Episode 440: The Future of Development: Is Software Disposable?

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_04 (00:01):
Welcome everyone to another episode of Dynamics
Corner.
This is a very exciting episodein how we use AI in our
day-to-day lives.
I'm your co-host, Chris.

SPEAKER_02 (00:11):
And this is Brad.
This episode was recorded onDecember 12th, 2025.
Chris, Chris, Chris.
Almost at the year end, andwe're finishing up strong with
this year.
And here we are again withanother amazing episode talking
about AI, software development,business use, and the future of
business with AI.

(00:32):
With us today, we have theopportunity to speak with Matt
Striploff.

(00:55):
Nice to uh see you again.
It's uh it's uh it's been awhile, and I feel like the world
has changed in that short while.

SPEAKER_04 (01:05):
We talked about that last time.
It's like you know, for a shortperiod of time, a lot has
changed.

SPEAKER_02 (01:11):
That's exactly what we said.
Who knows where the world wouldbe with this in a few short
months, and here we are, a fewshort months later.
The world is still here, yeah.
But man, what a whirlwind it hasbeen.
And I saw something this morningtoo.
I think it was what a year agotoday that agents, the concept

(01:32):
of agents was introduced, Ibelieve.

SPEAKER_01 (01:34):
Yeah, it can't it seems like just yesterday.

SPEAKER_02 (01:38):
It does, but I I mean it's but it's look at
everything that's happened sincethen.
It does seem like justyesterday, but look at all the
advances of uh in technologythat have happened since then.

SPEAKER_01 (01:48):
So oh yeah, you know, and I I know I was excited
about all the AI first SDLCmodeling work we're doing at Red
Hawk, and we talked about that.
Um it's great to see you guys,by the way.
I'm excited to be back on yourpodcast.
We had a blast last time.
So uh the progress we made isjust it's uh it's insane.

(02:09):
Like I'll give you an example.
Um, since we're talking aboutagentic and agents, we have
successfully developed, deployedinto production, and now have
automated agentic workflow thathandles the software bill of
materials, detects uh commonvulnerabilities and exposures in
those libraries and packages,and automates remediation all

(02:32):
the way up to the pull request.
Wow.
But stuff like the softwareengineers, I describe that as
doing the dishes.
You know what software engineerswant to do?
They don't want to do thedishes, I don't want to do the
dishes.

SPEAKER_00 (02:42):
You can do it.
You know what I mean?
So we've automated.
You can prep the meal.
Don't want to do dishes, man.
I didn't want to prep the meal.
I just want to show up and enjoyit.

SPEAKER_02 (02:49):
But hey, that's that's that's it.
It's um uh before we get intothat, because there's a lot, uh
I don't even know where to beginwith this because I end, you
know, knowing what you guys doover at Red Hawk, and you know,
from the conversation last time,I wanted to I wanted to get your
take on a few things.
But before we get in that, canyou uh tell us a little bit
about yourself?

SPEAKER_01 (03:10):
Yeah, Matt Strippelhoff, uh CEO, co-founder
of Red Hawk Technologies.
Uh, we are a softwareconsultancy.
We develop, support, maintaincustom business applications,
uh, which is that takes avariety of forms.
It can be uh agentic workflowslike middleware solutions that
sit in between systems.
We used to call that middleware,but now we're calling it agentic
workflows because AI is a bigpart of that flow.

(03:33):
Um But uh we also develop a lotof custom field service
applications.
They might be uh web portals,mobile applications, things of
that nature.
Our primary focus is serving uhfast growing or growth-oriented
privately owned mid-marketbusinesses.
Um and what a great time to bedoing it.
Because a few months ago, guys,I was like, maybe a year ago, so

(03:55):
what's a what's AI gonna do tomy industry?
Yeah.

SPEAKER_02 (04:00):
But uh a lot I am speechless with what AI is doing
to the industry and some of thestuff that I have seen firsthand
since our last conversation.
It comes up to I had some textmessages this morning with some
peers and individuals, and myquestion to them was who writes

(04:22):
code?
You know, it's just like if youthink about that, it's it's it's
come to the point where uh theamount of code that you write,
in my opinion, is what I wantedto see in your uh from your
perspective, what you're doingwithin your organization and
maybe some of the other peersthat you may have.
Yeah, I I see there's a shiftfrom the development cycle.

(04:46):
And again, it's not global.
I mean, I I've talked to somedevelopers who haven't even used
it yet, and I've talked to somedevelopers who use it all the
time.
So I'm not going to say thatthis is what everybody's doing,
but there's a general, I thinkgenerally, if you haven't
started using it yet, I'd be alittle concerned.
Uh, and if you uh all in, great.
A lot of people are still inthat journey of ramping up or in

(05:09):
the what I call like sort of inthe middle still, right?
So you have the the extremes andthen you have the uh I'm still
dabbling with it, uh, you know,maybe a little bit more, a
little bit more than theautocomplete, right?
Because everybody has that youhave that cycle of you start
working with AI and it justbecomes as uh somebody that we
talked with our teeny siderssaid, it's like a fancy
auto-complete, right?

(05:29):
And then it and then youprogress up to a little bit
more, then you start workingwith agents and you know,
talking to the agent, saying,Okay, do this.
And then you start getting intothis real workflow of no, then
you start creating instructionfiles, uh, agent files, and then
you have uh multiple agentsrunning in the background.
Uh so it's it's you get all theway up there.

(05:50):
But uh to bring that back to thethought of it is how much
coding?
Oh, okay.
This is where it was.
How much coding do you see beingdone?
And I wanted to um, you know,from the development point of
view, that's being done by thedeveloper versus the AI.
And of that, how much of it ismore a splight adjustment of
code versus all out coding?

SPEAKER_01 (06:11):
So I'm gonna frame my response based on uh two
different types of projects.
You've got production largescale applications that you are
you've got a product roadmap andyou're kind of building things
out and you're you're umevolving that into a solution.

(06:32):
The the percentage of code beinggenerated by AI is at this point
kind of a kind of a guess on mypart, but maybe 50-50, maybe you
know, where the the the softwareengineers primarily focused on
being the orchestrator and thearchitect and then providing the
appropriate level of context inorder to get the intended
outcomes.

(06:53):
That's really where they need tofocus their expertise now.
Um when we're doing net newbuilds, it's probably more 80-20
where the agent's doing 80% ofthe coding.
And the work that they're beingdone is is uh for example, we
have we have eight uh highfidelity proof of concept

(07:14):
projects uh that we're executingright now that we will wrap up
by the end of December.
And the uh the suite of AI toolsthat we're using are allowing us
to maybe we're only doing 20% ofthe coding on those.
And all the front end is gonnabe done, by the way, uh uh as
far as these proof of conceptsgo.
And then when it's time for theback end, when you start

(07:36):
building out the context and andputting together your plans, you
can use AI to help with theplanning.
And depending on the systemsthat you're gonna be integrating
with, like Dynamics, CRM, etc.,or maybe it's Azure AI, what and
we've got a number of thosetypes of projects going on.
You're gonna you're going toinclude uh MCP servers in that

(07:58):
mix, which provides thatadditional context that's
necessary for the AI tools to dothe plans.
And the way tokenization'sworking now with these tools,
they can handle a significantmore amount of context.
And that was really the issuewith these tools early on, is
they start to lose track becausethey can only handle so much

(08:18):
through tokenization, so muchthey can only retain so much
context at any given point intime.
And you start to get further andfurther and further away from
the original strategicobjective.
Well, now your detailed projectrequirements documentation is
part of that context.
Everything that you've done tothat point is part of that
context, and your MCP serversare part of that context.
So we're seeing more and moreand more of the actual syntax

(08:42):
being written by AI.

SPEAKER_02 (08:44):
It's it's it's insane.
And to hear what you had said issort of what I wanted to talk
about.
You you you unpacked a lot inthere.
Uh you you mentioned a lot ofthem, we have to kind of unpack
it.

SPEAKER_03 (08:54):
Okay.

SPEAKER_02 (08:54):
So, first, the now with the context, the
tokenization, you know, you havea uh a larger uh number of
tokens, right, that you havewithin your context.
Now, I I've just finishedreading the book Vibe Coding by
Stephen Yee.
And geez, I can't forget theother author.
I don't even know if I said hislast name properly, but that's
what I do.

(09:15):
And they they summed up what atoken is perfectly, because
that's the first question a lotof people have that may or may
not understand it.
And it's it's a token's not aword, a token's not a character.
It's really how they break downwords, and they said on average,
you can figure it's like fourcharacters is a token, I
believe, right?
So when you think in thecounting of tokens in your
context, it's a certain numberof characters, basically.

(09:36):
And then they piece thosetogether.
I don't know how it does allthat stuff, but it pieces it
together like magic and it spitsstuff out for you.
So now you can have a largercontext, and this is where it'd
be able to create thoseinstruction files or to have a
lot of that information alreadyavailable for the model to use
or your agent.
I don't even know what to callit.

(09:56):
I tell you, I feel like thepeople because I'm even now I
found myself the other daysaying, Thank you.
That was good.
Like, okay, why did I just writethat?
I literally wrote thank you.
That was good.
Now can we, you know, add thisto it?
It's just it becomes sort ofnatural with it.
So that's changed a lot.

(10:16):
We can talk about MCP service,but you started talking about
something more about softwareengineers, and that's where I
was sort of going with wherethey are in the in the roads.
It's where do you see the roleof, you know, I use the word
developer generically here.
You know, you have softwareengineers, you have different
titles, but somebody whodevelops code primarily for a
living, where do you see thatcareer trajectory based upon

(10:40):
what we know today?
And we all know that tomorrowit'll be a whole new world
again.

SPEAKER_01 (10:46):
So um I think those that adopt the tools, learn how
to extract the most value fromthe tools, understand which
tools are applicable based onwhere they are in that from
concept all the way through toexecution cycle, that SDLC
cycle, will be successful.
But they gotta be willing tolean in and blow up traditional

(11:07):
thoughts around softwaredevelopment.
You just have to get your handsdirty and get in there.
Those who are gonna besuccessful are gonna take the
time to learn how to bring thesetools together to deliver really
what their job is, is to solveproblems for their customers.
So if they can shift theirmindset to the intended business
outcome and just recognize thattheir value is not attached to

(11:29):
how much syntax they write intheir code, that's the first
thing they have to do is justmake that mental shift and start
to recognize that really theircontribution is the solution,
not the how they get to thesolution.
So I think back when there's allkinds of other examples of
advancements in technology, it'sjust happened much, much slower.
Um and I think last time that wetalked, Brad, we talked about a

(11:51):
a nail gun for a carpenter.
You know, uh you gotta startusing the tools that are
available to expedite uhthroughput.
Um at the same time, a nail gunin isn't gonna make you a
carpenter.
So you still have to have theskills and the knowledge and the
understanding from anarchitectural standpoint,
sustainability standpoint,what's the right way to approach

(12:13):
uh crafting a solution for acustomer uh or specific business
outcome that's scalable andsustainable.
So we hear a lot of buzz aboutvibe coding, and you see a lot
of really aggressive marketingfrom tools like Label and
Revelit, and they're saying, youdon't need engineers, you know.
I built an app that just likeSpotify in 30 minutes.

(12:35):
That's not what they built.
They built a prototype thatprobably doesn't have the the
the uh any of the architecturein place to actually be
sustainable in a commercialenvironment.
Yeah.

SPEAKER_02 (12:51):
So again, your point just to to pick up what I've
heard is it's not the tool thatyou use, it's the product that
you deliver that becomesimportant.
So the role of a developer isshifting to you still need to
understand the code, you stillneed to understand architecture

(13:12):
and design, but you're justgoing to use different tools to
deliver the solution.

SPEAKER_01 (13:17):
Yeah, yeah.
And we've not successfully beenable to just use purely vibe
coding in any effort.
Um, not not all the way throughto production.
So uh we're able to use it forrapid prototypes like high
fidelity prototypes concepts.
Yeah, but then even then it itgets a it can get squirrely

(13:39):
pretty quick.
Like you might only be a couplehours into something and you
can't figure out why thecalendar selection feature is
wonky from a visual standpoint.
You're gonna tell it over andthis is a specific example I had
last week, by the way.
I'm trying to get the days ofthe week to line up over the
columns of the of the weekdays,you know, just on a little
calendar selection feature.
Doesn't matter how many times Itried to get it to correct that

(14:01):
formatting issue, and it wouldyou'd see it write code, oh
yeah, I understand.
Very kind to me.
You know, yeah, that makes a lotof sense.
So it's very complimentary, youknow, nice conversation.
And it would write code and itwould render and say I fixed it,
and it didn't fix it.
So um ultimately you may findyourself, and this is what our
experience has been, is that youconnect uh the source code

(14:22):
repository and then you startshifting which tools you're
using until uh you find thetools that are gonna give you
the best output.
And sometimes you're actually inthe code.
Writing it.
That's that 20%, right?

SPEAKER_02 (14:36):
So yes, no, it's I I think it but what I go with that
too is so it's it's becomesdelivering the solution.
You're talking aboutprototyping.

SPEAKER_03 (14:47):
Yeah.

SPEAKER_02 (14:47):
How do you see it?
It's how do you see it changingin the landscape of business in
the sense that you have softwareengineers if you're delivering
software, but then you also mayhave some business users or or
business mindsets that areworking with this.
Where do you draw the line?

(15:07):
Because now theoretically, youcan have some business users
that don't understand code beable to create some of these
prototypes, right?
For the developer.
So if you're working, say, youknow, I'm sitting with a
customer and we're talking aboutsomething quickly, and they say,
Oh, I have this idea.
You can basically type up thatidea and then create a prototype
for them to then deliver over toa software engineer to further

(15:30):
refine.
So is it are we getting to thepoint where we're drawing those
individuals closer?
Right.
So before you used to havesoftware engineers and you used
to have business, I call thembusiness users, but it's not
even maybe, you know, businessconsultants.
And they used to talk to thedevelopment group and say,

(15:51):
here's what I need.
Can you design, develop, give mea prototype for it?
They'd have to talk to them,they'd wait a period of time.
And now it's okay, I can come upwith a prototype.
As I'm sitting with thecustomer, it functionally works,
right?
I mean, it's again, everything'sin scale.
It depends on what you'rewriting and what you're doing.
If you want to create, you know,a simple web page that you can

(16:13):
enter some fields and savesomething, I'm sure AI can do
most of it, right?
Because I've been able to dosome of that stuff quickly.
If you want to do some other uhmore advanced stuff, again, it's
not when for the sake of thisconversation, but you know, it's
it's not a one-size-fits-allwhen we're talking about
situations and scenarios.
This is more some casesgeneralizations because there's
always an edge case foreverything.
Um, but are we bringing thosetwo worlds closer to where a

(16:36):
software engineer, instead ofbeing the person that was
sitting in the back writing thiscode, not wanting to talk with
anybody, has to become morebusiness consulting aware to be
able to talk with business usersinstead of needing someone to
translate it in the middle.
And the business user isbecoming more familiar with

(16:58):
technology and the tool becausenow they can talk to an agent
almost like they spoke with asoftware engineer.

SPEAKER_01 (17:06):
Yes.
And I I I what I would recommendto anybody in the audience that
maybe is either in the processof getting an education and and
investing in developing softwaredevelopment skills is uh it's as
important, if not moreimportant, going forward in this
career path to become a businessanalyst.

(17:26):
That's really because we wetalked last time, uh Brad and
Chris, I think we were talkingabout the the you know the
future, the most popular uhsoftware programming language is
going to be whatever yournatural language is.
That's we're pretty much therein a lot of ways.
So um connecting the subjectmatter expert, maybe it's the
it's a senior leader.

(17:47):
Like for example, I was I'mworking with a CEO at a company
uh currently, and in two weekswe turned around a prototype
that will expedite theirworkflow so significantly, his
mind was just blown.
Um because his uh the way hedescribes what his business
needs as a CEO is he's thinkingstrategic outcome uh growth

(18:10):
without having to, you know, yougotta you want to outsize your
growth compared to youroperating expense to create that
desired margin.
CEOs and you know, businessleaders are thinking along those
lines.
How do I scale in a way that'ssustainable?
And right now my team is doing agreat job, but at this current
scale, it's okay that they'vegot manual workloads and

(18:31):
spreadsheets, right?
But if you're about to go downthe path of stacking on five,
six companies a year throughacquisitions, guess what's not
sustainable?
Living in spreadsheets becausenow you're gonna multiply the
manual labor and efforts and theissues that come from copy paste
errors, etc.
Um, through that process.
So this is a way for me to kindof come back to answering your

(18:53):
question is it depends on um theleader in the organization and
their role as to how they'regonna describe the business
outcome they're trying to getto.
And a really good strategicallyminded analyst is gonna be
critical in developing customsoftware applications, somebody
who can have that conversationwith the CEO and say, okay, I

(19:14):
understand that.
Because if you're um in somecases, if you're down at a
director level, uh maybe anoperator level, somebody,
somebody who is um adopted thestandard operating procedures
that are those manual workflows,it could be very well
documented, by the way.
And their personality profile,because you know, as a leader in
my organization, we're veryaware of personality profiles

(19:36):
and where people might fit basedon their natural inclinations.
Like what do they like to do?
Are they black and white rulefollowers?
If if you and you want black andwhite rule followers who who are
just gonna adhere to the SOP andthey get really uncomfortable if
they're asked to do anythingoutside of that routine.
So if you're talking to somebodyat that level and they're
defining what they want in thesoftware, you're gonna replicate

(19:56):
the manual process.
Eliminate error probably copypaste errors because you're
going to do some level ofintegration.
But is it really going toachieve what the CEO's strategic
vision is?
Like in the in this particularexample for the CEO I'm working
with.
He has uh analysts that can onlyhandle two projects uh per week

(20:19):
maximum.
And each project takes asignificant amount of hours
because of the level of researchthey have to do and and the
process of which they develop uhthe results and ultimately
they're deliver deliverable fortheir customers.
That's a major scaling issue forthem.
And his uh KPIs for success aregonna be I want them to be able

(20:42):
to handle six or more uhprojects per week.
Without increasing the expense.
Well, yeah, he's also able toarticulate his cost per project,
which um in conversation withhim, and this is another way to
think about how AI impactssoftware engineering with
without going beyond just uhthings like cursor, right?

(21:04):
Uh is in the interview processwith the CEO, I'm capturing all
the conversation in transcripts.

SPEAKER_03 (21:12):
Yeah.

SPEAKER_01 (21:12):
Then I'm taking all that information and I'm bibing
with an AI tool to create thedetailed project requirements
document, which I need ascontext.
But I take I add another step inthe process, which is where you
get some really strategicoutcomes, which is fascinating
to me.
So I take the PRD, I take all ofthe artifacts from the client,

(21:33):
including their SOPdocumentation and sample
outputs, and the transcripts formy conversation as to what's
important to him, what is reallythe business goal.
And I'm going beyond himdescribing features and
functions, because we tend ashuman beings to start with the
solution and not the problem.
So if you can get somebody inthat flow, whether it's a BA or

(21:53):
a really strategic-mindedsoftware engineer, they need to
work with the stakeholders tounderstand the business problem
and desired outcome, and thenshift the conversation back to
features and functions.
But let's use AI to recommendsome features and functions
based on all that context.

SPEAKER_04 (22:11):
Yeah, that's a great point that you uh you called out
because um right now uh I'veI've spoken to uh different
individuals where they're usingand to going back to Brad's
question is will it blur theline?
I think it's gonna bridge thatline between a business analyst
in your case and a developerwhere they have a better
understanding because in betweenthem would be an agent, right?

(22:33):
Where an AI would help kind ofhelp bridge that gap.
In addition to that, you cantake a functional design
document, like as you had said,take all that content and put it
in this functional designdocument where a business
analyst can vibe code, I guess,to get the concept put in place
and have AI help you build agood portion of that.

(22:54):
And then when you want to moveforward and say, hey, I think
this is gonna meet our results,that's where it's helped it
helps with the developer have abetter understanding of the
results that they're expecting.
And so their focus would be howcan I contribute to the result
that they're expecting?

SPEAKER_01 (23:09):
Yeah, it gives everybody context as to why
we're doing this.
Exactly.

SPEAKER_04 (23:13):
Yeah, yes, yeah.

SPEAKER_01 (23:14):
And so the Precisely, yeah.
Yeah, so so it one littleexample here to tighten this up
a little bit for this exampleI'm sharing with you guys is I
took all of that context and inthis example I used a chat GPT
account.
Um no, actually I used Gemini,the most re Gemini 3.0 uh
Redhawks account.
I put the project requirementsdocument.

(23:36):
It was less functional, it wasmore project specific, which
identified all of the types ofusers, features, functions that
they needed.
But I combined that with all thetranscripts from interviewing
the CEO.
And then instead of saying, giveme prompts I can use in my vibe
coding tool to create theprototype, I added a step in the
middle.

(23:56):
And the step in the middle was Iacknowledged that the tools that
we're using are automaticallygonna create a landing page
after you log in.
They just are.
They have to just if you don'tdefine it, it's just gonna make
something up.
The user logs in, this is whatwe're gonna show them.
Generally, it's gonna be somekind of metrics, whether they're
important to that user or not,but it gives you something to

(24:18):
work with.
Instead, what I did is I said uhtaking in context as to uh all
this information, all thesources provided, and
understanding the core businessobjectives, recommend a landing
page for the analysts once theylog in.
Jim and I came back and said,Wow, considering, and I'm not

(24:41):
kidding, it's like wow, youknow, it's very complimentary,
which cracks me up in thesetools, by the way, guys.

SPEAKER_02 (24:45):
It's like they're I told you you feel like you're
talking with a person.

SPEAKER_01 (24:47):
I feel like I'm talking to a person and they
really like me.
They're very encouraging andthey love all my ideas.
So it just makes me feel goodabout myself the whole time.
They motivate you.
Man, they motivate you for sure.

SPEAKER_02 (24:57):
But it I'm going to tell my after this, I'm going to
tell mine.
I don't want to interrupt yourstory.
I'm going to tell mine to talkto me a little less friendly.
I wanted it to yell at me.
I want to see what it does.
I'm actually going to put thatin the instruction file or in
the agent file.
I'll do it in one project.
I'll just put like in one of thecopilot instruction files for
the project.
And I say, you must be mean tome and you must call me bad

(25:19):
names.

SPEAKER_01 (25:19):
You're going to have to let us know what that's like,
man.

SPEAKER_02 (25:21):
I'm going to.
We'll listen to what it's likebecause that's what I told you.
I'm like, I'm like, thank you,and I'm doing this stuff too.
And I just wanted to say.
Oh, you suck.
Why are you being so stupid?

SPEAKER_04 (25:29):
You should have David Goggins voice that.

SPEAKER_01 (25:31):
Yes.
There you go.
I'm I'm going to use that, butI'm going to use a prompt that
says, uh, hold me accountableand don't and be rough.
So anyway, the outcome that Igot for that analyst landing
page, guys, it blew my mind.
And it was just arecommendation, right?
It wasn't the prompts I neededfor the uh for the proof of

(25:53):
concept, but it said,considering the CEO's business
objective of being able toincrease velocity in terms of
how many projects an analyst canhappen per week, you don't
really need a dashboard with KPImeasurements, but you need to
gamify the experience.
I would recommend a Kanban boardthat shows them the status for

(26:15):
each of their projects based ontheir flow.
And let's show them a velocitymeter so they know that they're
operating within the S the newservice level agreement they
have with the CEO.
And then we can even color codesome of those cards, and then we
can give shortcuts and call outprojects that are at risk of
violating that new SLA with theCEO.
So we're driving the rightbehavior.

(26:37):
So AI gave me thatrecommendation.

SPEAKER_02 (26:41):
Isn't that insane, though, if you think about it
one?
Yeah, you mentioneddocumentation.
I get far better documentationfrom the AI than any I've seen
from any developer.
That I'll come out and say likeacross the board.
And two, it comes up with ideasand suggestions that you don't
even think about.

(27:01):
Yeah.
And it executes them.
So it's it's it's baffling tome.
And I also sit here because I dotalk to some that, well, AI, I
can do this.
Oh, I got to do this.
It's gonna take me 160 hours todo this.
And I'm sitting there, I scratchmy head sometimes and said, just
use AI because it will even giveyou ideas that you didn't even
think of.

SPEAKER_04 (27:21):
Yes.
That's the time saver rightthere.
Where and can you imagine ifyou, you know, can you imagine
if you were to do that withoutAI right now, the amount of time
that you'd have to take andtranslate what they're trying
what they mean.
And then you're just at based onthe fact alone.
And you're not thinking ofoutside the box at that at that

(27:43):
moment, not yet, right?
Until you start speaking tosomeone, you may need to step
away and talk to others.
Here, you can have that withinan hour.
You can have a fullconversation.
Say, hey, have you thought aboutthis and that based on this
context?
You could be up and runningfairly quickly in the amount of
time.
So that's like your output hassignificant significantly

(28:03):
increased.
Yeah, I uh boatload.

SPEAKER_01 (28:06):
A boatload.
Yeah, I had uh probably sixhours of effort to get to the
and that includes interviewswith the CEO, um, some QA via
email threads, receivingstandard operating procedures
and documentation.
So I gathered all that contextall the way to the PRD.
So less than a full day for thework.

(28:27):
Okay?
You don't have to read the SOPs.
No.
I mean I got I got context and Ihave some QA around those SOPs
from two one-hour sessions inthe interview process.
Yeah.
But that was it.
And then um I think I'm roughly,you know, the team roughly 35
hours into the clickableprototype.

SPEAKER_02 (28:48):
And what would that have been before based on this?

SPEAKER_01 (28:52):
Oh, uh 120 hours.

SPEAKER_02 (28:54):
So that you I you got 25%.

SPEAKER_01 (28:57):
At least, guys, at least.
Um, what's fascinating to me aswell, this is something that I
think people need to be thinkingabout, is the vibe coding tools
that they choose for these highfidelity prototypes.
Think about those um that havemore full tech stacks behind
them.

(29:18):
So I'll give you an example.
Right now, the darling for usright now is is uh Firebase
Studio, which is part of theGoogle Gemini uh stack.
The reason I like that is Geminias a large language model is
already built into that techstack.
So let me give you an example.
And this is part of thatworkflow for this analyst.
They have to go out and do websearches to find providers that

(29:43):
match the detailed criteria thatwas collected from the client
during the intake process.
So using Gemini in combinationwith Firebase Studio and asking
for ideas and suggestions, it'slike, oh, well, why don't you
know let's have an agent builtinto this workflow at various
stages.
So the um it will automate theprototype we've demonstrated

(30:05):
successfully at this point, itwill automate the searches in
multiple languages, by the way,because they provide service to
global companies, to findproviders that are best match,
and then it scores best matchbased on client criteria to
shorten the cycle that it takesto select the providers that you
want to ask for proposals.

(30:26):
Then you can it'll draft youremails, making sure that you
don't inadvertently leave outany critical details when you're
asking for the proposal.
So it takes client intakeinformation, puts that in, and
then you can tweak each one ofthose drafts and then hit send.
Okay.
What's really crazy is theworkflow that we built out in
this prototype will also connectto Outlook.

(30:50):
And the agent is analyzing theresponses from the providers,
and you got a communicationinbox inside of this portal, and
it's using natural languageprocessing through Gemini
because it's part of the techstack.
And it's highlightingcommunications that it believes
contains a proposal response.
It's also highlighting thingsthat uh it believes are changes

(31:12):
in project scope because theclient's saying, responding via
an email thread saying, Oh, Ineed X, not real, it's this
number's wrong that needs tochange to this number.
So their in-client intakerequirements change.
So you're using Gemini.
You could use AI to ingest thatinto the next part of your
workflow.
It's like, man, when I showedthis to my client, guys, he was

(31:34):
like, that's my mind every day.

SPEAKER_04 (31:39):
You're using Gemini.
You but um so I've been usingNotebook LM, so I've Gemini as
well, but majority of my deepsearch uh recently I've been
using Notebook LM for a couplemonths now.
Yeah, and it's fantastic becauseit can create like uh uh cards
for you, right?
Like even like a PowerPoint, Iguess, cards.
Um and so is that what your isthat where you contain all your

(32:02):
stuff, or you is it just the theGemini chat?

SPEAKER_01 (32:06):
I'm using a sequence of different tools based on
where I am from sales all theway through delivery.
So I'm trying to figure outwhich tools work best.
So I will shift things and Iwill take context and move it
from one tool to another becauseI'm figuring out which ones give
me the best output.

SPEAKER_02 (32:23):
And that is the key.
Yeah.
This is see, this is one thingyou just hit upon, is is there's
not one tool.
It goes back with the concept ofwhat we're doing.
I listened to the Joe RoganJensen Wang um podcast
interview.
Uh not everyone listens to JoeRogan, they may not like him or
whatnot, but listen to thatinterview.

(32:46):
He went down this road kind of alittle bit about where AI and
using the tool, then everyone'stalking about where is my job
going to be.
Listen to that episode and hesums that up.
But that's the key right thereis knowing which model to use
for which type of output andwhen to use the tools.

(33:07):
See, this is sort of where Iwanted to touch upon as well,
too, because it's not that tool,just like a uh a carpenter
building a house.
I equate software to building ahouse because I think it's a
good analogy.
You don't always need a hammer,you don't always need a
screwdriver, but a builder isgoing to have a hammer, a
screwdriver, a paint roller.

(33:27):
You have all these tools, andthey know when to use the tool
because that's the mosteffective tool for it.
And that's where I think theemphasis will need to be versus
trying to know all the coding,trying to stay within this one
portion of it.
Because also, I think where youget you go back to what you had

(33:48):
from the prototype, we'regetting into the world where you
used to have software engineers,they only know what they know.
Yeah.
And they're going to developwithin that box of what they
know.
Unless they're really energeticand they go out exploring and
seeing how things can be doneand how people do things,
typically they're going to giveyou estimates on time based on

(34:08):
what they know, based on thetechnology that they know.
This right here to me opens upbecause uh Chris and I had a
conversation with a guest theother day.
I did some vibe coding.
It presents things to you youdon't even think of.
Yeah.
And you're like, how does itlike, how?
Where does this come from?
How does it come from?
So it's almost like you havepeers.

(34:28):
Yeah.
And then the different modelswill give you different
suggestions using this loosely,right?
Um I'm not bucketing stuff intoit like to try to be really
precise, but it's it's crazy,mind-numbing, jaw-dropping.
I don't even know how tosummarize it.
It's I I just don't even know.

(34:50):
Uh sometimes I think it'sreading my mind in some areas.

SPEAKER_01 (34:55):
Yeah.
It's so crazy, man.
Like we uh one of my principalengineers.
So I think a little tip for thefor the audience here is it is
particularly business leaders,you want to be successful with
figuring out which tools to usein in throughout your cycle
within the business.
You gotta be willing to carveout some investment, give people

(35:16):
permission to fail.
Maybe you need some skunk worksprojects, things like that.
Um, seriously, you know, becauseit's it's and you need to you
need to figure out what yourknowledge transfer and adoption
plan is gonna look like once youget those key learnings.
You know, we've been on a uhAI-first SDLC modeling effort
for the past probably four orfive months.

(35:38):
And something I haven't eventalked to you guys about.
Maybe we talked about it brieflyin our last uh last
conversation.
We just rolled out our owncustom ERP for professional
service organizations servingRedhawk first.
Wow.
It was a 90-day turn, 230 engineengineering hours.

(35:58):
Now, when you're a professionalservices organization like Red
Hawk, your resources in that ERPare your staff.
And the attributes of each ofthose resources are their skill
sets and uh uh in relationshipto the tech stacks that they
work with and how they score oneto ten in terms of level

(36:18):
expertise against those techstacks.
We already had all of our flowsin place and all of our data.
So let me just preface it thatway.
Our data was super clean andready to go.
Um, but uh now um our uh projectmanagers, our principals, the

(36:39):
C-suite, we're all using FlightDeck on a daily basis.
It's in production.
Wow.
We can accurately forecastrevenue out over 90 days,
generally because we're you knowthe PMs are responsible for
working with the the clients andunderstanding what's in the
roadmap, what's gonna happeneach sprint.

(37:00):
So we're forecasting that work,but then you know, resource
planning for us is making surethe right resources are
available at the right time withthe right skill set.
So attributes on our clientsoftware assets are the tech
stacks and the skills requiredin order to deliver on you know
providing sport maintenance andenhancement services.
Well, that all just allows us todo skill gap analysis.

(37:21):
Like I can tell you at any pointin time where there's risk and
opportunity in my business.
I know who I need to hire next,I know who's overloaded, um, and
we also have all of our timeentry data.
So we're managing contracts andall this stuff.
So this will blow your mind.
It blew my mind.
Uh three weeks ago, uh my CTOwho's functioning as a product

(37:44):
owner on this initiative, heknocks on my office door, he's
like, Matt, dude, I hope you'resitting down, I gotta show you
something.
So he brought up the developmentenvironment of our ERP, and he
said, navigate to the analysistab, and I did.
And he goes, Click on a hawk'seye view.
And if I click on it, and itgave me a couple of parameters

(38:06):
to select.
It's like select the date range,and I just selected the last 30
days, rolling 30.
And then I could select uh umaccount representative or
leaders within the organizationto kind of filter down the data
set if I wanted to, but I justleft it just tell me everything
about the business.
It generated an executive reportthat had different cards for

(38:29):
different types of information.
So it identifies where we'reperforming at a high level, it
identifies where there's risk,it makes strategic
recommendations about areas tofocus on based on client volume
and work and where maybecomments use the natural
language, right?
Um, where things maybe need tocourse correct.

(38:50):
Uh it's it blew me away.
And I asked Ron, I said, howlong did it take to create this?
And he said, well, it justoccurred to me that you're, you
know, you always need thishigh-level CEO level summary of
state of the business, right?
So I just went in to the toolthat we're using and I said,
Hey, uh we need a dashboard forthe CEO that kind of gives him

(39:13):
the state of the business.
He didn't define the cards andhow the information was going to
be analyzed.
You know, Jim and I was alreadybuilt into this platform, right?
And this is no exaggeration.
It took him five minutes, andthat included the time to think
of the prompt he was going touse.
That's wild.
That's like everyone's dream.

SPEAKER_02 (39:35):
I just I I laugh at that because uh I just laugh at
that.
I mean, you you hit some keypoints, and I laugh at it
because it's not that it'sfunny, it's mind blowing, and I
laugh at how many businesses arestruggling.
They think that they can't theythink that the individuals, they
think the teams can do so much Idon't want to say better in a
sense that having the people doit, but they take so much

(39:56):
longer.
And what I'm seeing in somecases.
Is this technology now allowsyou to adapt faster?
Because before it would be toyou told your CTO of their the
sponsor of the project, I needthis dashboard.
This is what I need in thedashboard.
Okay, we'll go back and developit.
It takes three months to get itdone, two months to get it done,

(40:17):
whatever it may take to getdone.
Yeah.
In three months, that may nolonger be relevant because the
world changes so fast.
And it and it's it's not justsoftware business, it's other
businesses too.
And I this I do want to touch onAI outside of software
development.
Uh, but the a few other thingswith software development I want
to touch on too.
But that's what like some ofthese per, and that's where I

(40:39):
kind of laugh is we people don'tfocus on how we can use these
tools to be able to do what youhad just mentioned.
Give me this Hawk's eye view.
I like that name, by the way.
I know where it comes from.
I picked that up, by the way.
Um how you can adapt to the howyou need to look at the
information based on how you seeit today, versus having to say,

(41:01):
I would like this dashboard, andthen you have to turn around and
wait.
And oh, by the way, there's beenan economic shift because
tariffs were implemented ortaxes had gone up or something
had gone on.
And now all of a sudden, becauseyou remember the tariffs, and
again, it's not a conversationfor for or against.
I can't tell you how many peopleI spoke with that said, okay,
now we need to factor in anddevelop something for tariffs
into our system.

(41:22):
Yeah.
Whereas if you have stuff likethis, for example, you may be
able to adapt that quickly.
Now, so that's where I kind oflaugh at uh I think Chris knows
why I'm laughing, but um withthis and you being able to put
this out there um to kind ofconsent this back a little bit.

(41:43):
Where do you see software going?
Because now is software becomingdisposable because now at that
point of I can sit here, case inpoint, I talked with somebody
this morning.
We had uh we're going to betalking or having a conversation
on something, and theyprototyped and whipped up
something.
By the time I finish saying whatI think we should do.

(42:08):
If we can generate software atthat scale, what happens to
software?

SPEAKER_01 (42:14):
Software becomes, you said it, disposable.
You may you may have a specificstrategic need that software can
help you with that you neverwould have invested in before
because of the time and effortrequired to build that software.
But now you can build it soquickly that it's okay.
We're gonna start to seedisposable applications that
maybe only have a 90-day shelflife.

SPEAKER_04 (42:36):
Maybe just in time kind of software that you need
for a certain period of time.
You perfectly put well put 90days, it could be 90 days, it
could be 30 days I just neededit for this month without the
massive investment that uh youknow the old school uh approach
would have been.

SPEAKER_01 (42:54):
Yeah, yeah.
So so for example, where I thinkopportunity for disposable
software might maybe come intoplay are fast growing companies
and on the acquisition track.
Yeah, startups.
Startups, you know, it it ifyou're um or companies like Red
Hawks, for example.
So uh, you know, part of ourgrowth strategy is acquisition,
we're gonna buy other softwareengineering firms.
So if I've got a platform thathas all of our way of doing

(43:18):
things, and then I can dump inall of their contracts, because
if I'm gonna buy a softwareengineering firm, which I've
done before, uh we want to we'rebuying their assets, we're
buying their book of business.
Okay, right?
So I've already got theindividual components.
So it wouldn't take much effortat all for us to build a
platform that supports theintegration of the company we

(43:38):
just acquired to reduce theintegration time from call it
six months to nine months to amonth.
If I could dump in all theclient contracts, right, and we
already have the agenticworkflow that actually, Brad,
you'll appreciate this, we cancreate documentation on existing
business applications.
We deploy an agent that doesthat for us, and that agent then

(43:59):
hands it off to another agentwho that identifies the bill of
materials and all the CDEs, andthen we get detailed action
reactionable reports on how toremediate that software
application.
So if I can take the contractsand drop them in, and I can
connect the repos for all thecustomers that they're serving
into that workflow, then on theother side of that, it's gonna

(44:21):
tell me strategically, I thinkit's like another version of a
Hawks I view where do our teamfocus their initial efforts and
to support that integrationpath.
To me, that's a disposableapplication in the sense that um
if I'm only gonna buy onecompany, I only need it one
time.
But I'm willing to make thatinvestment because I know my

(44:41):
team can build that probably ina you know pretty short order.
I've already got all the agentsand the the pieces built out.
Um so integration is a biginvestment that's part of an
acquisition strategy for anybusiness.
So I can see that as being areally uh good example of a
bespoke disposable application.
And maybe you repurpose it andspin it back up when you need

(45:03):
it.
Yeah.
Yeah.
I'm geeking out, guys, man.
This is too much fun for me.

SPEAKER_02 (45:11):
No, no, no.
I I I'm almost speechless in asense because I agree with you
and I see so many things, itbecomes now I was going to
purchase something the other dayto do a task for software, and
then I paused for a moment, andnow I'm like, it was it was
forty-eight dollars for a yearsubscription.

(45:31):
Oh, okay.
Yeah, which is not a lot ofmoney.
I'm just trying to say this islike where it's going, right?
It's it was$48, and I'm like,oh, maybe I can just vibe it.
Yeah.
And I did.
Yeah.
And it took four minutes.
Yeah.

SPEAKER_04 (45:47):
There you go.
So that's where I'm saying.
And then it met your needs, andthen okay.
It met my needs.

SPEAKER_02 (45:52):
And again, and I I will say, you know, as Chris and
I had in some previousconversations prior to this one,
I'm not doing stuff when I dothis that's being installed at a
customer or you know what I'msaying?
It's not what I mean, it'sproduction for my home, yeah,
for me, for what I need.
I'm not the haphazardly doingthis for an organization.

(46:13):
But again, it was that type ofthing is I needed something to
do something for a few minutes.
Yep.
And I had to sit back and go, isit$48 that I pay?
Or I want to vibe it.
Whereas before you would pay the$48 because it would probably
have taken me a day or two toreplicate what it was doing, or
maybe even a lot longer, becauseif I didn't even know how to do
it, probably could have taken memonths because I would have had

(46:34):
to learn the stack, I would havehad to learn the language, I
would have had to learn what itis.
But$48 is well worth it.
But now you're just like, I needsomething that does this and it
does that.

SPEAKER_04 (46:43):
Only a handful of times, and you're paying a full
year subscription when you onlyneed it for like two months.
Yeah.

SPEAKER_02 (46:49):
Yes.
I I need this was uh again, itwas a probably a one or two time
use function.

SPEAKER_03 (46:55):
Yeah.

SPEAKER_02 (46:56):
But that I needed.
So it goes to your point.
Like, did I really care aboutcertain things?
No.
Could it have been probablybetter?
Of course.
Uh, anytime you developsomething, it could always be
better.
But issue resolved, problemsolved, yeah, and minutes.

SPEAKER_04 (47:14):
And and and really quick too, that that that's uh a
perfect example of use of AI isthat uh people tend to forget
that acquisition and mergerusually takes a long time.
But if you have AI in between,it's going to expedite that
significantly because itrequires less physical person to

(47:36):
like do all the work.
There's a lot of paperworkaround that, right?
A lot of gathering of discoveryand stuff like that.
If you have agents that will dothat for you, it's going to save
you uh a whole lot of time.
And that's another thing that Iwas I was curious because I have
a family friend that's in the inlaw.
You know, how do you utilize uhcould you utilize agents for a

(47:57):
lot of the e-discovery kind ofthing?
And I'm sure they're startingthat way, but it's a lot of
these things are gonna expeditesignificantly.

SPEAKER_01 (48:06):
Yeah.
It's amazing.
You know, uh Brad, I think it'simportant that you and I like
how you articulated that some ofthe things that you're doing are
you're not rolling out intoclient or production
environments because there arestill a lot of cybersecurity
concerns around these tools aswell.
Absolutely.
You know, so a really criticalpart of our software development
lifecycle is yes, we're we'readopting AI first SDLC.

(48:28):
That just means that ourengineers who are experts are
leveraging AI first as opposedto deciding they're going to
write all the syntax.

SPEAKER_02 (48:35):
Correct.
And that's why I wanted topreface it.
That this in the context of myconversation, it wasn't for
somebody else that I was doingit.
It was for a task that I had athome where I had a controlled
environment where I knew I knewwhat I was doing in the sense of
whatever risks I was taking,which is extremely important,
which goes back into that.
You can vibe code anything, butyou have to be careful if you

(48:57):
don't understand what theresults are.

SPEAKER_01 (48:59):
We are seeing the tools advance pretty quickly.
One um, you know, I mentionedFirebase Studio and something we
experienced here just this week,is we have we have uh dev
staging production environments,backup routines, et cetera, and
this is for our proprietary ERP.
And our CTO is ready to releasea new feature, and Firebase
Studio said, uh-uh- uh-uh,because it detected a C V E in a

(49:22):
React library.
So it didn't even let him get itout into staging, which was
fascinating to me.
So um we didn't even know thatwas going to be a new feature.
There was no release schedule,there's no promotion around,
it's just all of a sudden thetool just all of a sudden it has
something new.
You know, which is it's the rateof these tools getting more and
more mature.

SPEAKER_02 (49:42):
It's yeah, I can't even think about it.

SPEAKER_04 (49:45):
I do like that you you brought up the the
cybersecurity component becausethat is it's grow it's going so
fast that people just want touse it and see what it can do
for them, but they do forget thethe uh the security aspect of
it.
I mean, there's been countlesstimes where I have conversations
with SB space, yeah, where theyhave I mean their employees are

(50:06):
using Chat GPT with theircompany data, and they don't
have any AI policy as a startingpoint.
I mean, even as simple as thatshould be a good foundation of
how to use AI within yourorganization.
And there's been like, oh, wehaven't done that yet, but we've
got Suzy in accounting askingquestions about the best

(50:28):
approach and then feeding dataabout their company.
I'm like, oh pretty dangerous,guys.

SPEAKER_01 (50:34):
It happens all the time.
You know, it surprises me howmany how few businesses have
completed acceptance usepolicies and training around
that.
And anybody who thinks thattheir employees are not using
AI, they're kidding themselves.
It's happening at every level inthe business.
So you gotta get a handle onthat.

SPEAKER_02 (50:55):
What are you telling me that people aren't people
aren't using AI?
Like you can you can stop them,or they're not using it, or
everyone's just unaware?

SPEAKER_01 (51:02):
Give them the secure pathway.
They're gonna use it whether youdo so or not.
So expedite that process.
That's you have internet access.

SPEAKER_02 (51:09):
Well, that's still it is.
It's it's Chris brings up thegood point with the AI use
policy, and you also bring upthe point of uh you know getting
that in motion now because Iwould rather have my employees
and team members know how to usethe tool, understand how to use
the tool instead of just seeingwhat they hear, like

(51:31):
conversations with ours saying,oh, we could create an
application for you in minutes.
Yeah.
Granted, I mean, we all workwithin the industry with we're
leaving out some of the aspectsof what we know about software
development and about technologyand about business.
But it does sound if you hearsome of these examples that
people put on there, and if I,you know, for those that may use
social media, you have theseshorts that are like, oh wow, I
did this in 30 seconds.

(51:53):
It's it is important tounderstand because, like you had
mentioned, they're going to useit, even if they copy and paste
it on their own somewhere else,not in your work environment,
they can do so.
You can't stop people fromdoing, or even if they had to
type it.
Oh, and by the way, did you knowyou could take a picture and
drop that into AI and we'll tellyou exactly what that screen

(52:14):
says?
Oh, that's what I'm saying.
Like, there's so many, there'sso many ways that people can get
information.
The only way you can stop themfrom getting information is to
not have information.

SPEAKER_01 (52:24):
That's right.

SPEAKER_02 (52:25):
I mean, and that that's what it comes down to, is
it's if you want anybody, andthen it's also you can be more
restrictive.
And don't get me wrong, securityis extremely important, of
course.
But there's a point where youyou have to balance security and
productivity because if you wantto be, and that security has to
match the level of theinformation that you have, and

(52:46):
then also how do you hinderproductivity?
Because I see sometimes peoplego the other way.
You get a little too securetrying to satisfy every edge
case where you have tounderstand what is the risk of
that edge case and am I willingto accept it?
That's the key.
Identify the risks, identify, amI willing to accept that risk?

(53:08):
Right?
That's that's a point that somepeople leave out.
So I just want to be clear aboutit.
We get I get a lot of feedbackon some of these things when I
say stuff.
So I have to be very careful tosay, I'm not saying you should
be haphazard with security, I'msaying be secure, but also
understand is the effort of whatyou're trying to do worth the
productivity that it maysuppress.

(53:30):
Right, yeah.
And are you willing to acceptthe risk of not doing something?
But like you just said, uh, whenit comes to the use of AI,
people will use it.
People are using it.
Uh people aren't in a nobody'sliving in a vacuum in 2025.

SPEAKER_04 (53:46):
Even Googling right now has an AI mode.
Yeah.
It just summarizes or even getyou in AI mode instead of having
a conversation with it, doingyour research.

SPEAKER_02 (53:56):
Everyone's using it.
That's interesting.
Yeah.
Um but AI is productive outsideof software development as well.
And because I have uh talkedwith others that they uh it's
it's amazing what people do withthe AI ecosystem, I call it.
I I've seen people have it gothrough all of their emails and

(54:17):
generate draft responses, andthey just review the draft
responses.
It's almost like you don't haveto do anything anymore.
Who types?
I'm just getting to the point,like who types?
Because now you can use voice.

SPEAKER_01 (54:30):
I got two great examples, and it's not even
related to software engineering.
Um, one is, and and Chris,you're probably doing something
similar since you mentionedNotebook LM.
Yeah.
Uh for sales for sales trainingin our organization.
I created a notebook and droppedin all of the sources I needed,
including talk tracks, serviceofferings, capabilities deck,

(54:50):
you name it.
And you we use that notebook forsales training.
And then we have a newopportunity in the pipeline.
My flow looks like this.
So um in fact, I've got afirst-time appointment with the
with a group this afternoon.
And before I meet with them, I'mgoing to use Chat GPT in a

(55:11):
thinking or deep research mode,drop in the URL address to their
business.
I'm going to drop in a suite ofartifacts from Red Hawk's
business to drop that in.
And the email thread thatincludes all the introduction
and other artifacts thatexplains what they're interested
in.
And it will give me uhrecommendations.
It's a sales dossier that Icreate in ChatGPT.

(55:33):
Then I take that sales dossierfrom ChatGPT, which by the way I
include LinkedIn profile PDFs ofthe key people that I'm going to
be speaking with that may I havemay have not even met yet.
Right.
I drop all that into Notebook LMand I create a sales-specific
opportunity notebook.
So when I close this deal, whichI hope I will close this deal,

(55:54):
then the process for my team,instead of a written project
brief, they go have anexperience with that notebook.

SPEAKER_00 (56:03):
Yeah.

SPEAKER_01 (56:04):
You know what I mean?
They can put it in, they can putit in uh uh run the audio
feature and they've got apodcast that they can interface
with and interrupt and askquestions.

SPEAKER_03 (56:13):
Yes.

SPEAKER_01 (56:14):
They come into that next technical discovery
meeting, dialed in, and this hasnothing to do with engineering.
This is off the shoreoff-the-shelf tools in the right
order of operation.
I mean, these sales dossiersguys are like it is crazy.

SPEAKER_04 (56:30):
That is crazy.
You can even do mind map, likenotebook ln.
They've added that.
Infographic, you'll create allthat.
Pete Pete uh PowerPoint quizzes.
You can do all the things youneed.

SPEAKER_02 (56:41):
I love mind maps, by the way.
I use X Mind.
I use X Mind.

SPEAKER_04 (56:45):
Maybe I have to switch over to the Notebook LN,
man.

SPEAKER_02 (56:47):
You'll have to show me it's stupid.
You will have to show me whatyou're doing because maybe I
will move over to that.
It's it's it's tough.
There's so many tools, it'stough for me to keep up.
I've been spoken I focus most ofmy time more on the development
portion of it.
Yeah.
Uh Notebook LM sounds like it'smore content management
simplified.

SPEAKER_01 (57:07):
So I have to you're control controlling your sources
and it's giving you insights onit.
So I'll give you a quickexample.
So I've got uh another uh dealin the pipeline right now, and
it was um uh an intern whorecently got hired full-time in
supply chain management, likethat's his area, and they need
an inventory, like a lower um uhbut custom inventory management

(57:32):
uh solution for their shippingyard.
Okay.
It was intelligent enough for itto identify based on the
LinkedIn profiles, backgroundinformation, it has to be good
data, right?
So his profile is prettycurrent, along with uh the
president of the company whojoined the meeting and all the
artifacts and my sales dossier.
It said this gentleman's careerpath is X, Y, and Z.

(57:53):
And based on what they're askingyou to do, this could be the
flagship project that helpslaunch his career.
So it's it's it's it's giving methat level of insight, like
that's how important thisprobably is to that individual
based on their careertrajectory.

SPEAKER_02 (58:10):
I'm like, Chris, we have to we have to schedule a
demo.

SPEAKER_01 (58:14):
Yeah, reach out to you guys.
I'll take you through a coupleof these things.

SPEAKER_04 (58:17):
All I'm doing is I do want to know, yeah, we'd love
I'd love to see how you useNotebook LM.
I I know you know, we work atthe at least for me, we work in
the Microsoft ecosystem.
I do use other products outsideof the co-pilot world.
Yeah, I I I am very impressedwith the Gemini and Notebook
LM's capabilities, even from apersonal perspective of just

(58:40):
putting all the things that's inmy mind into Notebook LM and
then have it deep search andthen reference to the things,
and then you say, Hey, can youcreate a video for this that's
pros and cons?
And it'll like tell me like whatare the pros and cons.
And even Audi, like you said,you know, little podcasts that
will have a conversation.
Oh my gosh.
And mind map, Brad.
Like, yeah, I need you to putthis in a in a in a mind map as

(59:03):
well.
Yeah.

SPEAKER_02 (59:06):
I have to see this notebook LM because um I'm just
thinking now of a lot of thingsbased on what you're telling me.
That the world is changing.
It's too fast.
It's almost um I don't know.
It's it's too fast.

SPEAKER_01 (59:24):
It's fun, man.
I tell you what, I was freakingout at at you know at one point,
like what's gonna happen.
The industry's gonna, you know,be heavily impacted and we're
gonna lose business.
But it's been the it's been theopposite.
It's been the opposite.
We're seeing more and moreopportunity in the pipeline is
getting uh it's just full ofopportunity.
Because what's happened for usis in serving that mid market,

(59:48):
the number one reason forcustomers not making an
investment was time and effortand the cost related to doing
something to pursue theirvision.
We couldn't find what theyneeded off the shelf, that's
when we would be brought into.
To the conversation, right?
So we still think that buybersis build arguments make sense,
right?
You know, go go with theMicrosoft stack, use uh ERP, CRM

(01:00:09):
platforms available that areavailable off the shelf, because
they're going to build in AIagents and tools to help you
along the way.
Custom's not always the rightdecision.
However, when you can't findwhat you need off the shelf,
they would typically turn to asoftware consultancy like Red
Hawk.
But by the time we get throughum the discovery phase when we
put something together in atraditional SDLC, sometimes the
RI ROI wasn't there.

(01:00:32):
The problem was it expensive orbig enough uh of a problem for
them to make the investment tosolve the problem.
Well now it's the conversation'sso different because these
bespoke software solutions canbe developed in such a
compressed timeline andinvestment that I'm hyper
optimistic.
Like, I mean, some of the thingswe're building for folks right

(01:00:54):
now, like R E R P and you know,the other example I share with
you guys, like I'm pumped.

SPEAKER_02 (01:00:59):
Yes, I think it's it's um it's it's just a repeat
of my thought, and I'm stuck onthis thought now, is something
that you touched upon earlier.
It's all about the data.
Yes.
I think I think the you andChris were talking about
Notebook LM and how you're usingit for data.
You're talking about how youcrafted an ERP system for

(01:01:19):
yourself that you're using uhbecause you had the data.
I think it all comes down tohaving a core data system that
you interface with these toolswith.
So you think of the concept ofMCP, which is supposed to be a
standard interface for you to beable to communicate with
external systems or separatesystems or however you would

(01:01:40):
like to look at it.
It's all going to be a matter ofhaving the data available
because without the data youcan't really do much.
I mean, you can create some coollittle applications for stuff,
but if you're looking for abusiness point of view or even a
personal point of view, you needto have the data available for
consumption.

SPEAKER_01 (01:02:00):
Agreed.
Quality of data and adherence tostandard operating procedures
will determine your level ofsuccess.
100%.

SPEAKER_02 (01:02:11):
Wow.
Well you know, my I think we'llhave to um I I'm at a loss of
words to be honest with you withthis.
This is just um it's uhstimulating, I guess you can
say.
But I do um I do appreciatetaking appreciate you taking the
time to speak with us again onthis.

(01:02:31):
I think we will do the samething.
Let's schedule a follow-up inanother quarter.

SPEAKER_03 (01:02:34):
Yeah.

SPEAKER_02 (01:02:34):
Uh, and we'll have some notes and we'll have to
just start thinking of what hashappened since December 12th,
2025, and the next day that weschedule it.
So all of us will have to keep aa log of things that we have
witnessed with the advances ofAI and uh things that came out,
and then we'll have to talkabout them and we'll see where

(01:02:56):
we are then.
Because will MCP exist?
Will agents exist?
I'm sure they will exist, excuseme.
Will they be the topic of thetime?
Interesting.
Uh, when I say exist, so we'llhave to see.
Um, but again, thank you verymuch for taking the time to
speak with us again thisafternoon.
We really appreciate it.
Thank you for blowing my mind.
I'm sort of speechless.

(01:03:17):
Uh, but if anyone would like tocontact you to learn a little
bit more about AI, learn aboutRed Hawk and some of the great
things that you're doing, what'sthe best way to get in contact
with you?

SPEAKER_01 (01:03:24):
You can look me up on LinkedIn, Matt Stripple Hoff
on LinkedIn, and uh reach outwith a connection request,
reference this podcast.
That'll be my cue to go aheadand accept that invite.
Um, or uh just check us out atredhawk-tech.com.
You can fill out a form there toreach out and schedule time with
me as well.

unknown (01:03:40):
Yeah.

SPEAKER_02 (01:03:41):
Great.
Thank you very much.
Uh I look forward to speakingwith you again in about three
months.
I'll talk with you.

SPEAKER_01 (01:03:46):
Always love talking to you guys, man.
So love it.
Thanks for having me back.
Fantastic.
Thank you.
Thank you, man.
All right.
See you.
Ciao.

SPEAKER_02 (01:03:53):
Bye.
Thank you, Chris, for your timefor another episode of In the
Dynamics Corner Chair.
And thank you to our guests forparticipating.

SPEAKER_04 (01:04:01):
Thank you, Brad, for your time.
It is a wonderful episode ofDynamics Corner Chair.
I would also like to thank ourguests for join joining us.
Thank you for all of ourlisteners tuning in as well.
You can find Brad atdeveloperlife.com.
That is D V L P R L I F E dotcom.

(01:04:22):
And you can interact with themvia Twitter, D V L P R L I F E.
You can also find me atmantalino.io, m-a-t-a-l-in-o.io.
And my Twitter handle isMatalino16.
And see you can see those linksdown below in the show notes.

(01:04:44):
Again, thank you everyone.
Thank you and take care.

All Episodes

Episode Transcript

Popular Podcasts

Stuff You Should Know

SmartLess

My Favorite Murder with Karen Kilgariff and Georgia Hardstark

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Episode 440: The Future of Development: Is Software Disposable?

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}Stuff You Should Know

SmartLess

My Favorite Murder with Karen Kilgariff and Georgia Hardstark

All Episodes

Episode 440: The Future of Development: Is Software Disposable?

Stuff You Should Know