440- When Knowledge Stops Being the Gatekeeper w/Leonard Boord

Doug Camin: Welcome back to today's episode of You've Been Heard, the platform
for IT leaders. I'm your host Doug Camin. And today I'm talking to Leonard Boord, the
director of IT at Carrot Express. Welcome to the show, Leonard.

Leonard Boord:  Thanks. Pleasure to be here.

Doug Camin: So before we jump on our session here, I see you're a big user of AI
tools here. I see a lot of posts about other things like that. You're like, hey,
I ran some AI stuff to give you some, kind of like formulate, the things we
might discuss and things, here's my content that I might have and stuff like
that. Tell me a little bit about how you're using these tools to change the
nature of the work you're doing.

Leonard Boord: Yeah. So, we happen to be at a point in time where humanity has
this incredible opportunity to be insanely productive, more so than ever before.
it's kind of like the invention and it is like a brand new tool. So when the
hammer came out, hitting nails was super easy and before you had to use other
tools and it was complicated. Best effort. So we have tools today like cloud
code, and that's the one that I tend to lean towards the most. And having access
to cloud code or GPT codex, whichever, pick your poison has enhanced everybody's
overall productivity to an insane degree. So one of the things that I try to do
the most is build it instead of buy it. It tends to be a lot cheaper. but what
this has also led me to see is that the horizon is a lot less stable than what
it's also ever been in human history, right. So, I'm a big believer that if you,
scale this out and I don't want to say far enough, but if you scale it out
three, five, ten years, it's very difficult to see who's going to still be a
player in the game. So a good example of this was nearly the end of my time over
at MC A, we started getting into ticketing systems because there weren't a lot
of cases, there were a lot of emergencies. So it was more you have to go on
site, there isn't time. So we couldn't really triage remotely, although we would
do the best that we could. But if you have to travel from one airport to another
airport, there's little that you're going to do on the phone.

Doug Camin: Mhm.

Leonard Boord: So I started getting into the world of ticketing and this is many
years back. And I guess the term is called ticket triage, right? Where you have
to get your own tickets. You have to audit them. You have to make sure that
they're consistent, that they have good data or you get bad data, fast forward,
to closer to today and after using many different ticketing systems like Jira,
Trello in the past, I converted right into a ticketing system. You can grab
monday.com and you can turn that into a form of ticketing system. You can use
Microsoft Forms and make ticketing systems there. And there's many ways to set
up ticketing systems. and I'm sure there's many good ways and many bad ways, but
at the end of the day, if you ask me, their value is in the ability to audit
them, extract the data, determine what is causing fires and put that out.
Otherwise you continue to have fires. There's a lot of people who like the idea
of fighting fires. And it's ticket in closing, first in, first out, first in,
first out. But if you start to notice that, and let's just say like Miami is
suddenly getting fifty percent increased ticket volume. The question is why? And
if you're just fighting fires and putting out case after case and you never get
to the why. That's when I've seen teams just grow outlandishly, but they don't
solve the problem. So then suddenly it's Miami. And since I live in Florida,
then it might be Fort Lauderdale. And now you're seeing things that start to
spread. But if you don't solve the cause, and it could be something silly, like,
an integration failure that's gone unnoticed. So after a while, every time you
reboot it, that problem doesn't come back until a specific condition in the
integration triggers. And nobody's aware because nobody's looking at the data.

Doug Camin: Mhm.

Leonard Boord: So you would be able to say things like all of this type of one
POS is suffering. Okay, great. So I know it's that one vendor. Oh, but it's not
just that vendor. It's a specific device in that vendor. And then okay, so now
you have vendor device and then you start asking questions like, what are the
conditions under which you've seen this happen? And when I was at Popeyes with
three thousand locations, if you have somebody who's constantly calling out,
hey, it's every time that a credit card times out after 60 seconds, the pin pad
device starts to get impaired. Oh, okay. So now we know it is this POS vendor.
It happens to be this terminal, and we know that we need to investigate this
condition. So then you bring in the other vendor who sells you that. But anyway,
so the case in point is that ticketing it sounds familiar.

Doug Camin: Of course the classic problem we run into all the time.

Leonard Boord: Right. So and I, went off on that tangent because we have tools
today that we're all built. And it's no fault of their own. Like, the ones that
I mentioned, Trello, Jira, Zoho Fresh service, etc. that were all built without
AI at the forefront. So I took it upon myself earlier this year to kind of just,
hey, how difficult would it be with cloud code? And I think this was with opus
four point six, to stand up a ticketing system that is the core product, an AI
tool, but you can turn off the bells and whistles and get back to an original,
what I would call a dumb ticketing system, since it doesn't have like agents and
AI capabilities or intelligence scripting or routing.

Doug Camin: Mhm.

Leonard Boord: And it turns out that you can do it in about three weeks. And mind
you, all of these tools are bespoke, right? So it might be unique to you. It
might be something that could be sellable to the masses. I know there's a lot of
litigation and things that come into play around eighty percent sweat a brow,
etcetera. but what that enlightened me to see, at least from my perspective, and
maybe I've got a hippie too. but there's a lot of uncertainty in the future. So
with the ability of standing up that agentic system, right. So it has the
ability internally to scrape documents. And I tested it with across a ten
thousand documents I scraped, and that was from the entire Microsoft repos or
libraries that are currently available from tools like restaurant three sixty
five bill pay, Paytronix, square toast, etc.. So for a camera system, I just
scraped all of that with just a mundane scraper that clod built embedded it
through voyage. because I like anthropic, that is.

Doug Camin: Mhm. So you just talked a lot about a lot of these tools you're
using, both the way you're using it. But right at the beginning, you made some
statements about the changes that you see coming. And So there's a lot of
schools of thought here in terms of the changes these tools are going to bring,
what they are, what they aren't. Is it hype? Is it not? I think it sounds like
it's fair to say that you're operating in the space of like in ten years. Like
half of us are going to be wondering what we do with our lives, type of deal.
because there's a tool, it's a bot that was able to replace a lot of the,
fraction of the daily tasks or all of the daily tasks or some context here. So
I'd love for you to elaborate a little bit on that and what you see as the
future of this space.

Leonard Boord: So it won't be the first time that, we as humans go through a
transition like this, right? The whaling industry vanished overnight. We used to
be ninety seven percent farmers and now we're three percent farmers. So these
transitions are commonplace, if you ask me. they're large, epic events that
happen in our history. And if you research them, you can find them. I don't
think that we're necessarily going to be wondering what we do. I think it's
going to be a change in what we do and how we do it. So for example, knowledge
won't be the resource that is, let's just say the gatekeeper to many things.
It'll be creativity. It'll be, how can I use this tool to figure out what is a
problem that I can solve for somebody? Or what is a pain point that I'm
currently having today? And how can I solve that for where I'm currently at
today?

Doug Camin: I think that aligns somewhat where I'm at too. Like I look at the
tools here and it's funny, I end up owe myself, I end up with bosses that are,
like, who could we replace with this? Or what can we do with stuff like that?
And the answer oftentimes is not who can we replace? It's what can we make
better. So, if we've got five staff on the help desk, how do we make that five
staff handle, twenty five percent more tickets without burning them out? so
like, the goal here is not to somehow just burden them with more work, but if
they had through the process of the work that they were doing, we could identify
a series of menial tasks that they were, not value added, adding less value, the
process of, coding up and closing out the ticket for just as a very basic
example, could we shave fifteen percent off of that? And if we can, then that
means that they could probably handle a half dozen more tickets in a day and be
comfortable doing it.

Leonard Boord: Right. So, over a carrot express, we have a very lean IT
department. It's just kind of me and a couple of MSPs with a couple of vendors
for boots on the ground. Right. So it's kind of like a quarterback situation.
and that's why I really started on this ticketing system because there's a lot
of low hanging fruit, a lot of, we can call them like tier one or layer one
issues that, are maybe not their tier one issues that are coming in that are low
hanging fruit, they're easy to do. They're very mundane and a simple document
can walk just about anybody through it. And if they've got a couple of
questions, it doesn't need to involve necessarily a human, although a lot of us
do like to talk to a human. Mhm. So and that's what I built with the ticketing
system. if you guys want to check it out, feel free to post it if you'd like.
but the idea was how can we grab all or almost all of the low hanging tickets of
did you reboot or did you plug it in? are the lights there? have you tried this?
Is the setting enabled in the POS? If it isn't, how do you get to it? And that
was, in Excel. How do I create a particular type of formula. And the way the
ticketing system was resolving that was, is that there's, I guess kind of like
the secret sauce is prior to fetching the embeddings, in the loops, it goes
through a tagging system. So the way that the process kind of looks is you
scrape it, you tag it, and then you go into the chunks that are tagged. And the
nice thing about that is, is that it saves time. So I saw savings and it was
something like thirty to 40 seconds while it was trying to find something and
bringing it all the way down to seven to eight seconds. So it feels fast. And
since it's using haiku, it's not causing a lot of API burn in terms of dollars.
So the idea was can haiku first find the right document, surface it, and then if
a human wants to review what's happening live, it detects that, oh, hey, I've
been opened by a human. I'm still going to behave the same way. But the moment
that the human decides to intervene, I now start coaching the human. And he can
have a separate chat with an agent, which is the I call it Atlas, but he can
have a conversation with Atlas or the XI and Atlas will provide them feedback
based off of the context of the ticket. So then as you're going the agent could
say, hey, Atlas, we've exhausted all of these things and it's continuing to
happen. could you find a couple of other related articles or provide some other
suggestions based off of the ticket summary? And you would send it, and then
Atlas would go ahead comb through the entire case. And if it needs to, it would
escalate to Sonet. And Sonet would then start inferring additional things to,
keep costs down. And, it's connected to Twilio. So people could phone in as well
and they can have a conversation with Atlas. eleven labs is included in that
process. So it was a very cool end to end project and I found a lot of value in
it.

Doug Camin: Cool. And so see there are two things I want to make sure we move to
a different kind of general topic. But one of the things I wanted to just
highlight for our listeners is the space that you occupy. You're in a retail
food service. and in that space, you mentioned about the, additional things, the
joke at it is like you call it, the first thing you get asked is like, did you
reset it? Did you unplug it? Is it plugged in? Is it turned on? but in your
space, like depending on the space. So, I've worked in, many years in county it
currently for a nonprofit. Like if my folks ask the people if the thing is
turned on, they will be mad. they'll be like, what kind of idiot do you think I
am? But in your space, like that's a, real, legitimate question because you're
talking about people who are in food service don't know anything about the piece
of machinery that's sitting in front of them, other than the fact that it's
supposed to like spit out a receipt or the screen is supposed to be on, and I
don't see it on. So like their troubleshooting skills well, it's a retail, low
wage environment, people are churning, you've got new people all the time. in
your case, asking them, is it plugged in and are there lights on is a fully
legitimate question that probably resolves, some large fraction of the tickets
that you have sitting out there.

Leonard Boord: Yeah. So you have to put yourself in their shoes. Right. and
whenever I work with an IT team, especially in the food and beverage, you call
it retail QSR, fast casual, I think is the.

Doug Camin: Thank you for clarifying. Yeah.

Leonard Boord: the thing is you have to put yourself in their shoes. So I, I
normally like to grab the IT individual and I put them two or three days inside
of the location so that they can kind of see the flow like, hey, be the cashier,
have a cashier beside you so that you're not, fumbling, but when you've got
fifteen people in line and the person in front of you is, trying to place their
order and the system is locking up, freezing the check isn't coming out. They're
demanding the check. That's the stressor and the pressure that they're under. So
they don't want to like, yes, I turned it off and on. they don't want to think
about that. So when they get on the horn with you, there's a couple of questions
that you, tend to kind of feel out. And the first one is, hey, listen, is it
actually turned on? Do you see lights? And if they say no, okay, now we know we
need to triage power. do you see lights? And there's no connectivity. Now we
know we need to triage that and are they plugged in? Are they seated properly on
both ends. And that will resolve the lion's share. But then you get into
trickier things where how did the system get into test mode when it requires
only a manager access? The manager wasn't on site and that is more of an
operational opportunity. And where we have to teach people not to share their
pins, because that is highly indicative that a PIN was shared from a manager or
somebody guessed it. And then we have to go into, IT security. And now we got to
go into best practices and the why did this happen or who did it? And then you
got to review cameras and I'm sure you're aware of all of that fun stuff.

Doug Camin: Mhm. Yeah. So all right, that's great. one of the things that when
you work in different industries, every industry has kind of its own
terminology. So yes. you're the fast casual space that's actually a great segue
to what I want to talk about next, which is, you and your history. So you've
been in this space for a while. Let's walk back through time. You're currently
the director of it for Carrot Express, but you've been in the fast casual food
space, it looks like for a while. can you tell us how you got into it, where
you've been?

Leonard Boord: Yeah. So professionally speaking, I, started at a golf store, so
on that tangent, when we had a simulator in the store, it was always fun to kill
some time to just, swing at, a club at a ball and just kind of see where it
would go and pretend. Oh yeah. Tiger Woods. Mhm. But, so I, started there and I
actually, I kind of started like muscle, if you will. So I, started in shipping
and then I moved over to inventory. And then from inventory, there's got to be a
better way. and that led to me investigating for computers for that company,
believe it or not, it was like two thousand and seven, two thousand and nine,
and they were doing everything on pen and paper, which is wild.

Doug Camin: Mhm. Mhm.

Leonard Boord: so we, got them into computers. And then from there, somebody
needed to own the systems and the architecture. And I naturally wanted to move
into that space being as a hobby. At that time, I was an avid gamer. So I loved
computers and it just felt like the natural next step. And then from there, I
did a hop over to a seven years over at master concessionaire and I started as
break fix. And it's funny that they won. I remember speaking to my boss and he
was telling me, hey, we've got a system down in Fort Lauderdale. Good luck. Talk
about birth by fire. so I drove over to Fort Lauderdale. I'm like, all right,
I'm here. I see the system is on. It's got power. All the cables are working.
what is the natural next step? What do you want me to do? What's the playbook?
And you just gave me a support phone number. Call these guys. and that's kind of
where it started. So I spoke with them. We eventually got the system up, took
three to four hours. There was a hiccup in an integration at some point and the
system was just hard down. if I, remember correctly, we were using at that time
something like Eve restaurant and then we transitioned to a different POS
provider.

Doug Camin: Mhm.

Leonard Boord: But yeah, so that was master concessionaire those days. And I
started as break fix and then I slowly worked my way up into it project manager.
And then I made the hop over to VI and then I got poached by RMG and now I'm at
karat. I'm happy to go into more detail in any one of those hops, if you'd like.

Doug Camin: Yeah. So just tell me a little bit about, so your first role as the
like name, I call it name director of it. So what have you brought to the space
that you feel is like the differentiator? and I'm asking to give you context
here. we got it leadership podcast. We got listeners here that are looking at
coming up. Some of them are in in the space already, but there's a number of
them who are coming up. So like what did you bring in? What did you say? How did
you get into that first IT leadership spot?

Leonard Boord: Yeah. So the biggest thing that I would say that I, brought was
system understanding, right? So a lot of what I've seen and when I was at RBI
franchisees is that they have these Frankenstein connections, where point apps,
I believe is what the industry term is, right? So I want to have a labor
scheduler. So I get the best in breed labor scheduler Oh but I want an
accounting system. I got a best in breed account. So there's tools, ERPs that
kind of try to reconcile all of these systems.

Doug Camin: But everything's got its challenges for sure. And what you're
describing about like best of breed versus consolidated. Actually, I went
through this where I'm at, now where we started with, siloed applications. And
then the question was, how do you get to something consolidated? And we looked
at NetSuite as the solution for that, because it was the only one that truly
offered, at least in the context of our business, a back end that was unified.
and I've always made this joke about the flow goes back and forth where in it
where like somebody will come in. siloed systems, they'll be like, you know what
you need, you need a consolidated system. And then when you get to the
consolidated system, somebody will come in and be like, you know what you need?
You need a specialized system to do this and this and this, and then you. So you
kind of ebb and flow back and forth as the different people come in and out.

Leonard Boord: Right. Well, so a lot of it is because people love a legacy
system, and they love a system that they got accustomed to over five years. And
if you ask them on day one, if you could go back in time. Time travel and speak
to them, they would tell you on day one, they hated that system. And then after
five years, they learned the system, and now they love the system. And then
change happens and now they hate the change. So, step one was and it takes about
to do it well and to do it right and to speak to all the different departments
and all of that jazz. It takes about three months to kind of understand what
everybody's real big pain point stresses are to build a plan and prioritize that
plan adequately. I'm sure some people can do it faster, but I, like to get very
detailed and I like to have those meetings and get the C-suite engaged so that
we have a good flow of what we're going to tackle and why we're tackling it,
because things will pop up, things will try to derail you. So that was stage one
and stage two was are we making data driven decisions or are we making emotional
decisions? Because emotional decisions are great. That's how you get that hunch
feeling. And sometimes you really hit something like you make a viral post or
whatever. and one of the, things that I've consistently seen is that depending
on which company, but it's all the same when I ask them, what is our number one
or top ten products, the data painted a different picture. I like X, Y, and Z
because it's my favorite or it's got this top, where it tastes like this or it
hits a craving, but your craving may not be your public's craving. And when you
audit your menu and you look at an actual pea mix and you start to make more
informed decisions. things tend to go better. So outside of understanding
problems, kind of like a parallel path happens to be data. Do we have it know?
Can we get it? Yes. How do we get it. And in the past it used to be a lot of
SDS. you need software developers in order to build out these APIs. And now with
cloud code or, a few years ago, you could use a g T before, anthropic got
really, into the top spot. And what I would do with those at that time is, hey,
I'm using this system. I'm very curious. first, do they have an open API? And if
they would say if it would come back and it would say yes, I would say, great,
help me plan the architecture on how I would extract this or that so that I
could be optimized within reason. Because you're always going to have pitfalls
because you're ignorant on these things. To a certain extent, the person who's a
full stack dev may not understand operations and operations, has no clue what's
going on in full stack. software developers may not understand the end user. And
this happens a lot. So, figure out how to get your data, figure out what's the
best way to organize your data, figure out how to store your data and have a
game, just a general game plan when walking into a new company. Hey, how are we
making, the determination on what product we want to sell, what product we're
adding? Why are we doing it? Figure that out. This is at least the fast casual
side. and you can do it for retail as well. When I was over at MCA, they had, I
believe it was half of it was retail like you would concessions if you will. And
they had worldwide market, Miami gifts to go.

Doug Camin: Okay.

Leonard Boord: and then you have, a fast casual. So you would have non full table
service restaurants. So we did have Chili's when I was over there as well. So
you had that as well. Mhm. so figure out the pain points, figure out the data
situation. And then how do you surface it? and the biggest thing is have a plan
to keep costs down. And today this goes back to part of our initial conversation
is it's never been easier to keep costs down. Never, never. And it's incredible.
You could rent a virtual machine on OB for twelve bucks a month, and that thing
will be a stallion for a while. personally, I've got an on prem server here that
I can touch. It's right next to me. It's in my office. New definition to on
prem.

Doug Camin: Uh huh.

Leonard Boord: And, it's running a bunch of stuff. It's got five or six VMs on
it. It's segmented. It's got a database or two on it. And that's also segmented
across many schemas with many tables. And the idea is I wanted to keep egress
down. So first let me build what we need. Figure out calculate what egress would
look like. Does it make sense going to the cloud or should we get storage in a
data center. and there's pros and cons to on prem and you learn along the way.
Do you want to use proxmox. Do you want to use Hyper-V? Do you want to use,
Windows Server?

Doug Camin: What's the VMware anymore? Screws put down, those ones for any small
and midsize business. They really, brought on really came for those customers. I
know that was something I moved out of. And I hear that story pretty frequently
about folks that had, VMware installation. So then that's the Broadcom
transition happened. It just came down on a part because the price increases.

Leonard Boord: Right. Well and then there's the other angle of like tech debt
right. So when you're servicing these. and I live this through carrot express
and in a very rapid succession, we had a ton of reporting and it was all done
based off of Excel CSVs. and it required humans to go into, into areas and pull
stuff and combine it, aggregate it, you're running into Excel files getting
corrupted because you're using them as a mock database because you don't know
SQL or you don't know SQL exists. you can't invest the time because you're
already working ten hours, twelve hours a day. So when do you find time to
sleep, eat with your family and work? It gets very challenging. So at that stage
in, I call it like the pre AI era, that's where you would need a team to come in
and they would do the full stack and then they want to make money, right? Mhm.
So and I lived it over at RMG where, we did bring in a team. I was doing my
stuff on the side to get data for credit cards, credit card failures, high
declines. Why are we declining? Is it demographics? how can we make the guest
experience better speed of service? And at that time, I didn't have the ability
to build out API pipelines like Uber Eats, DoorDash, and Grubhub because some of
them were invite only. So this other team would have that role. So by surfacing
these reports, I was able to extract a ton of data and we were able to make
better, more informed decisions.

Doug Camin: Mhm. So, one of the things you mentioned about data and you really
pulled that thread through the things you brought up is that, so I have a friend
who he works at Cornell University and, he does application programming. He's
also big into the AI space with them and, has really changed some of the
dynamics, on some of the teams he's working with by using some of these tools to
develop code and other stuff like that. But the one piece that's really critical
is that data at the bottom end of it, so like data cleanup, data management,
data organization, like the stuff doesn't work if you have bad data. And that's
really a critical baseline function of data management data cleanup. The other
was you mentioned about coding up stuff like at my place right now, I have a
team that's, working on coding some things, you mentioned about whipping up a
ticketing system and stuff like that, but I'm gonna go out on a limb and say
you're sophisticated enough that you're not fully coding it, but you are, doing
prompt coding and that type of stuff. And, the changes that are coming in the
coding space are pretty significant. I'm not a coder myself. I mean, I learned
code as part of my degree twenty five years ago, but I have never been in the
space of being a programmer. And the feeling right now that I've seen is that,
and this is what I said to my friend at Cornell about. And I have folks that are
trying to code applications while they can develop the applications. Like
there's, generally like three pillars for my understanding of like how software
development has traditionally worked. You've got the people who write the code,
you've got the people who, kind of like review or vet or understand like the
linkages and those types of things. And you've got a support cycle that goes on
the other side of it. And then those things kind of like, flow back and forth.
how do you have the infrastructure to get what people have coded into a
releasable format. And then once it's out there, how do you maintain it and
update it accordingly and stuff like that. And some of these like five coding
tools have made the first step incredibly easy. So like all of a sudden we can
generate enormous levels of code very, very quickly. But haven't quite yet
figured out the path or there isn't a tool yet that can really, truly answer the
path of how do you sort through all this new code that's been developed
effectively. And that's been a challenge. I know that's surfaced in a lot of
other companies. So I think it's a story about Amazon having this problem where
their coders became so productive in terms of just the volume of code that
they're putting out. But then it became filled with errors because there wasn't
people to review it based on the volume that they were now generating instead.
so the system broke down in other areas where one of the other like kind of
three legs of the stool wasn't able to support the first leg becoming so much
more robust.

Leonard Boord: Yeah. So I have always wanted and I always considered myself to be
like an aspiring coder. So back in twenty three, I took like thirteen classes
and in a very short amount of time. And I got certified in data science with
IBM. Okay. I've taken, a plethora of Udemy courses with Python. just even for
like unreal game engine development, C++. And that's probably one of the more
complicated ones. And while, I can read code, I can highly understand code, it's
very sophisticated and I can follow the, the methods, functions, the
definitions, all of that stuff.

Doug Camin: Yeah.

Leonard Boord: I wouldn't, I mean, I can write code, but I will never be as good
as cloud code. Never. You should lock me in a room for ten years. And I would
only code on my mind and I wouldn't be as sophisticated as. But to go on this. I
built a ticketing system. I built a stream deck competitor. I'm building a
discord competitor. I'm trying to build a hotchpotch of like a GitHub needs for
perforce. So I understand where you're going with this. And the single biggest
takeaway that I have determined from all of this is that the shift is not. That
we're suddenly these amazing coders. No, we have to be amazing debuggers.

Doug Camin: Mhm.

Leonard Boord: And normally, as part of when I code in AI and I vibe code or
prompt engineer and I try to do these things, is that it is imperative that the
AI understands the actual ask so that it doesn't go rogue and it needs to build
what I call a cdg, a change dependency graph.

Doug Camin: Mhm.

Leonard Boord: And then it has to use Subagents to keep context down. And what
I've been able to do through that is I've been able to keep contacts without
compacting so that we know what we're doing. And then Subagents have produced
something like six to eight millions of tokens consumed. So it's a lot of
delegation. I haven't orchestrated this orchestrators going through these
subagents and we're staying on topic. But for example, we produce an output and
somebody has to test it. And who better than the person who's designing it? So I
think the old school method of I'm a builder, you're the tester is going to kind
of die. So I think it's going to be I'm the builder. I'm the tester.

Doug Camin: Mhm.

Leonard Boord: And then you move on. So these three core pillars are probably
going to come down to two. And then the support system is probably also going to
eventually get combined in through AI and agents, because that's just the
fastest way to surface results. But you're always going to have the human in the
loop. But I don't think in the next ten years, we're going to be able to get rid
of the human in the loop because you need somebody who has oversight, somebody
who can understand, the end user, the AI. They don't want to just go back and
forth. And, at some point it happens to me with IVR systems all the time. I get
an IVR system and it's like, man, I just want to ask a thirty second question
and I got to go through seven minutes of IVR.

Doug Camin: Mhm.

Leonard Boord: So, I think that's the big paradigm shift. So for example, the
discord competitor, as you mentioned, is seventy thousand lines of rust. in a
year probably couldn't produce seventy thousand lines of rust comprehensively.

Doug Camin: Mhm.

Leonard Boord: That would need more time. but as every time I build something in
there, be it like using live kit to have communication, adding a form embedding.
What ends up. You always suffer the risk of regression. So you have to do
regression tests. And as part of that hey listen, we just built out this
obviously in different terms. But hey, listen, we just built out this thing. now
I need you to produce a QA doc. What did you change? How did you change it? Why
did you change it? Now let me go through it. Let me see if it passes. Let me see
if it fails. If it fails, why did it fail? Here's the code. Here's the console
logs. Here are the outputs. Here are some images. This was the expectation. This
is the UI. This is the UX. And the amazing thing is that AI can do all the back
end tests. You can do all the smoke tests. But what it can't do today is all the
UI, UX. Okay, I tried, I really tried, I spent a good two weeks trying to build
a team that would go through it. And what I ended up getting was hundreds upon
hundreds of the same picture. It was like, hey, I tested this mighty little
thing and here's a picture of something totally random, like, here's the home
screen. Send a picture of this. Send a picture of that. Let me see it so I can
quickly audit, quickly verify and do this so that I don't have to go in and
reproduce it. And then what I realized is there's no better tester than the
person who's building it because, the intended workflow, if you hand off to
somebody else, now there's an inference layer. What did they build? Why did they
build it? What are we trying to solve? And the person who's building it can own
that space.

Doug Camin: Yeah. So I want to see there's another topic I want to go on. I just
want to make a comment real quick though. the part where you're talking about
like, the prompt engineering and stuff like that, the first thing that comes to
mind is like Star Trek, like next generation, like Geordi La Forge, like talking
to the computer, like computer. What if we did this? And it's like, oh, well,
this, and it kind of like you negotiate back and forth until you get to the
answer and it's like, oh I didn't think about that. Like, maybe we're heading
towards something like that at some point in the future, which would be pretty
cool. so I want to move into, more fun topics here. Talk a little bit about your
history. What have you done? in the past. So can you tell something somebody
wouldn't know about you that listeners wouldn't necessarily know about you or in
your history that you've done?

Leonard Boord: Okay. Well, I want to say twenty twenty to about twenty twenty
two. I was a streamer.

Doug Camin: Okay. Yeah.

Leonard Boord: I was on mixer. It was something like two hundred and thirty two
concurrent viewership. and I played a bunch of different games and that was a
fun time. I did it for about two years. it was great as a hobby and great as an
outlet, but due to the amount of hours that I was investing into it in terms of
cutting content, producing content, learning how to use the the Adobe Creative
Suite and all of that. I would have to peg it as the worst job I've ever had. It
was fun, but it was the worst job.

Doug Camin: Wow. All right. Okay. So going back in time for you, mentioned about
your Twitch or Twitch streaming and other streaming things, but how did you get
into technology? Was this always a thing for you or did you kind of stumble into
it? Tell us a little more about what was your what was your first experience
with computers and how did you get into the space?

Leonard Boord: So I have a very vague memory. I don't remember the game, but one
day my father was selling computers. I don't remember the name of the. It could
have been like a digital or something like that. And he came home with a
computer, connected it with one of the old CRT monitors and like, what do we do
with this thing? And he thought it would be very cool. And there was like this
little tank game, where you were like a green square and you had like a triangle
and a red line, and you would just shoot these, like, little orange blobs. and
it was just this little tank game. So from there, I was kind of hooked. it's not
like the old story of where I started on pong. I would have to talk to him and
see if that's, what it could have been. but from there, we always had like a, I
think my very first console was the Sega Genesis. So. Well, the thing is, I was
too young.

Doug Camin: Too young.

Leonard Boord: Yeah, yeah. So yeah, but we eventually got into that. There was
also like some Voltron games, and there was also some games that, kind of spread
out from there. And I don't remember the exact progression, but if I had to peg
it, it was something like Sega Genesis and N64 into the Dreamcast, I think into
a Gamecube, into an X-Box three hundred sixty. And then for me, the consoles
kind of died. I would always try to buy one in case I wanted to play a game, but
for me, I've always been a computer gamer.

Doug Camin: Mhm.

Leonard Boord: So even when I was like twelve years old and BellSouth was still
around and I'm trying to play a game instead of the days of no saving
progression during a download. I remember trying to upload or play lineage two
forever ago in two thousand and five, and I'm six hours into like a nine hour
update because of internet speed, and then internet goes out and I spend four
days going through this. So I would have to call BellSouth. BellSouth. Hey guys,
I don't know what's happening. I keep updating this at night and it's just like
a twelve year old me having a conversation with them. Well, we need to dispatch,
attack, send attack out. I've already power cycled it. I've learned your script.
I've done this twelve times in the last six months. I know it's not us. Send it.
And surely enough, there was some. Something had degraded in the line or
whatever. And I was too young to understand what it was.

Doug Camin: Yeah. All right, so I'm going to go into a little bit of the
lightning round as we're coming up towards the end of our podcast here. All
right, so what part of your role fires you up the most?

Leonard Boord: Problem solving?

Doug Camin: MM. All right.

Leonard Boord: so, I think idle hands are the devil's play thing when, people get
bored, they, tend to look for challenges and solutions elsewhere. I don't, like
showing up to work and having nothing to do. so having a laundry list of things
to always be solving or working on is definitely great. I think that AI works
against me in that situation.

Doug Camin: Mhm. All right. So make a prediction eighteen months from now, what
will everyone in it be talking about that people are ignoring today?

Leonard Boord: I wouldn't say necessarily that people are ignoring this. I think
people are being slow adopters and late adopters and they're fighting it. It's
AI. If you're not adopting AI today, in eighteen months, this might be a kind of
a hot take, but you may be forgotten.

Doug Camin: Mhm. Yeah, there's definitely, I feel like, there's some agreement
there that, from me, you've got to understand the technology while, our job is
to understand not just the application of it, but the limitations as well. so
how do you deploy it effectively to do the things you need? Like, in my
organization, I did a whole series of, collaborative learning tools. So we, not
just a pilot, but we actually pull people together from across the organization.
We gave them tools, we pulled them together on a regular basis to try and build
an understanding. Of course, this is like eight months ago. So now everything
we've learned is, kind of like already gone. but It helped us shape and
understand where the investments could be made the best, so like, because, I
would get talk from my boss initially to be like, well, why don't we just give
these AI agents to everyone? And I'm like, well, yeah, except that they're,
like, for instance, like copilot for Microsoft three hundred sixty five is an
additional thirty bucks a month per head. And that's not practical. Like the
budget will not only will the budget not bear that, but we know full well that
these people won't use it. So we really work to identify like, okay, this like
subset of people really need it. If they request it and they want to use it,
then we'll give it to them and that type of stuff to both manage cost, but also
build the flywheel to start like spinning up usage, enhancements and the
productivity that comes with it.

Leonard Boord: Yeah. So, I kind of said it very open ended. the reality is, yes,
you need to know its limitations. Like don't tell it to delete system32 that
would be very bad. but I definitely think that just about anybody can use it as
a productivity enhancement tool.

Doug Camin: Mhm. Yeah. I agree, like as far as baseline enhancement, the tools
are already there, even like super basic stuff. Meeting notes, meeting minutes
Summarizations. How do we do this? pull the basic information out of this
meeting. here's a transcript of the meeting where we sat for an hour chit
chatting about things. Tell me what the important salient points are. All of
that is like table stakes. Now these things are baked in the cake, and everyone
who has to do that type of work should already be using these tools to do that
stuff right now.

Leonard Boord: Well, even plugging it into Excel.

Doug Camin: Mhm.

Leonard Boord: Hey, somebody sent me the sheet. I don't like the way it is. I
want to see it this way or that way or help me understand this data. What is it
telling me that I'm not seeing? And, even at a very basic and fundamental level,
if you have a folder with a bunch of CSVs, you can have a conversation. I mean,
the other day, I didn't want to throw something into a database. I really did,
because standing up a database requires building out the table, setting up the
schema, figuring out where it's going to. There's maintenance involved. So I
just did a full data dump of like, thirteen years worth of exports. Just threw
it in there. And I said, hey, Claude, I want to know how many people, build me a
C suite level KPI dashboard, which is, explanatory answers a lot of questions
with regards to how many downloads, how many users, how many active, what iOS
and in four minutes it had it. You know, if I were to hand that off to an
analyst, it would take five days.

Doug Camin: Yeah, for sure, for sure. All right. So we're right here at the end
of the podcast before we go, we always love to hear from the leaders that we've
got on the show sharing, two things. One is, what's your advice to fellow CIOs
or IT directors out there? And the other that I always make sure to ask is,
what's your advice to the people who are coming up trying to get into the space?
what do you have advice for each of those spaces?

Leonard Boord: Yeah. So for the first question, I would say talk to your peers.
And if you don't know any peers, then, have lengthy conversations with the tool
of your choice. If it's GPT, have the conversation with GPT with regards to your
decisions that you're trying to make. Try to strong arm it into not being a yes
man. Tell it to be Gan or use adversarial questions. Any idea that you have.
Instead of saying, hey, I have this great idea. No, no, no, I have this idea. I
want you to poke holes in it. I want you to to find gaps that I may have not
thought of. This is what I have thought of. What do you think? Attack it. Tear
it down. If you can. And then that'll show or expose areas that you haven't
thought of. And then you can work on those. And then for the people that are up
and coming in the space, I found that one. If you can get access to resumes of
what people have done in the position that you want to go, it is highly
informative, because you'll see credentials, you'll see kind of like their
career path. And nowadays LinkedIn is kind of like having your resume exposed.
So try to figure out who is in a position that you would like at a company that
is similar to what you would like to be doing, and then you can kind of look at
that as a high level summary. You can copy and paste it. Once again, here we
are, dump it into AI and say, build me a career path through courses on Coursera
or Udemy that will give me the exposure, that I would need to be able to have a
conversation around this and to show knowledge so that I could eventually one
day obtain a position there. And then internally in your own company, ask
questions on what would need to be done in order for you to obtain those titles,
because nothing makes, finding these positions easier than having that title
already.

Doug Camin: Mhm. Awesome. That's some sage advice. All right. Leonard, thank you
so much for investing your time with us on the podcast today.

Leonard Boord: Thanks for having me.

Doug Camin: That's a wrap on today's episode of You've Been Heard, the platform
for IT leaders. I'm Doug Camin, and we look forward to coming to you on our next
episode.