked for, but this is going to get you
closer to the outcome that you desire.
So you have the soft skills
of convincing people to do it.
Your way on top of the, , scar tissue
of knowing what the better way is.
Ben: Right, right.
And this is where some experience
helps and or the ability to leverage
the experience of others that you
can refer to and say, Hey, , I
hear what you're asking for.
Let me tell you that
I've seen this go wrong.
I've seen this break bad.
In similar situations, here's
where I've seen in the past this,
how this can go the right way.
And let's talk about what that
looks like, in your situation.
And those discussions are hard.
, because I think people get, they
get anchored biased, , by , the
thing that they really want or
that somebody told them like where
the conversation started there.
And they think in their head, they've
already got the solution and they come
to you to simply just implement it.
And those of us who have been
around the block long enough know
that when it comes up and , if a
client or a customer is not able to.
To take our advice.
Well, you know what?
We can always say no, right?
You come to me for the, and I used to
run in this in creative work too, right?
People really had this really strong idea
and I would go, okay, well, , that's not
something that I would provide for you.
Cause that's not what I do.
Right.
If you want that kind of style, or
if you want that kind of thing that
you want to go to these other people,
but they'd be like, no, no, no.
We heard you're really good.
We want you to do it for us.
And I would be like, well,
part of why I'm really good.
, is that.
I maintain that standard through being
able to say, , here's what I know works
and here's , what I know would work for
you or I believe would work for you.
And if you don't like that,
then, I can always take
another client is the thinking.
And that's a level of privilege
that it took a long time to get
to and eating a lot of ramen.
, to be able to maintain,
for in the early days.
CJ: Amen to that.
Right?
Like, I think one, I think you really
just define like what as consultants
we get paid to actually do right.
Like we don't get paid to say
yes, we get paid to say no.
Right.
Ben: right.
You're buying our expertise.
, not necessarily.
If you want somebody to just do the work
again, I'll take that house metaphor.
If you want just somebody to do what
you want done and you want to design
it yourself because you believe, you
know, best, just go hire a framer or
contractor and tell them what you want.
Like you might.
You're not going to get , the level
of expertise where you're going to
get potentially happiness out of what
you just built, or you're going to
have to learn the hard way, , that
the things that I was just trying to
explain are like, here's, what's going
to happen when you build it that way.
Or , if you design the solution
this way and , that's cool.
Like you, maybe you can get it done
cheaper because you're not having to
pay for that, that hard one expertise,
and maybe you'll be happy with it.
But I'm gonna you just guess it
probably you're not gonna be right
and you're gonna have that little
bit of regret and going Like man, if
I'd only just you know, actually paid
for the expertise to do this work.
I wouldn't be in the situation.
I am now
CJ: And
Duke: know, you know, what's
crazy though, is that it's the
whole payment for expertise thing.
And I'm not saying it's all partners,
but there are more than one partner where
. The expertise, how do I say, right?
They get, , they take
their mid tier resources.
They slap a senior rate on a senior
resource who isn't really that
senior and a company gets engaged
with the whole partner, right?
So they say, this is the
partner we're going to use for
all of our service now stuff.
And it's like, all of a sudden,
I'm getting, like, the fraction
of a fraction of a air quote,
senior resource at a crazy rate.
and on top of the wrench turning rates.
When there are,
Ben: That's a good way to
put it wrench turning rates.
Duke: yeah, but there's, tons and a lot
more every day people like Corey and I,
who do like freelance vendor, agnostic
architecture, just toot our horn a
little bit where our objective is not
to put 1 more billable resource on your.
Cue, , our objective is to make sure
you get the best result and you pay
a little bit more for it, but you pay
it to not be sold some new services
and to only obsess about your outcome.
Almost like a lawyer relationship,
Ben: right,
Duke: you know what I mean?
CJ: Like a lawyer relationship, right?
Like we're looking out for you.
That's the contract, right?
Like your lawyer, when you show up in
court, your lawyer can think you're wrong.
Right.
But they're going to zealously defend
you to the best of their ability anyway.
, so it's the same thing for me.
Like, when I show up to a client,
, I'm absolutely your person, right?
, you can be engaged with the partners.
Actually, a lot of times
I prefer that, right?
Because there's sometimes there
are some projects that are bigger
than my ability to deploy right
within a reasonable amount of time.
And so.
Okay.
Call me in and let me help you get the
most , out of the investment that you've
placed in other partners or call me in
before you pick the partner and let me
help you pick the right partner for the
thing that you're trying to get done.
Right?
Because I can do a lot of stuff,
but I'm not the best at all of it.
Duke: Or just act as a foil,
CJ: Yeah, that too.
Ben: right, right.
Duke: I think 1 thing customers
don't understand is that people
who are employed by the partner
to do the work at your site.
Are beholden to the partner 1st.
And so, like, you have, you have
to tell your employers line, right?
Not your employers customers line.
CJ: I've been trying to figure
out how to frame that dude.
That is so, that's so great
because that's the thing.
Duke: yeah, and that's why it helps to
have, , an advisor of last resort almost.
To say, okay, yes, what
they are suggesting is a way
here is an alternative way.
Ben: I like that.
I like that a lot, right?
It's the difference between, , the
fiduciary responsibility of a CPA, right?
Who's, who's managing your books and
helping you make good decisions on
taxes and other investments versus,
calling up the investment firm and
they're there to sell you a product
and make a commission, right?
It's, Yeah.
CJ: no, this is exactly it.
That's exactly it.
It's been, , you reached
out, , earlier in the week and
just a , little quick pivot here.
Like you reached out early in the week.
You say you were yelling at us through
the, uh, through the speakers in
the car, like on the last episode.
So tell us a little bit about that.
Ben: So, uh, you know, obviously
everybody, the year 2023 has been, , the
year of keynote speeches where somebody
just walks up and says, they tap the mic.
They're like Gen AI.
And then like,
Duke: bye.
Ben: and then like the crowd
goes wild and they drop the mic
and they walk off stage and.
I think sometimes when we get into
talking about what is the architect's
responsibility, , in this day and
age , of where Genitive AI fits,
, it's more that kind of co pilot.
I think too many people were
getting really excited about
being able to replace resources.
, it also isn't , the kind of
10X, , productivity booster that I
think everybody makes it out to be.
You still got to have the kind of
overall vision and understanding
to play this at this level, right?
And you got to know where it fits
and where it can be deployed.
And you'll note that ServiceNow
has been , really not slow moving,
but slow moving by some standards.
I think people have been like, hey,
you guys are late to the ballgame here.
But I think we've been really careful and
thoughtful about where it is appropriate
to deploy and really testing and making
sure that when we put it out there, It's
not going to produce bad outcomes, right?
, we're only putting it in the places
where we know it can be successful, , at
, the level of quality that, we want.
And I, I know a lot of people are really
excited about what's possible, but this
is one of those things where pivoting
back to some of the data governance stuff
and complexity and thinking about where
your data lives , and where it goes.
If we're training these large language
models, where, , who's holding on to the
training data, , which APIs are we using?
Are we using, , our own internal LLMs
versus, , are we leveraging , the
Azure OpenAI, , API calls, , are we
doing it over the open internet or are
we using something like, I referenced
earlier, the Microsoft collaboration
proxy to do it privately between, your
ServiceNow stack , and the Azure API.
So it's like, Making sure that the people
that are playing at our level in the
game have an understanding and , can
speak intelligently about some of the
concerns that people might have about
privacy, , where their data residency
is, , being able to speak to what's
being, what their data is going to be
used for and how it's going to impact
some of these next generation tools.
So I was yelling at the radio about that.
Some of that.
CJ: And regulations, right?
Like regular, like we can't leave
that one out because regulations,
cross border data, , data access
and things of that nature, right?
, I know we're not particularly strict about
this in the U S but everywhere else is.
Oh
Ben: latest executive order, , basically
put an end to what we used to jokingly
call, bring your own API, , where
GCC and GCC high accounts were
getting all juicy about trying to,
and that's the government, regulatory
like FedRAMP , and DOD and GovCloud.
People were getting all
hungry for , Ooh, cool.
We can do more with less so we can,
start leveraging some of , these APIs.
And, , that executive order was
like, Hey, wait a minute, , let's
just slow our roll for a second.
Where is this data going?
Where is it living?
, and that kind of put an end to that.
, cause there were some legit , concerns
about where this data might end up , and
who could see the training data and.
, there's just questions I think
that need answers first before
we can start rolling with some of
those tools in those environments.
Duke: There's almost like a direct
relationship between what you just said
and the amount of hype for something new.
Ben: Mm hmm.
CJ: Yeah.
Yeah.
There
Duke: harder they want you
CJ: on that.
Duke: like the harder they want
you to adopt and, , be part of the.
Excitement capture, the more I can
guarantee there's , really important
stuff, , that hasn't been thought of yet.
We don't even know
where, , the booby traps are
Ben: Right.
Duke: right.
A few months ago, I'd say, . When AI
hype just seemed like every day there
was like 10 YouTube channels that would
advertise, here's like 15 AI tools that
came out this week that you need to
get or else you're not going to keep
up and you're going to be poor, right?
Ben: Right.
Duke: After you get past all of that and
the just absolutely dizzying, hypnotic
marketing megaphone screaming in your
fucking ear, after you get past that,
then you get the whole things like, Oh,
Well, we put all of our, , company IP
to be trained by this thing so we can
help us design products and now people
can use a clever prompt and ask this
AI to , give them my trade secrets.
And the marketing people don't stop for
1 second to consider that because don't
think about product, just buy product
and then get excited for new product.
Ben: Right.
CJ: Yeah, I mean, it is, it's definitely
one of those things , where are those
downstream, , vulnerabilities, don't
necessarily surface in the marketing.
Right.
Duke: I, I do want to say though,
quick, Corey, sorry for interrupting.
Like after I say that I have looked
at what ServiceNow is doing, like for
the things it intends to use Gen AI
for, I am legit, legit, super stoked,
Ben: Right.
And we're doing it in a safe and
thoughtful way because, we've got
people much smarter than myself even
looking at this stuff and thinking about
all of these really hard questions.
, and making sure that we are
being reasonable about it, right?
We're trying to move fast and
fix things, kind of Francis Frey
style rather than break stuff.
, and we're not pushing, we're
not pushing the hype cycle.
Cause if you look at some of the
hype cycle stuff, the only people
that really benefited from that
were some of the YouTubers and the
streamers that were making lots
of, , ad revenue clicks, right.
People watching their thing to
make money fast using Gen AI.
. And none of the people that
bought into that made any money.
The only people that.
, got anywhere were the ones making
the videos, convincing everybody
else to make a lot of money, , using
Jenny and nobody else made money.
. Funny that works that way.
And the only, but only people that
benefited were those handful of
influencers and the companies with
the training data that made really
good use of that training data
where everybody was pushing it.
But yeah, , we're moving a
little bit more thoughtfully.
And I think that's the differentiator.
Somebody that wants to claim
enterprise architect is thinking
about the big picture questions, the
big risks, not just, can we do it?
Should we do it?
What are the trade offs?
What are the risks?
, let's.
Weigh this out.
CJ: Yeah, and those are things
we didn't have to think about
like way back when, right?
, because everything was really
confined to the instance, right?
, and you just, you really just thought
about, , and it was mostly it only
even , back then, so there wasn't
a whole lot of business, , data.
Yeah.
Yeah, scope was smaller, right?
Like, , so the vulnerabilities that
could arise weren't necessarily
enterprise risking, right?
But now they are,
Ben: Yeah.
We don't have to think about the
data government side back then.
Duke: I can't believe it happened
again, but we are 40 minutes in for
Ben: love it.
I love it.
, and thanks for letting me, be old
man yells at cloud for a minute.
, and hang out with y'all.
Duke: sure, man.
How will the youngins know otherwise?
Right?
Ben: It's true.
It's true.
Somebody has got to do it.
Duke: Yeah.
All right, folks.
This has been episode number 99.
We still don't have a proper outro yet.
, but we are done for the year
and, , we have got something
special planned for episode 100.
Please keep the eyes on the feed,
, because we're hoping episode 100 is
going to be a big blow up for us.
So, , can't wait to show
off what we got planned.
CJ: All right.
And that's a wrap.
Duke: All right, cool.