Episode 431: Vibe Coding and the Future of Software Development

In this episode of Dynamics Corner, Kris and Brad are joined by Matt Strippelhoff, CEO of Red Hawk Technologies, a company he founded in 2008 that specializes in developing, supporting, and maintaining custom software applications for mid-market clients. They dive into the evolving landscape of software development, exploring the concept of "vibe coding" and the impact of AI agents on the industry. They discuss the buy vs. build dilemma, emphasizing the importance of viewing software as an asset rather than a project. The conversation also explores the new trend of Software Development as a Service, the challenges of managing cognitive load in a rapidly evolving tech landscape, and the potential future applications of AI in transforming business operations.

Send us a text

Support the show

#MSDyn365BC #BusinessCentral #BC #DynamicsCorner

Follow Kris and Brad for more content:
https://matalino.io/bio
https://bprendergast.bio.link/

Episode Transcript

Speaker 1 (00:01):
Welcome everyone to another episode of Dynamics
Corner.
Brad, buy or build.
Build a table or buy a table.
I don't know.
I'm your co-host, chris.

Speaker 2 (00:14):
And this is Brad.
This episode was recorded onAugust 27th 2025.
Chris, chris, chris.
Buy versus build that's aquestion that I hear often, and
today we had the opportunity totalk about buying versus
building as well, as a topicthat everyone seems to be
talking about these days iswhere is the position of AI in
software development With us?

(00:36):
Today, we had the opportunityto speak with Matt
Tripp-Strippaloff.
Good morning, sir.
How are you doing?
Doing well.
Nice to see you, brad.

(00:56):
Nice to see you as well.
Nice to hear you too, by theway.

Speaker 1 (01:00):
Likewise, yeah, finally we'll hear each other
now.

Speaker 2 (01:04):
Yes, yes, yes, sometimes we have those
struggles and challenges, but,uh, we always make it through,
which is good.
Uh, you know, I have beentalking about it on these
episodes as we start about theyou know my new fun thing that
I'm doing.
I'm doing a lot of vibe coding,okay, right, you know, you know
it's good, bad or indifferent,but I'm able to sit down and I'm

(01:28):
doing things that I never eventhought I would be able to do.
How well I'm doing them, I donot even know, but it's piqued
my curiosity to where I'mcreating some cool things, and I
had mentioned in previousepisodes I purchased a bunch of
Raspberry Pis, oh yeah.
And now I'm just writing abunch of Raspberry Pis, oh yeah,
and now I'm just writing abunch of code and creating these
little applications when you'revibing when you're vibing code,

(01:49):
vibe coding, are you using this?

Speaker 1 (01:52):
I mean, are you doing it with a particular
programming language?
I know you had mentioned Pythonlast time.
Is there any other language inyour vibe coding that works
really well.

Speaker 2 (02:09):
Right now for the vibe coding aspect.
I'm using python.
I do have it create, not fromthe vibe coding point of view
where I'm creating anapplication, but I've created
some small things.
Uh, you know, professionally Iuse it with al right.
I don't call that vibe coding.
I use github copilot agent toassist with coding, but I don't
call that vibe coding becauseI'm still reviewing and
processing.
I knew nothing about pythonformally when I started this.

(02:29):
I'm picking up and learning thesyntax and structure.
Don't know other libraries andstuff, uh, but I I'm using it to
create powershell scripts to dosome things that I need to do.
And it was really actuallypretty cool because even on the
python I I was able to ingestAPIs.
I ended up creating a Dockercontainer with MySQL, a Docker
container with Apache and thencreated a vibe coding Python

(02:53):
tool to process and read an API,process the data and insert it
into MySQL.

Speaker 3 (03:00):
Fascinating.
We were talking about this atmy office yesterday.
It's like one of our seniordevs.
He's been with us for about 16years and he's probably pretty
far along out of most of ourdevs.
He's been using Cloud Code alot and the way he described it
he said the new programminglanguage is English which is

(03:23):
English.

Speaker 2 (03:23):
Thank you for saying that, because if you go back to
a few episodes, I talked aboutthat specific thing and and I do
have a question I want tofollow up on the vibe coding and
that's what one of the hosts onthe podcast had mentioned.
This was several months ago.
He said you know, within 10years, what do you think will be
the most?
Or even five years, I forgetright at this point.

(03:43):
He said what do you think willbe the most common programming
language?
And he answered the humanlanguage and it's almost
becoming to that point where youcan sit down and think of
something and ask for it andagain, there's different models,

(04:04):
there's different tools but itwill write the specifications
for you, it will write theapplication for you, it will
write the tests for you and youcould just basically give it
something and go to bed.

Speaker 3 (04:19):
I don't think we're far away at all from disposable
applications, purpose-built,short lifespan things that we
would never have done before ina traditional SDLC, because it's
too much of an investment,takes too long to build these
things, and now businesses aregoing to be spinning up
applications that are purposebuilt and maybe they only need
them for a few months, maybethey only need them for a

(04:40):
specific purpose, and thatpurpose is time-based or
time-sensitive and it doesn'thave to hang around.
It's crazy.

Speaker 2 (04:48):
Yes, oh, it is.
And it leads to the topic ofwhat I wanted to speak with you
about, what we wanted to speakabout when we were talking
earlier.
But first, if I ask you, whatis your definition of vibe
coding?

Speaker 3 (05:02):
Well, you know, I think of vibe coding in terms of
like true vibe coding, I thinkis a terrible idea, and what I
mean by that is my definition ofvibe coding is you're vibing
with the AI agent that's doingall of the coding.
You know the tools that we'reusing.
We can see the code that it'sproducing.

(05:24):
We can course correct, we canrely on the agent to make
modifications and updates.
But I think true vibe coding,using some of these tools that
are no code, you don't even seethe code.
It's just creating things foryou.
There's a lot of hype out there.
I guess the way I would definetrue vibe coding is you're just

(05:45):
vibing, using natural languageto create things.
I think it creates a lot oftechnical debt.
I think that if you're notcareful and you don't understand
what these tools are doing, youcan choose tools that are
putting things into your codethat maybe you wouldn't want
there if you knew better.
Got to watch out forcybersecurity risks.

(06:07):
Slop squatting you know as aterm.
If you guys talked about slopsquatting on any of your you
have not no, so the the the slopsquatting is, is starting to
show up more and more.
Basically, what's happening isbad actors are polluting the
models with misinformation aboutframeworks or plugins, rather,

(06:28):
that aren't real, but they getso much information out there
that the vibe coding tools willwill pick these things up, which
has they're basically injectingbad code through slop squatting

(06:49):
.

Speaker 2 (06:51):
So a bad actor is slop squatting, yeah, slop
squatting, yeah.
So a bad actor is injectingmalicious code by training the
models to create the maliciouscode, with the assumption that
the individual using these toolsdoesn't understand what they're
doing.
I could see how that's verydangerous.

(07:11):
And that brings us to thiswhole vibe coding talk.
Again, it's vibe coding doingsomething at home.
I'm having fun experimenting inmy own isolated environment,
which I don't even do, to behonest with you, it's not even
on my home network network.
I have a separate network thatI use for vibe coding for that
reason that you just mentioned,because I don't want anything to
for lack of better termspollute or do any, uh, harmful

(07:35):
actions on my existing network.
Uh, so, but again, at home it'sit thing, but at work it's
another thing, which comes tothe whole question that we often
speak about.
Chris and I work, we assistwith individuals enhancing or
extending a business applicationwhich that business application

(07:58):
has pre-existing well, notpre-existing has tools available
that you could purchase, or youcan have customizations
developed for you which we cantalk about.
This whole AI agent thing whichbrings up the topic of is it
better to buy or buildapplications?
And with that, before we getinto that if you would tell us a

(08:19):
little bit about yourself.

Speaker 3 (08:20):
Sure, yeah, yeah.
Well, I've been in this line ofwork for about gosh over 25
years.
I started Redhawk Technologies,a software development company.
We create custom businessapplications, middleware
solutions that integrate withcommercial software applications
as well and mobile apps, youname it, and that's really our

(08:41):
focus.
We've been in business for alittle over 17 years.
Just made the Inc 5000,000 forthe second year in a row, so out
of the 5,000 fastest.
Congrats.
Thanks, man.
Yeah, we ranked 1,062 on thatlist this year, congratulations.

Speaker 2 (08:56):
That's awesome, 3,000 .

Speaker 3 (08:58):
Yeah we're really happy about that.
But a big part of that successcomes from our unique business
model.
We develop, support andmaintain custom business
applications under a managedservice model, so it's like it's
development and service, so fora fixed monthly fee schedule.
We're basically a fractionalDevOps team.

Speaker 1 (09:16):
We create support and maintain these things.
That's interesting, it is avery cool model.

Speaker 2 (09:21):
Yeah, Now I don't know what I want.
I don't know which we shouldtalk about.
Should we talk about vibecoding, AI agents, building
versus buying, but the managedservice model that you have.
Maybe we'll blend them alltogether.

Speaker 3 (09:32):
We can blend them together.
You know it's, it's yeah.
Wherever you want to take theconversation, I will tell you
that when we discuss new buildswith customers, they have a
vision for something that theywant to create for their
business.
If it's clear to us that theyhaven't looked for commercial
off-the-shelf solutions firstand we know that there are
solutions that might work forthem, we would advise them to

(09:55):
buy, not build, Even thoughthat's not our business.
We're not out here as avalue-added reseller selling
Microsoft products or Salesforceor tools like that.
Sometimes we think it makes alot more sense just to buy
what's available off the shelfand customize suits your needs

(10:15):
before you build somethingcustom.

Speaker 2 (10:17):
For sure.
So what considerations shouldsomeone take before they make
that decision of buying versusbuilding?
And I see oftentimes go ahead?
My mind will run on this one.

Speaker 3 (10:31):
Yeah, I think the first step is really determining
the level of customization thatmight be required for something
commercial to suit yourspecific business needs.
The more customization you need, the more likely it is that you
would build something custom.
If you think that you can adaptyour business to work with the

(10:54):
off-the-shelf solution maybeit's a blend of those things you
might modify some of the thingsthat you do so you limit the
amount of customization requiredin an ERP or CRM platform.
Then you would lean towardoff-the-shelf.
That would be the way that Iwould look at it.
So I think you have to have apretty good understanding of the
business needs before you startselecting the tech stacks and

(11:16):
the commercial softwaresolutions that you want to
evaluate.

Speaker 2 (11:20):
And with that, what consideration now?
We talked about the vibe codingand we speak about how easy and
I use that word loosely in thiscontext it is to create
solutions and I'm seeing thatmore and more as well, where
everyone thinks oh, my famousquestion, how hard can it be?

Speaker 3 (11:41):
How hard can it be?
How hard can it be?
How hard can it be?
When I talk to customers andprospects and we're evaluating
their requirements, we'retalking things through.
One of the things I try todetermine is how many exceptions

(12:05):
to their rules will they have?
Businesses always have a yeah,but I needed to do this, except
for in this scenario.
And so if they have lots ofexceptions, then there's going
to be a lot of complexity, and Icall them exceptional companies
because they have all theseexceptions to their own business
rules.
That'll drive up a level ofcomplexity.
Whether you're vibe coding orusing AI as a code assistant to

(12:26):
create something, or you'redoing something, customizations
with the commercialoff-the-shelf ERP, the level of
exceptions that they have willdrive up the complexity, the
effort, the time, ongoingmaintenance, you name it.
It gets more and more expensive.
I don't care what approach youtake.
You name it, it gets more andmore expensive.
I don't care what approach youtake.
If you have an exception toevery business rule.

(12:47):
If you're an exceptionalcompany, you're going to pay an
exceptional fee for it.

Speaker 1 (12:52):
There's always an exception.

Speaker 2 (13:02):
That is a good way to phrase it, because a lot of
businesses again, everyone wantsto have a differentiator to
make them better than theircompetitors and what it takes,
what they think thatdifferentiator is.
But oftentimes thatdifferentiator isn't as
significant as they may think.
So, like you had mentioned,it's the consideration of buying
versus building and youmentioned it's time.

(13:23):
If you're building something,it takes time.
You also have the risk, again,assuming that the commercial
product that you're looking tobuy is properly developed,
vetted and it's a functionalapplication more error-prone, I
would say because you also haveto go through a good testing

(13:43):
cycle and make sure that theapplication works to your needs.
You have maintenance on that and, as you had mentioned, with all
of that is a higher cost.
Now that higher cost may bemitigated based upon the
complexity of what you havebecause it generates revenue for
you.
But there's that technical debtthat oftentimes they don't.

(14:05):
Someone doesn't factor into theequation of now I need to move
to a newer version of thatapplication or technology's
changed.
You know, as you talked about,you can extend or customize
commercial built software aswell.
Where does that technical debtfall into play?

Speaker 1 (14:22):
so this is where I can go down the road on this, I
think, I think lot of that whenwe deal with those clients at
least for me from my experienceis usually it's short term, like
we're trying to solve a problemnow.
So when they're looking for asolution, they're really trying
to solve okay, we have thischallenge that we're dealing

(14:43):
with, but they don't really lookat it from a long-term
perspective, like you had said,brad, like the technical debt.
What does it take now tomaintain that and also build
upon that?
So a lot of times they justkind of like let's just solve a
problem now.
And many times theconversations I've had is yeah,
it was right at that time, itwas the right tool at that time,

(15:05):
but then it's been two yearsnow.
It's no longer a fit for us.
And I was like two years,that's a very short period.
You're not getting yourinvestment back.
I guess I would say I don'tknow how common that is for you,
matt.

Speaker 3 (15:20):
It's the whole reason we developed our business model
.
Honestly, those earlyconversations with clients we're
talking to them.
The first thing is we want tohelp them realize that they need
to change their perspective onthe solution that we're
developing.
That it's not a project.
Oftentimes they think of it Ihave a project.

(15:42):
I need to find the right vendorto execute that project.
But if you have a projectmindset, there's a beginning,
middle and an end, and thereality is with software,
whatever you put out there isnow in an ecosystem that it
doesn't control.
There's all these factors.
Think about it like if we thinkabout it in organic terms.

(16:03):
If we think about it in organicterms, it's in an environment.
It's designed to thrive in thatenvironment, but the
environment is changingOperating systems, hardware, et
cetera.
To think about it in terms of anasset versus a project that's
one of the things we try to helpour customers recognize is that

(16:23):
you now have another asset thatis part of your business.
You only made this investmentbecause it's an asset, and an
asset is creating value, whetherit's streamlining workflows and
reducing the amount ofworkforce that you need, so you
can scale and you don't have toscale through people.
That's a great reason to buildsomething custom, but that's an
asset.
If you have a fleet of trucksand you're in the service

(16:45):
industry, you're changing theoil.
You know you're maintainingthose assets.
You think of those things asassets.
So, as far as remediatingtechnical debt, updating
packages and frameworks, that'sall part of what we're doing
under our software developmentas a managed service, doing
under our software developmentas a managed service.
So we're responsible forhelping that asset continue to

(17:08):
thrive as the ecosystem aroundthat asset is changing, so it
could be other things changing.
It could be that they'rechanging to a different ERP and
the integration needs to bemodified.
The cool thing about what'shappening with AI coding
assistance now are all the MCPservers that are out there for
these commercial platforms makeit a whole lot easier to create

(17:28):
a technical plan.
In an AI-assisted tool likeCloud Code is a good example you
connect in the MCP server forAzure, for example, and then you
can basically create your plan,review the plan and then have
it execute the plan to build outor modify an integration.

(17:50):
A big part of our conversationis because we primarily focus on
serving mid-market, andmid-market customers don't have
DevOps teams in-house, sosometimes there's some education
we have to talk about what asoftware bill of materials is.
They're like well, what's that?
Well, that's the list of all ofthe frameworks, components,
libraries that we're using.
They don't really care abouthow the sausage is made in the

(18:12):
end, but they kind of need to.
So we have to educate them alittle bit about software bill
of materials routines that wehave as part of our service to
scan that SBOM for commonvulnerabilities and exposures
that are emerging.
Because if the bad actors arethe only ones paying attention
to vulnerabilities showing up inthose libraries and you're not

(18:33):
patching those things as amatter of routine, that asset's
going to turn into a liabilityovernight.

Speaker 2 (18:44):
It's a great way to think of it.
I'd like your analogy just tobring back to what you you
started that topic on it's aproject.
I mean, it's not a project.
You're purchasing an asset andthen you're going to do
maintenance on that asset, justas if you would do for as you
had mentioned a fleet of trucks,whereas a lot of times

(19:05):
individuals exactly like you sayit's a project.
They start the project, theywork through and implement the
project and now they're donewith the project.
Because we always say or Ialways say and I talk with
others in the same field and Idon't want to put words in
Chris's mouth ERPimplementations are never done.
And it's not that they're neverdone because the implementation
is poor.

(19:25):
It's never done because youhave to look at it and frame it
as maintenance, as if it's on atruck, because you do need to
replace the oil.
And again, that could beequated to technology changing
or even your internalrequirements changing as the
economic climate changes as well.
So it's a great way to think ofit and I can honestly say, in

(19:47):
the couple of years that I'vebeen doing this, it's the first
time I heard someone say look atit as an asset versus a project
, and you may absorb thatimplementation differently,
because now I'm purchasing afleet of trucks, I'm purchasing
something to of trucks, I'mpurchasing something to provide
value.
I'm kind of repeating what yousay because it's really sinking

(20:09):
into me.
Yeah, I really like that tooWith me with me.

Speaker 1 (20:13):
I like that very, very much because you're right.
A lot of people, a lot ofbusinesses coming in especially
small, medium-sized businessescoming in and getting an ERP
system at least in our worldhere is that it's a solution
that would solve all theirproblems and that's it right.
But they don't realize that thethat Brad mentioning it doesn't

(20:37):
end.
You continue to improve uponthat and you change your
business process and it becomesan asset to your business so
that you can be more profitable,profitable, profitable and
increase revenue.
But they don't realize that ittakes some work to do that, to
maintain it just like an assetrather than just like, hey, I
want to buy this and it'ssupposed to solve all my
problems, make me more money,yeah.

(20:57):
So they got to change thatmindset, definitely.
Yeah.

Speaker 3 (21:02):
Yeah, if they, if they don't.
And I think it still happens alot today where yes

(21:23):
no-transcript.
That's if you want to addfeatures.

Speaker 1 (21:28):
The other two are really bad man.

Speaker 3 (21:29):
They're really bad.
Right, something happened.
It's a cybersecurity issueoccurred and you can track it
back to that.
Now, liability no longer anasset, because nobody was
maintaining it and it was youknow.
Some vulnerability came up andit was exploited by the bad
actors your operating budget islicenses.
Yeah.

Speaker 2 (21:48):
That's not enough, I don't know what you're talking
about, but yes, you're right.
This is a great viewpoint tohave.
As you had mentioned,oftentimes people don't budget
for the maintenance on theirsystem, whatever that system may
be.
They either purchase or developor, as Chris, you've mentioned
a lot of times, they'll budget.

(22:08):
Okay, well, my annual licensefee is X, so I'm going to pay
that annual license fee, not themaintenance on that asset.

Speaker 1 (22:18):
They should have a budget for maintenance or new
features right, like maybeimplement a new feature which
requires training and requiresimplementation and things like
that.
That would improve yourbusiness and that could also be
enhancements too, that theyshould budget.
But they don't think about thatLike it's just license, that's

(22:39):
it.

Speaker 3 (22:40):
Yeah, it's a good perspective.
Yeah, In our model we'reproviding what we call EMS hours
.
It's just good perspective.
Yeah, In our model we'reproviding what we call EMS hours
.
It's just enhancements,maintenance and support.
So that's part of the blendedservices that we're providing
under that fixed monthly feeschedule.
So it's all part of it.

Speaker 2 (22:58):
So your implementation approach is
software development as aservice.
Would they pay a monthly feefor software development?
Is that through the entireproject or is it once they have
the asset developed?
The ongoing maintenance on thatis a software development as a

(23:19):
service.
I like this model.
I hear of managed services morefor system monitoring, system
patching.
When I say system, I'm notsaying application system, I'm
saying operating system.
To put it to different terms,we have our network operating

(23:40):
systems, we have our networkhardware, we have our network
performance monitoring, like alot of that, network operations
center type management services.
This is the first time I'mhearing of software development
as a service.
I'm interested to hear aboutthis.

Speaker 3 (23:55):
Yeah, so we have options that our clients can
choose from in terms of how thecontract is structured.
If they have the capital andthey want to use that capital
for the initial build, we canset up the contract in a way
that the first 12 months thateven billing schedule, that

(24:15):
fixed monthly fee schedule isaddressing their investment
required for that initialgreenfield development effort,
in combination with some budgetfor maintenance and support,
because it's not going to take12 months to build.
Most things take six to ninemonths depending on level of
complexity.
So, um, we're still going tobuild it in that, in that window

(24:37):
, like it were, as if it weretime and materials project, but
their fee schedule can bedistributed out over that 12
months so they still have thatfixed monthly fee schedule.
If they want to preservecapital, then we will strike a
deal over and generally ourcontracts are three to five
years.
We'll strike a contract withthe customer that allows them to

(24:58):
amortize the cost of thatgreenfield development with the
bundled services over 36 months.

Speaker 2 (25:06):
Interesting, interesting.

Speaker 3 (25:07):
Yeah, it works great because oftentimes what we find
is, especially in the mid market, they have some idea of what
they're willing to invest.
No one ever shares budget withus.
You know what I mean, and a lotof times the way things are
approached is there's a championon the client side and they've

(25:27):
been given permission to go shopand the first conversation I
like to have with them is well,what is this going to do for
your business?
I need to understand that it'sgoing to generate some value.
Otherwise, whatever proposal weput in front of that customer,
that prospect, it's just goingto be pure cost and they're only
going to make the investment ifthat cost falls within whatever

(25:48):
number.
They pulled out of their hatthat we don't know what that
budget is.
Right, that's the only reasonthey're going to make the
investment is because they foundsomebody willing to do it, for
what they're willing to pay forit.
But they might not even knowwhat it's really going to do for
their business.
It just seemed like a cool idea, so we look at things like we
want to.

Speaker 2 (26:07):
Welcome to our everyday.

Speaker 3 (26:09):
Everyday, this is Everyday.

Speaker 2 (26:17):
No, this is.
I don't mean to cut you off,but that is like the mindset is
so true is is.
It's a recommendation tobusinesses explore what the
value you're going to get fromit.
And that budget question,nobody wants to ask because you
can have someone come to yourhouse to do work and they'll say
what are you willing to spend,how much budget do you have?
You're like I'm not telling youhow much is it going to cost,
because everyone wants to toplay that out.
Well, they want to make surethat everyone wants to feel like
they're getting the value forthe dollar, and not somebody

(26:41):
just saying I want to exhaustthe budget, which is interesting
.
So everyone plays that danceand I often will try to say well
, you don't have to give me anumber, let's just come up with
a ballpark.
Because if I'm purchasing a newcar, I'm either going to go buy
a Toyota or I'm going to go buya BMW.
There's a lot of costdifference in there.

(27:03):
But are you in the market for aToyota, like a common standard
vehicle?
Are you looking for a high-endluxury vehicle?
And then to your question why?
Like, let's look at the valuethat it's going to give you,
then spending that capital maynot be so hurtful, because you
can see the return on yourinvestment, as we'll call it,

(27:26):
yeah, uh, in this case that's uhthat's a good approach.

Speaker 1 (27:31):
that's a hard mindset to change, though, in the
industry, right, because I I, Idon't know where who's making
these decisions.
I don't know if it's someone infinance or, like you, just have
this much money but don'tunderstand the actual what are
you trying to solve, what areyou trying to do?
You know what are you trying toinvest on, what kind of asset
you want, and so forth, and soit kind of yeah, you do, we need

(27:56):
, we need some sort of educationin the industry where, like you
have to change that mindsetwhen you're bringing in software
into your business.
It's a pipe dream.

Speaker 3 (28:08):
We created some tools , really simple tools, there,
and anybody can go to ourwebsite at redhawk-techcom and
go to downloads.
But one of them is just areally it's a glorified Excel
workbook, guys.
That's all it is.
But but what it does is it putsthem on rails to think about
their manual process androutines and cost of errors and

(28:32):
omissions, because most of ourwork ends up coming back to
workflow automation, doingthings with machine learning
even AI these days creating sometype of integration between
systems, so you're gettingpeople out of spreadsheets.
It can be BI implementations,things like that.
But for them to understand thevalue that it creates, they need

(28:54):
to think about the cost of notchanging.
What is the cost of them notmoving forward with the solution
that they're going to incurweek over week, month over month
, year over year?
And so it's an ROI calculator.
So basically, all they do isthey fill out information about
who's involved in thoseprocesses, how frequently

(29:17):
they're doing them, how muchtime those processes require.
If a mistake's made, what isthe cost of an error and
omission, if there's an audit,and then it'll calculate what
their cost is to stay withstatus quo, to stay with status
quo.
And all of a sudden, whenyou're looking at a custom

(29:39):
software or even a commercialsoftware implementation, you
have something to compare it to.
Okay, if I make this investment, these costs go away.
How long does it take for me toget my ROI?
And so it helps them make abusiness case?
And you know what, if theycan't make the business case,
they shouldn't make theinvestment.

(29:59):
So I'm speechless.

Speaker 1 (30:03):
That's a hard pill to swallow, I know.

Speaker 2 (30:08):
But it's a good approach to.
You know, this is our buildversus buy conversation.
But now this is even justchoosing to move beyond that
implementation, it's somethingbecause it is, it's, it's a,
it's a.
it's a nice way to approach whatyou're doing and then you're
treating it as purchasing anasset, where, if I'm going to go

(30:28):
buy a refrigerator from my home, I know how much money I want
spend what it's going to do forme.
You could equate that tosoftware.
And why are you willing tospend that money, versus just
somebody saying, oh, we have$10,000 to spend, see what you
can get.

Speaker 3 (30:45):
Yeah, yeah.
I've got a great example foryou guys.
One of our clients is a creditunion, and you remember when the
feds were changing rates everycouple of weeks there for a
while?
Well, all of their financialproducts that they had published
online were hard-coded.
The agency they worked withjust built out the website, all

(31:07):
of you know.
But it all comes back to theinterest rates, and so when
their team's going in and havingto modify dozens and dozens of
digital assets to modify theirrates and the impact of what
they're offering because of therate changes, they were spending

(31:29):
tons of time every month.

Speaker 1 (31:33):
Manually changing rates.
Manually changing Because it'shard-coded, all order codes,
Well, on all their financialproducts.

Speaker 3 (31:39):
Right, yeah, and it's also a highly regulated
industry.
Whatever you're offering,that's what you've got to
deliver, and if you misssomething out there and a
customer comes in and takesadvantage of that, I mean you've
got to honor it.
So they said, well, what wouldit cost for us to have a custom
just a basic custom webcomponent where we can just
indicate what the interest rateis and then all of our financial

(32:01):
products update automatically?
Well, we gave them a proposalfor that and they're like whoa
man, we don't want to make thatinvestment.
I'm like, okay, well, let's usethis little worksheet I've got
and we plugged it all in and Isaid, look, you're going to pay
for this project twice in thenext 90 days If you don't do it.
Yeah, You're like oh well thatmakes a heck of a lot of sense.

(32:24):
Let's go ahead and get thatdone, and so within two weeks we
had a custom solution for them.
It was really simple basic webapplication and solve that
problem for them forever.

Speaker 1 (32:34):
Wow.
So they didn't think about thelong-term right.
They just can say, oh, it's tooexpensive.

Speaker 3 (32:38):
It's too expensive, We'll die from a thousand cuts.
You know what I mean.

Speaker 2 (32:45):
It's almost like it's not a software buying problem.

Speaker 1 (32:50):
It's a behavior mindset problem that you're
trying to solve at the verybeginning, before the software
solution you're bringing,because at that point it doesn't
really matter.
It's a bigger challenge istrying to change the way they
think when they're bringing in asolution or bringing in an
asset in their business.

Speaker 3 (33:08):
Yeah, Now with vibe coding.
I want to connect us back tovibe coding here real quick.

Speaker 2 (33:11):
Oh, yes, you know what I mean.

Speaker 3 (33:12):
Yes, what's happening now is people're gonna people
who aren't thinking asset versusproject, which are a lot of
people right thinking projectnow they're gonna just have
their bas or their subjectmatter experts, vibe code the
solution like it's a project andthey're just gonna put that
into their ecosystem and lookout, man, I mean no, it's, it's

(33:34):
that I'm seeing more and more ofthat uh type of thought process
.

Speaker 2 (33:41):
And just to dovetail back to a dovetail, or just to
jump back to chris, the commentthat you made, it's it's also
not necessarily someone offeringthe service or someone
implementing software to changethe mindset.
It's important for whomever ismaking these decisions, you know
, for those that are customersof or consumers of these

(34:03):
products and services that maylisten to this to change your
mindset so you can reallyunderstand and how to calculate
the value of what you're tryingto do, to see if it's more than
just a good idea, because, asyou had mentioned, if you can
really get the return oninvestment calculated properly,
it should tell you right thereis it even worth doing it?

(34:23):
Versus, oh, it's a nice shinyobject, it's cool to have, let
me pay for it and now have tomaintain it and bring that into
it.
But to go back to your otherpoint, there is it's becoming
more and more of how hard canthat be without them
understanding?
Oh, anyone can use AI to createan email.
Anyone can use CREAI.

(34:44):
Oh, we have someone, like yousaid, they created this really
cool little application at homelast night vibe coding.
Why are you going to charge methis much money to do it so?
yeah it's good luck inmaintaining that.

Speaker 3 (35:02):
Yeah, vibe code I use another analogy a nail gun
won't make you a carpenterexactly, that's right.

Speaker 1 (35:10):
That's you could go buy that stuff off Home Depot.
Right, he's not going to get it.

Speaker 2 (35:17):
But learning how to use the nail gun will, yeah, and
someone with a nail gun thatknows how to use it will be
better than someone who has ahammer doing the same thing.
So it's the whole thing that wekeep saying AI is not going to
replace you.
Somebody using AI will, andthat's a big thing to think

(35:41):
about.
It's not the using AI, becauseAI is doing it for them.
It's AI is helping them withsome of the mundane tasks or
some of the minutiae, as you hadmentioned.
You can set up these agents.
These agents will go do andhave some functions.
It's almost the equivalent ofhaving a team of junior
developers.
Tasks or some of the minutiae,as you had mentioned.
You can set up these agents.
These agents will go do andhave some functions.
It's almost the equivalent ofhaving a team of junior
developers.
You still bring the code back,you review the code, you make

(36:03):
sure the code's sound.
Then you implement it, not just.

Speaker 3 (36:08):
I could go on for days about this we're trying to
and we're having quite a bit ofsuccess with this with our team
here internally at RedHawk iswe're helping the engineers
realize that they now will beorchestrating, leveraging their
many, many years of softwareengineering experience to
orchestrate the AI agents.
The AI coding agents are typingway faster and they have access

(36:34):
to knowledge in seconds, so I'ma big fan of using AI coding
agents, if done properly andsecurely.
Secret standards.
Yes, yes, and right now we'regoing through a very specific
project, internally facing tomodernize our sdlc end-to-end,

(36:58):
to identify the, the toolingthat we're going to use, writing
the acceptance, use policiesand determining the impact.
And, guys, I'm telling you, man, it's a, it's a new world, it's
a, it's a, it's a new world.
I think we're having the samechallenge right, brad.

Speaker 1 (37:14):
That's something that you have to change your
standards now because you'rebringing in AI agents when
you're developing, so you haveto fit that into your now new
standards of SDLC.

Speaker 2 (37:28):
We can shoot off of this hard, but it's like you you
said coming up with thestandards and individuals
orchestrating and then using thetooling, and then the one of
the big things with this isaccepting that you're
orchestrating.
It's almost everyone is a morecritical of ai than they are of

(37:53):
people.
And, to go back to what youwere saying is, ai has access to
more information, it can go wayfaster and it can make mistakes
or it can hallucinate, as, aspeople use the term, I can do
the same thing, but we're goingto say that Brad made a mistake.
Okay, well, let's fix it.

Speaker 3 (38:13):
But oh we have this AI agent.

Speaker 2 (38:15):
Yeah, we have this, this, we have this ai agent that
made a mistake.
Right, it still has a,potentially, and not in all
cases, and I'm not trying to sayit's better than a human or
anything like that, but it's.
Ai can do something 99 betterthan a human, for example, and
that one percent that it missedand the human could do the same
thing and miss 20, and we won'tsay anything about the human.

(38:36):
But but we'll say that the AIis flawed.

Speaker 1 (38:38):
Yeah, you can never trust it again, right?

Speaker 2 (38:41):
It's comical to me in a sense, and I'm not saying
it's better.
I'm just saying it's learn howto use the tool and understand
the limitations of the tool,just like we talked about the
limitations of a hammer, amanual hammer, versus a hammer
with a compressor right, an airhammer, whatever they call those
things.
I'm not a carpenter, I don'teven know what they call it Air

(39:03):
gun right.

Speaker 3 (39:04):
Is that what they call?

Speaker 2 (39:05):
it Nail gun, Okay.
So it's the same type of thing.
You have to know thelimitations and how to tweak it,
Like how much pressure do youneed you limitations and how to
tweak it.
Like how much pressure do youneed, you know, do you want to?
You know, do you want?
Where do you want the nail head?
Do you want it to be recessed?
Do you want it to be on thewood?
You still have to makeadjustments to get that nail
exactly where you want it beforeyou execute.
So with that, how do youeducate a team to accept it and

(39:35):
then work in unison with thisorchestration?

Speaker 3 (39:38):
Well, the first step for us was to identify the
champion in our business seniorengineer with more than 15 years
of experience, who just wantedto dive in and really understand
how to use these tools.
Basically, think of them as anAI czar within the business.
So figuring out what the newstandards are, which tools we

(40:04):
should be using, the acceptanceuse policies, what they should
look like, how do we protect ourclient's IP, how do we collect
our IP, you know, protect all ofthose things.
And the next phase for us isand this is still early, I think
a lot of companies are stillearly is answering that question
, brad, I don't know that I havea good answer for you yet

(40:24):
because we haven't completed thejourney, but we need this
leader in our organization totrain, leader in our
organization to train, and it'scoming up within the next week,
two weeks.
He's going to be trainingmultiple engineers, onboarding

(40:44):
them on using cloud codespecifically for some of the
initiatives that we've got inthe business right now.

Speaker 2 (40:51):
Will the journey ever be complete?
No, because in my opinion, thisis going lightning fast,
because that MCP server thateveryone's talking about today
go back two months.
I don't think it exists.
I don't know the timeframe, butwhat I'm saying, what it feels
like, is two months ago itdidn't even exist.

Speaker 3 (41:08):
Yeah, it's only been a few months, yeah, and it's a
huge game changer, because thebiggest problem with vibe coding
early on was that it would losecontext.
Because the way tokenizationworks, you know you start having
a conversation and it has ashort-term memory issue.
It doesn't understand thecontext of the larger
application, so it can startgetting really wonky really fast

(41:30):
.
And now you've got MCP serversthat house all of that context
Scan the entire code base of anexisting application using cloud
code, create a markdown of that, make that part of your MCP
environment and guess what?
Now it has context of the wholething.
It's a dip man.
That's why we're really leaninginto it now.

(41:50):
Without MCP servers, I thinkwe'd still be really skeptical.

Speaker 2 (41:55):
Yes, no, I like it that for the whole context and
it gives you the opportunity nowto even use multiple MCP
servers for context, becauseeveryone's popping up with
document libraries, like you hadsaid, it's even from our world.
I've seen individuals trying toset up MCP servers, again with
the context of an entire codebase.
Then you have one for theentire context of learning

(42:16):
material, the libraries ofdocumentation, and then now you
can work with these together andit's.

Speaker 3 (42:23):
It's unbelievable.
Now this is where cybersecurityrisk comes into play and where
I think standards need to beclearly defined for an
organization.
Open source MCP servers are themost likely candidates for swap
squatting and malicious codefinding their way into a
solution.
So it's different when we'retalking about Microsoft Azure

(42:46):
MCP.
That's different To me.
That's going to meet thestandard because that's
Microsoft's MCP server, so Ithink that's something that
developers have to be verycareful about.
What MCP servers are theybringing into the ecosystem?

Speaker 1 (43:04):
A lot to consider.

Speaker 3 (43:06):
Man, oh man, I'm telling you.
And then I think softwarecomposition analysis becomes
more and more important too,because you know the developers
are deciding on their librariesand frameworks traditionally,
but if you're letting AI makethose decisions for you, are you
aware of any open sourcesoftware licensing risks that
might be part of your solution?
You know there are some opensource libraries out there that

(43:31):
have strong copy leftrequirements, meaning you can't
monetize that solution unlessyou make your solution available
as an open source product.
So these are things that seniorengineers and people that's
been in this industry for a longtime, these are things that we
think about, that clients don'tthink about.

(43:52):
Now, if the client's cuttingthe experts out of the
conversation and going to createtheir buy coding, they have no
idea what risks they might be in.

Speaker 1 (44:03):
I'm sorry, there's a lot to think about now as you're
building all these things out.

Speaker 2 (44:10):
It's a fascinating point to me and I think about
this a lot.
I see this a lot and oftentimesI wish I could go back to 1793,
have electricity and hot waterand just work on a farm, because
life was much different, butnot to shift a little bit.
But it's something that yousaid is they rely on the experts
, and I get to speak with a lotof professionals in the software

(44:31):
industry and they're alltalking about the human language
, as we spoke about a fewmoments ago, becoming the
language of the future, thatworking with MCP servers, as you
had mentioned.
They can keep the context andyou're orchestrating a bunch of
junior developers and I'm sayinga lot of this loosely.
I don't want anyone to listento this and say, wow, you're
using the wrong terms or you'redoing something.
I'm just trying to say,generalizing, in simple terms,

(44:54):
what someone's doing or theperception of it.
So now you're relying on anexpert to manage a bunch of
juniors.
I'll ask this question to a lotof individuals and I think you
can see where it's going is ifwe no longer need juniors, right
, and we need the experts, howdoes someone get to be a junior
to an expert?
When AI first started, what Iwas saying.

(45:16):
To be honest with you, just tobe transparent.
I said AI back then thelanguage models was wiping out.
The middle.
Juniors could come in.
They still had function.
The individuals in the middlewere where the language, the
large language models even ifyou're using local models on
your own were taking out sort oflike the middle, where the
language, the large languagemodels or even if you're using
local models on your own weretaking out sort of like the
middle group, and then you stillhad your experts coordinating

(45:38):
the middle and then your juniorswere learning from using these
tools and then being mentored bythe experts.
But now it's almost likeeverybody wants an expert to
manage all of these juniors sothey can do code reviews.
How do we get experts if we nolonger need juniors?

(46:01):
It's a rhetorical question in asense, but do you see this
yourself?

Speaker 3 (46:07):
Absolutely.
We talk about it a lot.
One of my peer groups that Imeet with on a monthly basis we
all own software consultancies.
We're a non-competitive market,so it's like a.
It's a great peer group becausewe can have open discussions.
We're not directly competingwith each other, and the topic
of training young talent andwhat that looks like comes up

(46:31):
often and none of us havefigured it out, so I don't know.
It's different now.
It's really different.
I think that your futuredevelopers will be groomed by
seniors that have learned toproperly orchestrate the AI
tools and they'll be taught howto orchestrate.
They're not going to bespending a lot of time in the

(46:54):
syntax.

Speaker 2 (46:56):
You're right.
I've heard this argumentbecause the software development
industry and again, softwaredevelopment industry doesn't
actually mean you have to createcustom applications or you're
building applications.
You could be in the softwareindustry if you're creating
enhancements to an existingapplication, such as ERP
software, or creating extensionsto other products.

(47:19):
So it's all encompassing andoftentimes that's what everyone
says is the software developmentindustry is going to change.
You're no longer going to needto code, You're going to be more
of a orchestrator, promptengineer, all these things that
people say.
But if you go back to what wetalked about a few moments ago,
you need to have an expertreview and analyze the code to

(47:44):
make sure that it's not to makesure that it's sound, it's not
malicious, that it's using allthese proper syntax and proper
libraries.
So now they don't have thatdevelopment experience to do the
review, they only have theorchestration experience.

Speaker 3 (48:01):
Yeah, I have to wonder how, I don't know.
What I'm thinking about rightnow is software composition
analysis, and there are tools,some really good tools, out
there.
We use Cast Highlight, forexample, for a lot of our
customers.
I think you're going to see AI,checking AI.

(48:23):
You know Cast Highlight, forexample.
It does a great job of not onlymanaging the basics around
software, building materials anddoing CVE detection and
fingerprinting, open sourcesoftware licensing risk and has

(48:52):
a workflow built into it whereyou can say no, this isn't a
risk to our business because ofthe way we're using this
particular custom softwaresolution.
We're not in violation of thatlicense attached to that library
or that framework.
So I think we're going to seeAI tools assist in workflows to
check the quality of the code.

(49:13):
I think that's just where it'sgoing.
But we're all going to have tothink how much do we trust what
it's telling us?

Speaker 2 (49:23):
I think it gets to a point where how much do you
trust that the car you'redriving or the vehicle that
you're driving is going to workwhen you press the brake?
Yeah, I mean, it's one of thosethings, driving or the vehicle
that you're driving is going towork when you press the brake?
Yeah, I mean, there is going,it's.
It's one of those things.
You there's going to be acertain level of trust to using
these tools.
Like we have the trust now acustomer of any of ours that's

(49:44):
using something that we'veimplemented.
They're going to trust that itworks properly.
So that's what that's kind ofgoes back to.
What I'm saying is like therehas to be a certain level of
trust and understanding to thesethings and get to a point.
But then now you mentioned it'sit's just sort of ai is going
to train or check ai.
What if we get to the point nowwhere ai creates stuff based

(50:09):
upon what it knows and nobodyelse is creating anything new,
because everyone's using ai tocheck and create stuff?
Do we just stop advancing?

Speaker 3 (50:18):
we might.
You know, I I describe ai asthe great homogenizer.
It because that it's justtaking a blend of all the
information it's been trained onand it's going to give you the
yeah, it's going to give you themean.
It's going to give you the meanit's going to give you okay,
this looks pretty good.

Speaker 1 (50:35):
It's a next step to human evolution where we're
doing less of that.
I always believe that with theAI agents, I think more likely
it's going to be replacingspecific positions.
I start seeing a lot of theseagents coming out that is
targeted towards a specific roleof the organization.
So we see that right now, brad,right for ERP, and it'll come

(51:00):
to a point for people to trulytrust and I talked about this
with other people as well whereyou trust it better when there
is some form of familiarizationof human element.
We talked about why is roboticshaving a human face right?
Because we trust more.
We trust it better when there'sa human face.

(51:21):
So right now, we're hearing itby voice, right.
We hear it from chat, gpt, grokright, they're talking to you
like a human person.
So you start to feel like I'mtrusting this now because it's
giving me answers.
It may not be right, maybehallucinating, but I'm trusting
because it's a voice.
I can't imagine to a point whereyou have an AI agent, a

(51:42):
developer, right, an agent thathelps with your development,
that has a virtual face, as ifyou're talking like you and I
and we're vibe coding, or thathas a virtual face, as if you're
talking like you and I andwe're vibe coding, or maybe you
and I, maybe, matt, you're theAI agent with the virtual face.
I would know that I would thinkyou're real.
Then we can just have aconversation here.

(52:03):
This is what I want to do.
What can you?
And then you can reason with me.
That would be amazing to see.
Then you can trust it, becauseI'm like, yeah, matt looks like
a real person.
I could trust CZ, but I thinkwe're coming to that point soon.

Speaker 2 (52:20):
You're taking it to a whole new level.
Because, now we won't even needany human interaction, Because
I'll just create a Matt that hasa face.
I can sit there and say hey,Matt, do you want?

Speaker 1 (52:31):
can sit there and say hey matt, do you want to watch
the?

Speaker 2 (52:32):
game with me server right like there's your.

Speaker 3 (52:34):
Watch the game with me I need to sell my business
and go sell firewood.
That's I'm in shop and sellfirewood.

Speaker 2 (52:40):
I'm gonna have to do something different, man I hate
to say this in one respect andagain to be um.
I've been doing this for a longtime and I'm almost happy that
I'm on the tail end of my careerbecause, like you said, it's.

Speaker 1 (52:59):
We were just talking about this, brad, where you have
two systems that are talking,two ERP systems talking.
The idea that I had in my headwas you know, if you create an
agent that coordinates withother agents and just have two
businesses talk to each other,there's no reason to interact.

(53:19):
They're going to know what'sbest between two systems If you
allow them to talk to each other.
They're going to know what'sthe most efficient way to do
business together that benefitsboth.
I don't need it.
I'm going to play golf with theother owner.

Speaker 3 (53:37):
You need to develop that middleware solution and
then you know, be the first, youguys can be the first
billionaires without anyemployees.

Speaker 2 (53:44):
Well, that's exactly what it's going to be.
Because, if you even think it'slike the simple functions of
procurement and planning, yeah,if you can have two systems,
talk to each other, mrp, and youknow what you need for
materials, your requirements,planning, your procurement hey,

(54:04):
I need all.
I mean it's a calculation,basically Right.
And it's calculation based onwhat.
If you think of someone thatyou have doing forecasting and
again, I'm not trying totrivialize what anyone does,
don't get me wrong, it's not theoh, I just know I need this
number they're coming up with arationale for needing a number,
somehow based upon what and aslogistics improves and becomes

(54:25):
more reliable, yes, there'salways going to be exceptions
due to weather.
As such, you I could see christo your point, something in the
middle saying hey, run your mrp,tell us what you need, we'll
create the orders.
We'll tell the shippingdepartment, which again is
becoming autonomous in a lot ofplaces, to ship the product,

(54:46):
you'll get the product andyou're done.
Yep, yeah.

Speaker 1 (54:50):
Yeah, there's auto-invoicing.
Now I mean email responses areautomated.

Speaker 2 (54:56):
Matt, if you're working on that, let us know.
We'll jump in with you there,right, let's have a whole other
conversation, fellas.

Speaker 1 (55:01):
We're going to improve the MCP right.
We're eventually going to talkto each other, because we're
going to have all theinformation they need.

Speaker 2 (55:08):
Well, the MCP is a standard, so if you can have
again, you have APIs.
If you start creating thesestandards for systems, we could
go down this whole world.

Speaker 1 (55:16):
I know, man, I'm not saying anything more.

Speaker 2 (55:18):
You heard it here first, right somebody will
develop this before, before thisis even published now, because
we put it out into the universe,as they say.

Speaker 1 (55:27):
So it's that's what I want.
I'm going to come to you, buildme this.
You got it.
We can buy a code in an hour.

Speaker 3 (55:35):
How hard can it be?

Speaker 2 (55:38):
Exactly that's.
And it's a funny thing, it's uh, you know I'm a little loose
with some of the things I say,but I used to have people say
how hard can it be?
And I used you can do ityourself.
If it's that you know, you haveto ask that question.
How hard is it for you to do ityourself?
And you know, get back to mewhen you're done Now, because

(55:59):
there is a certain level ofexpertise.
So AI is changing the softwaredevelopment world, software
development as a service.
I like the concept, I like theidea and it's great to hear that
the buy versus build mentalitythat you have, or the approach
that you have to make sure thatthe value is there for the

(56:19):
customers, is important Becausea lot of times, business
decision makers who are goingand undertaking these
investments in assets again,I've picked up so much from this
discussion Now they can see isit really worth the investment
in that asset or should I justbuy something?
Or do I really need to havesomething customized for my

(56:42):
business?
And I am not going to say thatnobody needs customization
because that whole oh, just useout of the box, right, that's
what everyone talks about.
Yes, use out of the box as muchas possible From a framework of
thinking through your processis the way I like to say it is.
Let's look at the out of thebox system.
Can you use it?
Does it do what you need to doand what do you need to change

(57:06):
to do it?
Therefore, you don't have thecost for the customization.
But don't use it to say thatyou have to disrupt and change
your entire business to fit theout of the box.
Change the approach to be.
Let's analyze our processes,analyze our business to see if
we should redo our processes,because I'm sure you go through
it when you're identifying theproblems of businesses.

(57:28):
Why do we print this reportevery Friday at nine o'clock?
Because it's the way we'vealways done it.
The way we've always done it.
Yeah, you know.
So going through theseexercises really gives you the
opportunity to validate whyyou're doing something versus
just saying you need to changeyour business process.
So it is.
It's more of an education onagain, it's another thing.

(57:52):
It's almost like you need tohave educational seminars on how
to choose a software productand what to consider when
choosing a software product, andthis is for anything.
It could be ERP software.
It can be, as you had mentionedweb software for finance.
It can be anything that youneed to gain some efficiencies
in your business.

(58:12):
So I could talk about this AIstuff forever.
My mind is still.
I have so many things to bringup and to say.
I just don't know.

Speaker 3 (58:21):
Right now I'm terrified and excited and
motivated all at the same time.
It's just a mix of emotions.
I think what's going to be alot of fun is doing rapid
prototypes and proofs of concept, because that is so feasible
now, just to model things outvery quickly, especially on the
custom side of things, just whatwe've done in the past couple

(58:43):
of weeks with an internalsoftware product that we're
creating.
It's months worth of work donein a week.

Speaker 2 (58:56):
I've heard that, like individuals are doing stuff in
like it's, it's exponentiallyfaster.
It's not just oh, we've shavedoff a week.
It's something that could takemonths is now done in hours per
se yeah, we've got a customerthat has an iowa and web
application.

Speaker 3 (59:12):
It's an overall platform for a specific market.
It's very niche and we justlaunched all of this last week
and it's been months worth ofwork.
And a new feature that weneeded to roll out, a sizable
feature.
We estimated 65 hours ofengineering time Just to get to

(59:33):
a proof of concept ofablefeature.
We estimated 65 hours ofengineering time Just to get to
a proof of concept of thatfeature.
We used a couple of differenttools that we've been exploring.
We got nearly production readyon that feature in 25 hours.
Not proof of concept at 65hours, like we thought.
Nearly production ready in 25.
Wow, so it's coming.

Speaker 2 (59:58):
So, with that, right now, we value time for money.
Yeah, and the question that Ihave with this, just to build
with it, I know a lot have thisthought, so I want to paint the
picture properly.
A lot of individuals say, well,we'll use AI, you'll have more

(01:00:18):
time because you can do thingsfaster.
But we all in this world mostlyhere in the United States and
also in other countries theshift may be different, but it's
you work for 40 hours, we payyou for 40 hours and, okay, well
, you have tasks to do.
You do your tasks.
Oh, you have free time.
Here's more tasks.

(01:00:39):
How can we change the timevalue of money to get the free
time that everybody's thinking,because it's now more and I've
talked about this on otherepisodes and I try to ask this
to any conversation that we havewith talented individuals such
as yourself that work with thisAI space, with the efficiency
how can we change the thoughtprocess so that the efficiencies

(01:01:03):
that we gain we're not puttingadditional pressure on those
completing the tasks?

Speaker 3 (01:01:08):
I don't know if we can You're talking about a
massive cultural shift to beable to pull that off.
I think that small,independently owned businesses
probably have the bestopportunity to change the way
that they operate, to achievesomething like that where we're
generating the value, we'rebeing paid for the value that

(01:01:29):
we're generating, and maybe wedon't need to work 40, 50, 60
hours a week any longer andpeople can still be paid well.
But I think that's a tall order.
It's going to be interesting.
I'm in the professionalservices industry, so even
though we have our unique model,there's still a component of
trading hours for dollars.

(01:01:51):
So we also need to be thinkingabout well, either we're going
to need a lot more work becausewe can move things through the
pipeline the SDLC, so muchfaster Am I going to see a lot
more work come through that SDLC?
Or do I need to start thinkingabout how we change the way that
we price and we structure thesedeals?

(01:02:13):
And so now we're exploring andthis is early man, I don't have
any decisions on this or evenreally enough clarity to speak
to it in a meaningful way but wehave to start thinking about
value-based pricing.

Speaker 2 (01:02:28):
Yes, yes yeah, yes, yes, yeah.
That's a great way to put itbecause and I think you're on
the right track and why?
That hit me with the softwaredevelopment as a service,
because it's a service you'repaying for, not time that you're
paying for, but internally, areyou still budgeting it for the

(01:02:49):
hours for the individuals to dothe tasks, or or is it a value
you don't have to tell me?
This could be a rhetoricalquestion, just that I throw out
there, because I see there beinga shift of we're offering a
service, someone's paying forthe service.
It doesn't matter how long ittakes.
And again, I say this all thetime I want a deck built on my
house.
I have a contractor come out.

(01:03:11):
He doesn't tell me it's 60hours, he tells me it's $2,000.
And as you go through, if youhave changes, you run into
issues because they may berotting wood or some other
things that weren't visible.
You do a change order, anadjustment or figure out how to
mitigate that, but he's stillpaying his team by the hour to

(01:03:33):
complete the job.
So I see software developmentcompanies uh, in a loose term as
a classification doing the samething yeah, it'll still exist,
I guess at that point.

Speaker 1 (01:03:46):
Right, it's just a matter of like how you're how,
but that's what we need to shift.

Speaker 2 (01:03:50):
Yeah right, because we'll sell services for a fixed
dollar, assuming that that fixeddollar will cover our expenses.
Our expenses are going to betime-based for the most part,
because that's how our cultureis.
So how do we fix that?

Speaker 3 (01:04:13):
all of our employees are salaried, so my costs to
operate the business are fixed.
The sales and delivery side hasto maximize the investment we
made in the operating side ofthe business to generate the
margin that we need from thebusiness.
So if we can change theconversation with the prospect

(01:04:33):
to hey, we worked with you in aconsulting engagement Maybe
that's time and material still,I don't know.
I'm just spitballing ideas herethat we've evaluated what it
would mean to create this customsolution for you and the impact
it's going to have on thebusiness.
You're going to see $100,000 ayear ROI and we're going to

(01:04:54):
price this based on valuecreation.
You're going to pay us X andthen we deliver it and you know
if they're going to get $100,000a year return on that
investment, what percentage ofthat would they be willing to

(01:05:14):
pay the vendor who created thatsolution?
I don't know if it's going towork.
I'm probably going to starthaving conversations like that.

Speaker 2 (01:05:22):
Well, it's something I think that needs to be.
I think there will be a radicalshift in business in the near
term.
Either it's going to be thesmall companies or startups like
you had mentioned.
You know, chris and I would bethe next billionaires now with
no employees, uh, for example.
But I I say this all in jest,but it's it is it's going to

(01:05:42):
have to be a shift because I Isee a change in output versus
time, versus value, versusdollars.
There's a lot of variablesthere and everyone talks about
oh yeah, I would give you abetter quality of life.
What I'm seeing right now, it'sjust putting more pressure on

(01:06:02):
individuals, because now they'reexpected to produce more and
those that aren't using AIaren't producing more.
So, in essence, it's like thiscycle of you need to do more
because now you no longer haveto do something.
It's scary, because I said thesame thing to someone the other
day.
It was yesterday.
We were having a conversation.
I use AI to write an email tosomebody.

(01:06:24):
They're using AI to create aresponse and then it comes back
to me.
We're getting to that world.
There was everybody's using AIto write these lengthy emails
where it's almost like justwrite the one sentence and just
call it a day of.
Okay, I need.
I need a hundred dollars forthree hours to do this task
instead of this.
Oh, how are you doing?

(01:06:44):
We have to do all this.
That's but that's is.
It goes back to what you'resaying is AI is checking AI.
But look at these emailexchanges that I'm having.
And we talked about it.
It was all AI generated.
It's wild.

Speaker 3 (01:06:58):
And then it's like I think there's a piece that
people are missing right now too, and that is if you look at
hopefully you guys have these inyour organization, we have them
in ours an acceptable usepolicy for any AI tool that
you're bringing into thebusiness.
It creates all the guardrailsand training to make sure people
stay within those guardrails.
Well, a big important part ofour AUP is that the human being

(01:07:23):
is still responsible for what'sgenerated in output.
So if you can generate a lotmore stuff really fast through
AI, you still have to take thetime to look at it.
So the cognitive load is can befairly significant, because you
got to check everything thatit's generating before you like
I'm sure you read that emailbefore you send it, brad.

Speaker 1 (01:07:46):
Yeah, yes, you know, I do, I do read the email.

Speaker 2 (01:07:49):
make sure that the, the, the grammar is correct
because, again, the AI is notperfect, Even if it gets me 80,
90% of the way there, but Istill go through grammarly or
something.
It still will fix the grammar,but I do read it.
I don't want to say it's, I'mjust saying conceptually.
I didn't have to sit there anddraft the email.
I drafted the email in a matterof seconds.

(01:08:10):
I now read it, edit it versushaving to think of it completely
.

Speaker 1 (01:08:17):
Yeah, I think the flow of thought of writing that
it's much faster because youjust have to give it a proper
prompt.
It's to say, I need to respondto this email, blah, blah, blah.
And then you do have to read tomake sure the flow looks good
and it's like, hey, it's notfilling in incorrect information
.
So, a little bit, it's stillsome time there, but it's still
better than in many cases whereyou're writing it up and you're
you know, sometimes your thoughtgoes in different directions

(01:08:38):
and then it doesn't look likeit's flowing right.

Speaker 2 (01:08:41):
So co-pilot or ai can certainly help make it easier
for the reader to understandwhat you're trying to say but to
matt's point, the cognitiveload is greater on individuals
now, and that's the point I wastrying to get at is they're
responsible for more output.
Therefore, there's a biggerpressure today to review that

(01:09:05):
output in a faster time.
I don't know, have you?

Speaker 3 (01:09:11):
guys checked out the Notebook LM at all.
Yes, yes, we're using that alot in the early stages of the
sales cycle and talking aboutcognitive load, and I can create
and we're using enterprise soit doesn't.
You know, we can share itacross the organization, et
cetera.
Notebook LM Enterprise andwe're using enterprise so we can

(01:09:32):
share it across theorganization, et cetera.
Notebook enterprise.
I can connect sources,background information on a
company, a prospect.
I can drop in PDF outputs ofLinkedIn profiles of the
decision makers.
I can drop in transcripts fromthe first-time appointment sales

(01:09:53):
calls.
I can drop in transcripts fromother conversations that we have
with them in those early salescycles and drop in a template of
a project brief.
I always use project briefs toeducate the team before I bring
in the sales.
You know the engineering partof the sales cycle.
I can put all that together ina few minutes.

(01:10:16):
But then I've got all of thisinformation in there that I just
put in there in a few minutesand I generate the audio podcast
style output.
I got a 20-minute podcast I gotto listen to and then I got to
proofread the brief and I got tolook at all the information,
maybe go on and correct somethings.
Sometimes the, the, thetranscripts are a little wonky.

(01:10:38):
You know, if it doesn'tunderstand things or it doesn't
understand acronyms, you go editthose things and then you
generate some additional outputfrom notebook that you can use
in the in the sales cycle.
It's fascinating that you canuse in the sales cycle.
It's fascinating.
But the cognitive load man isno joke.
If I'm in a bunch of meetingsall week long and I'm doing that
all week long.

Speaker 2 (01:11:00):
I'm tired, yes, no, it's exhausting, it's tough to
keep up with, but it's achallenge for individuals to
come through when it's a strangeworld that we're living in.
It's a new world and I'mfearful that our brains can't
keep up with it, because ifhumans have been on the planet

(01:11:21):
for 6,000 years, if you look atthe advances in technology, it's
really within the past 40, 50years where we've had the most
advancements.

Speaker 1 (01:11:33):
A fraction of that.

Speaker 2 (01:11:34):
And tooling, and it's almost like the time in between
.
Now, mcp servers didn't existmonths ago.
Who knows what will be inventedtomorrow based upon this?

Speaker 3 (01:11:43):
Well, when quantum computing comes out, generative
AI is going to be a whole.
It's going to be.
I don't know, I don't even have, we don't have enough time to
even dive into that topic, butquantum's right around the
corner.

Speaker 1 (01:11:56):
Yes, and accessible.
Yeah.
Yeah, that's a whole topic initself.
I love that.
Yeah, I think my mind is you'reboth overwhelming me here.

Speaker 2 (01:12:08):
It's, you know, stemmed from a conversation that
I would, that we wanted to havea buy versus build again.
It's, uh, it's what I loveabout this.
It's um it's dynamic down theroad, yeah well, I, I see, and
the cost of this will come down,it's I'm sure the the cost of

(01:12:29):
ai right will come downsignificantly, I think once,
once they have the interest it's, I'm sure it would.
Yeah, the the cost of ai rightwill come down significantly.
I think once, once they havethe interest, it's just like
anything else.
I remember paying for four gigof ram.
I paid several hundred dollarsback in the day and even for a
1200 2400 baud modem.
I was like whoa, I just spent300 on a 2400 baud modem.
I mean, I think it would be thesame thing, like this email

(01:12:49):
used to be the same.
You used to pay by the email.
People didn't even realize thisright Again this is how old I
am.

Speaker 1 (01:12:55):
Younger generations didn't know Like personal PCs
were.
Like three grand, four grand atthat time, and now it's like
Three grand.

Speaker 2 (01:13:02):
What are you talking about?

Speaker 1 (01:13:07):
Oh yeah, the old original desktop.
I remember going to like whatthe Circuit City or CompuServe.
I remember my dad and I went tothe store and just seeing these
computers were like $3,000, youknow, two to $3,000 for a
desktop that could barely hold afew photos.
You know, in today's yeah, thatwas exactly it.

Speaker 2 (01:13:23):
That's what I mean.
That's why I was saying Ilaughed at the three grand,
because it's a $10,000.
Anyway, it's a nice walk downMaryland.
But, matt, thank you for takingthe time to speak with us.
I would like to see if we couldschedule a follow-up in a
couple of months to maybe talkabout the buy versus build but I
have a feeling that things willbe significantly different in a
few months and also maybecontinue on your journey of

(01:13:46):
value of time.
Time, value of money, I guessyou could call it as well as
some of the other things thatyour team has experienced with
AI and their workflow.
But thank you again.
We appreciate you taking thetime to speak with us today.
If anyone would like moreinformation on the services that
you provide, also your returnon investment calculator, which

(01:14:08):
I'm going to go take a look atafter this as well, to be honest
with you, see if we canincorporate using that, because
I like the approach that youhave what's the best way to get
in contact with you?

Speaker 3 (01:14:16):
Yeah, so they can visit our website,
wwwredhawk-techcom.
They can just search for MattStriplehoff on LinkedIn and
reach out to connect.
Happy to have a conversationwith your listeners, you know if
anybody is interested.
But yeah, this has been a lotof fun.
Yeah, I would love to come backand spend time with you guys.
It was a blast.

Speaker 2 (01:14:38):
Excellent, great.
We appreciate it.
Again, thank you very much foryour time.
It's truly the currency of life.
Once you spend it, you can'tget it back, so we truly do
value any time that you spendwith us, because it's time
you're not doing some greatthings with AI at the point, so
enjoy.
Thank you, ciao, ciao.
Thanks, matt.
Thank you, chris, for your timefor another episode of In the
Dynamics Corner Chair, and thankyou to our guests for

(01:15:00):
participating.

Speaker 1 (01:15:01):
Thank you, brad, for your time.
It is a wonderful episode ofDynamics Corner Chair.
I would also like to thank ourguests for joining us.
Thank you for all of ourlisteners tuning in as well.
You can find Brad atdeveloperlifecom, that is
D-V-L-P-R-L-I-F-E dot com, andyou can interact with them via

(01:15:25):
Twitter D-V-L-P-R-L-I-F-E.
You can also find me atmatalinoio, m-a-t-a-l-i-n-o dot
I-O, and my Twitter handle ismatalino16.
And you can see those linksdown below in the show notes.
Again, thank you everyone.

(01:15:46):
Thank you and take care.