Watts in Your Data

Denis: Hi, and welcome to a new
episode of the Industrial Data Podcast.

This is the third episode where I
will talk about the evolution of

OT data integration and have with
me on this episode Lonnie Bowling.

Lonnie welcome to the show.

Lonnie: Hi.

Thanks Dennis.

Thanks for having me, man.

Denis: Yeah, it's great to have you here.

I remember we had a couple of talks,
a few weeks ago, which inspired

me, in fact, to do this podcast.

So perhaps the people that don't
know who you are, could you say

something about what you do?

Lonnie: Sure.

Yeah, so my name's Lonnie Bowling.

My company is called Dimus, D-I-E-M-U-S.

And it's just a one person company.

Me and I do my, my main activities
these days are consulting for I.

Companies that are trying to leverage
their operational data into other

systems mostly around renewables
like wind farms and things like that,

and also just utilities in general.

I've been around a long time I've been
in this space before it even existed.

I started out as an engineer
in a manufacturing plant.

And even way back then, I realized
the value of data and, man we

could be using this manufacturing
information that is on the line.

If we could get our hands on it and
figure out how to collect it and

use it, that'd be pretty awesome.

So that was even, that was in
the nineties, kind of dating

myself here a little bit.

But anyway going all the way from
my early days in manufacturing.

Eventually I started working for system
integrators who did a lot of integration

around control systems, working on control
systems, you know, your typical PLC or

you know, SCADA systems, things like that.

And did a lot of that for a while
and still this data thing was always

kind of in the back of my mind.

And then, I guess it was around 2010
is when I discovered OSI Soft PI.

It was involved in a couple of projects
I was working on and and I was like,

wow, this is this is the product that I
have never seen that solves a lot of the

problems I've always been thinking about.

And so anyway, I became heavily,
involved in PI projects building

up a, an actual team around doing
integration around OSIsoft PI

systems and speaking at conferences.

And I've been doing that
for, I guess 15 years now.

And the landscape's changed quite a bit.

Since since those days, even just in
the last 15 years, things have changed

quite a bit, which I think we might
wanna talk about on this podcast.

We can talk about kind of like how it
started out and then where it is today.

'cause it's a lot different very
interesting story in my opinion.

But anyway, that's that's kind
of the Lonnie in a nutshell.

Denis: That's actually a good segue into
the topic of this podcast, which would be.

How did OT data integration evolve over
time since the nineties to let's say

2025 today and the role of o OSIsoft PI?

And I think you mentioned
there, wind and renewables.

I think

Be a good industry to focus as I
have also quite a bit of experience

working on SCADA data in wind parks.

So perhaps we can focus
the discussion at that.

And in that regard, the first
question you mentioned big changes in

digital infrastructure integration.

What are they?

Lonnie: Yeah.

I think we could probably break it
down into a few different periods.

I've never really like tried to
like segment it out like this

before, but in my mind there
was kind of like the pre-data I.

Era, which is important to understand
because these systems are still around

and they haven't changed a whole lot from
even back in the eighties and nineties.

So during that time you know, there's up
to early two thousands, there was a lot

of industrial there's a lot of automation
going on in industrial, systems, you

know, like manufacturing, utilities.

This is before we can't really talk about
wind and solar at this point because

they really weren't much of a thing.

The internet had just been invented.

So it just gives you an idea
of like where all this was.

Yeah.

I remember installing PLCs and a
SCADA system in the manufacturing

plant and networking it together.

That was like, the network
was just a new thing.

We just put it in and we're
connecting all this stuff together.

I go, wow, this is amazing.

And then the internet.

But, so that's I think a little bit
different than maybe where we are today

where you just think about, oh, you have
all these devices out there and they're

producing data and we wanna use this data.

It's, in my opinion, it's a little
bit of history of understanding, you

know, where how this kind of evolved.

So picture everybody's kind of.

Really concerned about automation
and putting in different types of

controllers and things like that.

They're not really thinking
about data at all at this point.

They're just really thinking about,
okay, how can we improve our processes?

Processes?

How can we improve our quality?

How can we reduce the amount of
labor and all that kind of stuff.

And so this was kind of like a big it
was a really big thing in in this space,

in our space, in the industrial space.

So that that, that started happening.

Everybody started doing it.

All the new systems going in had these
kind, these kinds of things happening.

And so that was, you know,
that went on for a while.

And then I think it was probably
like, I would say like the early

two thousands, like around 2005.

Or so maybe a little bit
earlier, but not really much

discussion up until that point.

But people started going so we
have all this information, all this

stuff being connected together on
networks and can access it remotely

and over the internet and so forth.

And internet really wasn't like much of
a, of a consideration at this point, but

people started realizing, oh, well we
could, we have a lot of information here.

You know, that we could potentially
use for other things other than,

you know, just a controller setting
there running a machine of some sort.

So that's when the first ideas
around data and getting data out

of these systems kind of happened.

I think there were some places
where it was happening maybe,

but that was like when I really,
you know, started noticing it.

And and it's funny because the
way that it kind of happened was.

Kind of in a weird way, actually, I, my
experience was that it was around SCADA

systems and in SCADA systems, I remember
it was a water wastewater treatment plant.

And they were they could
control the whole plant, right?

They could run it and they could not
turn the pumps on and off and pump

valves and do all that cool stuff.

Check levels and temperatures and
chemistry and all that good stuff.

But one of the things they didn't
have is they didn't have any trending.

So they never knew like what was
happening and what happened in the past.

So part of these SCADA systems
started offering historians.

And so a historian, you know, you know
what a historian is, I know what it

is, but you know, just for the audience
sake, in case somebody out there doesn't

know what it is, you know, it is just
literally collecting time series data.

So just values and data points.

So these historians, SCADA
systems started adding historians

as a feature to the product.

Then you can configure
those to store data points.

So this is when and this is kind
of how the whole thing around

tags kind of happened and why
tags are part of our world.

So a tag is a big problem for us, but
we could talk more about that later on.

But these tags, they have these names,
they have these weird names in 'em,

and you have to like organize 'em all.

You gotta know what it's for,
but it represents some value

out there in a sensor, right?

So if it's like a temperature
sensor, you'd have a tag.

It's on piece of equipment that
equipment ID might be embedded in

the tag name and stuff like that.

So all these tag names were in
the PLC code and in, you know,

these, in these these systems.

And so when you wanted to collect data
in the historian, you just told it what

tags you wanted, and then the historian
would go and look at it 'cause it

already had a connection to the system.

And then it would start collecting data.

So literally the whole system
was tag names and data.

And so you really had no other
information other than that.

So when you were in a SCADA system and you
wanted to look at something, you clicked

on it and you would see a trend pop up.

So that was like the first.

You know, kind of like use of data
in a very, you know, basic way that

still is to use today quite a bit.

You know, it's obviously a very
useful thing to understand, like what

just happened soon, but it's still
all around operations at this point.

So, I'll pause for a second, see
if you have any thoughts or if you

have any way you wanna, if you have
any questions or ideas on this.

Denis: Well, I found it interesting
that you mentioned that people were

not thinking about data back then.

I think that's a big change
I see happening today, namely

companies realizing that.

Data is in fact a valuable byproduct
of the production process that

should, in my opinion, at least be
treated with the same care as the

actual product they're producing.

'cause that data can lead to
improvements in operations.

Would you say in your experience that
we focus on the utilities industry, that

message has, landed, do those companies
value data as much as they should

or is there still a long way to go?

Lonnie: Well, yeah, I think people.

Most companies now are understand
there's value around the data.

Like unlike the 2005 point where
people were just there was a lot of

discussions about it and kind of like
conversations about it that people

weren't necessarily too convinced that
this would be a smart place to spend

money, you know, because it's like, yeah,
okay, a concept that might be fine, but

how, you know, when you start asking
me to write checks, you know, I better

get some kind of real results outta
this other than just saying, I did it.

I didn't get anything outta it.

So we're 20 years later now
and you know, so at least I.

The conversation around the
data is valuable isn't much.

It's not a, it's not a big
discussion at this point.

Mostly discussion is centered around
like, okay, it's valuable, but in

what way specifically, you know,
and then how do we accomplish that?

You know?

So those are like the things that I think
are a little slightly different than.

Then around you know, just like, okay,
hey I'm gonna convince you that you

have a lot of valuable data here that
you could probably use for something.

You know, that was like, once you
kind of pass that point, then but

we're still like, in my opinion,
it's kind of startling, right?

Thinking about it.

I mean, 20 years later, we're not
like, we're not like that much

further down the road with this.

We're still struggling.

There's a lot of struggle here, you
know, with, how do you really like,

get in there and do something amazing
with this data and these systems?

And how do you put 'em together in a way
that is gonna make it easier to use and

people are gonna be able to access it.

And so there's a lot of, there's a lot you
know, interesting challenges around it.

And things have been.

Progressing, the new technology has been
progressing, so we're still, there's

still, in my opinion, a pretty big delta
between kind of results and where we

are in the in our technology and what
our technology can promise versus what

we're getting, what's being delivered.

Denis: So let's focus on
the problems you mentioned.

At this point, we have the
various stacks Time series data

is being fed the historian.

We technically have the data and we
manage to make plots, and if question

would be, so what's the problem?

We have the data in our historians, right?

Lonnie: Yeah, exactly.

I think that's a good point where,
so, so a lot of the early problems,

I think were and maybe still are
to an extent, but there's a lot of.

There's a lot of messiness
that goes on in these systems.

You know, like I was talking
about earlier, thinking about

the history of how they unfolded.

You know, nobody was really structuring
these things to be easy to collect

data out of and things like that.

So, and when you go in and you you
put in new pieces of technology

or software or whatever it is,
and you connect to these systems.

They're they're not gonna
rewrite 'em for you.

You know, it's gonna, you're gonna deal,
you're gonna have to deal with what it is.

So there's a lot of
different types of protocols.

There's a lot of you know, the
tag naming all over the map.

Everybody still does it
differently to this day.

And I honestly don't believe that'll
ever change because there's standards

out there, but, they're really confusing.

Even if you try to use it, most people
don't implement it the same way each time.

So, so you'll have a, so, so you
have a just kind of lot of, like,

on the ground level, you just have
a lot of variables to deal with.

And when you get into like, things
like what we were talking about with

renewables, like wind and solar,
things like that, distributed systems

it becomes another layer of challenge
on top of that because now you're.

You know, you're not like in a place
where like, let's say a data center

where you just have a huge amount
of equipment sitting in one building

and it's all there and you don't
have to really worry about losing

connection to it and things like that.

But you know, when you get into these
remote systems that are distributed.

You know, connectivity becomes a
thing you have to really think about.

Like, okay, what happens
when you lose connection?

What's going on with that data?

You know, how do you, when you reconnect,
how do you get that data synced up back

into your main database and, you know,
and then also like, bandwidth, right?

You may not be able to just stream all
that data real time into your system, so

maybe you don't want to stream it all.

Maybe you want to do some kind of
edge type of thing where you only

send the important information.

And things like that.

So there's all kinds of weird and
interesting complex things that

happen where at first it seems going
back to our SCADA example with the

trend, it seems like so basic, right?

But now in two, in 2025, it's like
the, these systems are massive.

The amount of information that's
in these systems is massive and.

Then you're like, okay, well let's
collect all this data and let's put

it somewhere and do something with it.

You know, there's a lot of questions
around that, some of the things I

just mentioned, but also like, exactly
what data are you gonna collect?

Are you gonna collect it all?

Are you gonna collect some of it?

Like, what happens if you
don't collect the right ones?

After you make that decision, you
realize later on, oh shoot, I should

have been collecting this other stuff.

And so, you know, so there's all
kinds of all kinds of challenges

around this that make it.

Yeah, it just makes it, it
can make it, it make, it can

make it tough to get it right.

And I don't know if we've actually figured
out all the ways to solve these problems

yet, but you know, that's why you're here.

Denis: Yeah, you make a good point,
especially about the first issue

you mentioned about it's not easy
to get data out of your historians.

Well, let's assume there is an
external system, whether it's the

cloud or a Delta Lake or anything
else we want to get the data into.

a common problem I often encountered
in my work is proprietary formats.

Historians are made to store
data, let's say by given vendor,

Vendor, of course, is not very happy
of that data flowing out in open source

format 'cause now others can use and
makes this Historian, yes less valuable.

So I often encounter modules
you have to purchase in order

just to get the raw data.

Have you

seen anything of that?

Lonnie: Yeah, definitely.

So thinking about like how things evolved,
like after 2005 when we started, okay.

Historians became a thing and I told
you like, okay, historians were part

of a feature of a SCADA system, right?

And SCADA systems and PLC and
other control systems, they were

all like, you vendor specific.

So ge, Siemens Emerson these, you know,
all of 'em had their own versions of these

and and they all were, you know, and they,
and like you said, they really weren't,

they kinda like, were happy if you're
using their platform and all their tools,

but they really like, okay, well we don't
wanna make this like so open that anybody.

We connect to it with other tools
from other platforms, even though that

didn't really exist in the beginning,
but it became more of a thing later on.

Initially they were just putting these
in and they weren't really thinking

about like, okay, well we wanna make
it open so people could connect to it.

They were just mostly trying to,
you know, embed it within their

products, within their other products.

But, you know, and then O OSIsoft
was a little different in that they

specialized in just a historian.

So I think that's a.

Was an important thing that happened
because they were like one of the first

companies in our space to be okay.

We're not just embedding a
feature in an existing product.

We're actually creating a whole
new product that then can connect

to all these other systems.

And so that was, although it was still
proprietary and it still is today.

It was kind of a little bit of a step in
this open kind of world where, okay, now

you had a system that you could connect
to anything, and they really worked hard

in making sure it could connect to all
these different sc and control systems.

And and they created all these connectors
and so all this data could flow in.

And then you had this one
historian now, and you could

deploy it and you could connect it.

To all these other systems and that's
kind of where they really like, I

think they were unique in that way
and that's kind of why they got so

much traction in certain industries.

You know, like the utilities,
they had like a 90% market share

at one point, I'm pretty sure.

But back to your point is that now, okay,
even like with, if you wanna get data

out of a PI system, it's proprietary.

So you have to do it their way.

You know, they have their own
API, they have their own SDK, you

have to pay licensing for that.

And they, and as they develop
new systems, which they they

now been purchased by AVEVA.

So, and that's part of Schneider,
subsidiary of Schneider.

And that happened like
about three years ago.

So now it's AVEVA and AVEVA.

Owns a bunch of sc, they own
a Wonderware SCADA system.

It's all been rebranded.

They bought, they pur purchased a bunch
of stuff and they packaged it all up.

All that integration is happening, and
it's also, once again, like a platform.

So PI now is not like, it's still pi,
it still can connect to everything,

but it's, but it kind of has it's,
and it is kind of like static now.

It's not really being expanded now.

New systems are being put in
new systems are being developed

by AVEVA and other companies.

That are what I would say the next
generation of what a historian

is gonna look like versus the
pie and the other historians that

were built or in the early days.

So now we're in this new period
really that probably started like

20 18, 20, yeah, around 2018.

It's my guess.

So the last seven years now that the
technology around the historian and how

it works and what it's capable of and what
can connect to it has changed drastically.

With other historians being developed
and other systems being developed

that are based on modern technologies,
you know, like cloud-based and can

be you know, deployed to the edge and
all this other interesting things.

And Databricks has come up which I
know you're doing Databricks stuff

and I personally don't know a lot
about Databricks at this point,

but it's like a new hot topic now.

Everybody's very interested in Databricks
around OT data and these systems.

So we'll see where that goes.

Denis: Well, let's get to the
cloud and Databricks in a moment.

I first want to focus a bit more
on the actual role of a historian.

The way you described it, you had
historians developed before you had

this whole push for data integration,
and then OSI Soft PI came in and

helped connect the various systems
assume you meant SCADA PLCs, but

also even the MES and ERP layers.

Would that be correct?

Lonnie: Yeah.

not, it's kind of a nightmare.

And that whenever I get into the
M-E-S-E-R-P part of the world,

it's like they're really expensive
and, and companies that put 'em in.

It's usually a very long, big
projects and historians are part

of a an MES or ERP Pro project, but
because they have to have someplace

to put that kind of time series data.

But there's also a lot of back
and forth, the transferring data

back down to control level layer
and backup and things like that.

Still I think to this day kind
of SEP is a little bit separate.

Denis: Great.

So we are in fact talking
about time series.

Lonnie: with MES and stuff like that.

It's part of this time series,
but it's also a lot of like, you

know, the, essentially the big
idea behind M-E-S-E-R-P stuff is

is tying your manufacturing and
your production into your business.

So you're merging those two.

So when you get an order that
can go down to the plant floor,

let's say you're making chocolate.

I say that 'cause I actually worked
at a chocolate factory one time.

I was like.

One of the coolest jobs I ever had.

Anyway, they give us free samples too.

Like, and a sample was like a
15, a 50 pound bag of chocolate.

Denis: That's nice.

Lonnie: Yeah.

Yeah.

But anyway you know, like you wanna send
your chocolate recipe for the specific

order down to the to the equipment.

And the equipment has that
recipe and res it and then, you

know, a batch process like that.

And then it's done and all that
gets recorded into the system.

And, you know, you have
traceability if you.

Something happens and you get a
customer complaint, you can later

back, go back and look at it.

So that's a lot of like the M-E-S-E-R-P
stuff, which is a little different

than the OT data that we're talking
about in a lot of the industrial

use cases are centered around maybe
downtime, minimizing downtime, or

improving performance optimization.

You know, there's a lot of interest
in machine learning, right?

And where you wanna feed all that
data into these models and you wanna

be able to do abnormality detection.

And so there's a lot of
those kinds of use cases.

I mean, that's not the only
thing, but it usually is centered

around kind of trying to improve
operations in some way or another.

That seems to be like a big, the big thing
that everybody like if you go look at.

The conferences, that's like what
90% of the talks are gonna be about.

Denis: I fully agree.

I've attended a couple of aluminum
conferences in primary production,

and even there field that is very
traditional, very on methodology and

even there, you see year after year, the
amount of talks about AI and aml just

exploding is definitely a living topic.

In that sense, let's focus.

You mentioned here, indeed these new
applications that we want to do with data.

We don't just want to look at trends,
which means we have to get more

compute if we talk about ML and ai.

In my last podcast with Thomas from Times
here, we also established the fact that

many utility companies are now focused
on moving the data from the shop floor.

that's OSS of buy or story directly
into the cloud to analyze, for

example, with Databricks, do we also
observe this movement to the cloud?

Lonnie: Yeah, I mean there's definitely
been a, you know, I think in 20 15, 20

14, as when like cloud computing kind of
came into the scene in our world and and

there was a big debate around on-premise
versus in the cloud and things like that.

And your data's in the cloud.

so those discussions are kind of
mostly over at this point where most

companies are, interested in moving.

Some aspect of this
infrastructure into the cloud.

And it's not that, there's still
on-premise stuff and there's

still a reason to have it.

You just don't like put
everything in the cloud.

So, there's a lot of different
architectures now around how do

you implement cloud solutions in an
infrastructure kind of way, right?

But generally speaking, why do
companies want their data in the cloud?

You know, what is the, what's the
reasoning or the thought behind that?

Right.

And that's an interesting question
because sometimes they don't really

know they just somehow just think,
okay, I, I'm gonna put it in the cloud.

I have ideas around why they do that.

Everybody was thinking, well, let's
put all our data in, like a data lake

and, then we'll have it all there.

We can do these massive queries on it and
we can discover things about our operation

that nobody ever would have an idea.

And that really didn't get too far.

Don't really remember anybody
truly being successful with

just that general approach.

And I think that's where Databricks
is solving part of this problem in

that being able to, like I said,
I'm not a Databricks expert, my

understanding is it kind of gives you
gives you ability to structure and

query your data in a more friendly way.

And maybe it's performant.

I don't know, but I know like there's
one of the things I could say is like,

okay, a vva OSI soft pie, in the past,
they're working on a massive Databricks

integration, so their system will be
able to be integrated with Databricks.

So that's kinda like, oh, okay.

So there must be a lot
of driver behind that.

And I think, well why are you
putting your stuff in the cloud?

What are all these systems promising
people trying to do with it?

Like if your wind farms, going
back to that example where's

the data ad originating from?

Well, it's originating from
all over the place, right?

So you gotta put it somewhere and you
know, you might as well just put it in the

cloud and have it there in a data center.

You can put it on prem, but, you
know, it's just as easy to do it

in some data centers somewhere too.

And then you have access to it and
you can work with it from there.

But yeah, it's a big subject and I'm not
really clear at this point, like where

we're going with the cloud solutions and
what's actually gonna be accomplished

from it other than it just being kind
of a place to aggregate things and

maybe the new technology on top of
that allow us to work with it in a,

in a more effective way in the future.

Denis: Let's focus on the problem of the
lakes that you mentioned, the data lakes.

And I Absolutely hate the jargon
in the data world, but you have

the data lakes and you had the
data warehouses of the past.

The warehouse was like
organized data easy.

To query or was the lake was,
let's say, easy to put data into.

The problem was that those
lakes became data swamps.

No one found a way in it, and
the data was essentially bad.

What

the Databricks invented was
the lake house, warehouse, and

the data lake combined forming

a lake house, essentially means
that your day lake is more

organized and easy to query.

So that was one innovation.

And then to answer your
question, why you would want to

move the data into the cloud.

I think there are a
few compelling reasons.

What I see nowadays is that machine
learning is becoming more productized.

In the Past you had like data
scientists writing the code, building

the models, tweaking it very long.

Whereas now we rely on
off the shelf models.

That are living in the cloud.

So if your data is in the cloud, you
get access to a lot of pre-built stuff

by some very smart companies like
Microsoft and so on that have built some

very good models that you can leverage.

So I think it's to get access to this
additional intelligence, in my opinion,

is a good reason to go to the cloud.

And the second would be compute.

Like in essence, Databricks is.

Just a layer built on top of Spark.

It's a big data framework.

In a nutshell, what happens is that we
do distributed computing in a sense that

it's not only one CPO calculating or like
one computer, but many computers in the

farm, which makes it possible to compute,
let's say, some very large calculations.

Lonnie: Yeah that's good.

Yeah, you're teaching me some things here.

That's good.

, One of The things that is a challenge
around time series data, industrial data

that is tying into some of the things
you're saying is these systems have

become very large it used to be like a
pie system could, you could collect like

reasonably five or 6 million points.

You know, and it would be fine now.

I think it was in 20 13, 14 they scaled
that up to 20 million and like a year ago

was talking to a company that was, you
know, their PI system couldn't handle it.

And it was like, you know, they
had like 15 or almost 20 million

points and it was just like, just.

Not, you know, and, and, and
they needed, they needed more.

They needed more a, a system that
could actually like elegantly scale.

And so scaling isn't just about storage,
it's about retrieval and organizing

and all those other things too.

So, so the underlying, underlying
historians are starting to change

along with the tooling and the
things that sit on top of them

and how they all work together.

Then I think, yeah, the AI aspects
and the you know, all the models and

things like that are very interesting.

And it's not the same as it was in the
early days in machine learning and the

data scientists and it was very specific
on the things you, the things you

could do and the way you had to do 'em.

So.

I think we are entering a new period
right now that that people kind of really

wanted to have, and we kind of skipped
over what happened in the teens, but

that, that the, the teens were a lot about
getting stuff into the cloud and trying

to do machine learning and realizing
that there were a lot of problems around

that And data quality was, it is the
big one which I know you're focused on,

which is a really good thing to focus on
because still to this day is a challenge.

Going back to the.

The early days in the nineties
when nobody was thinking about it.

Right.

And so yeah, so anyway it, it is gonna be
exciting to see, to see what happens from

here because, you know, just the last two
years really, we've seen AI just come onto

the scene in a Real way, in a no joke.

This is.

The stuff works kind of way.

So I think in our space
there's slow adoption.

Companies are still trying to figure
out, okay, well what am I gonna it?

But at the same time, I think
everybody realizes that, okay, this

is something that we've gotta, like,
figure out how to utilize that.

And in order to utilize that, like
what we're just discussing, right?

Denis: Yeah, I think we painted a very
interesting timeline, essentially a map of

how it was in the past and how things just
kept growing on top, on top, and on top.

You also mentioned some consolidation,
especially with AVEVA purchasing

many other players and trying to
build one stack in that sense.

I'm curious about maybe to end
with some of your predictions.

Based on what you've seen in the past,
where would you say things are going?

Are we gonna have a
simplification of technologies?

Are we gonna have more technologies?

Stack on top of each other.

What do you think?

Lonnie: Man, that's a good question.

I'm not a hundred percent sure
about any Of this because there's

been some like attempts and failure
failures along the way in our space.

I just, I can't say specifically where
we're going to end up going to, but

I, I will say that that AI will have
a real impact and it'll be a positive

impact I believe overall it'll have
a positive impact on our industry.

And I think it's going to allow a lot
of these things that we've kind of been

trying to do for a long time around
equipment optimization, minimizing

downtime, improving, uh, things.

Also things just like being able to
operate systems effectively without

requiring humans to do every little thing,
which is still a lot of times the case.

So I think it'll have, I think, I think
we'll definitely have that happening

and I think these systems that are
being put together now will allow that.

But, it's, it is just still a lot
of moving parts in everything.

And, And so we, we've gotta figure out
some of these problems on how to deal with

them in maybe a robust way or something.

Like, I could tell you like,
one problem for instance is.

Like what we're talking around
structuring data or, or data quality.

And that's always been uh, an issue when
you go to get, when you go to access data,

you know, well, how do you access it?

How's it structured?

How do you find what you're looking for?

And then if when you do get it,
is it actually like good data?

Are you missing data?

Do you have bad values or where that's
been a problem like since day one.

And I think with AI.

And a lot of these new tools,
they can actually be used to

help solve those problems.

So I'm not talking about like having
all your data, like kind of there ready

to go and you can feed it into an AI.

I'm talking about like AI working
at a lower level to determine when

the data quality's off or to figure
out how to structure that data that

that normally, you know, if you
have a 15 million point system, it's

really hard for a human to do that.

It can take months and months
of a team to figure out it.

What data, what points are what, but with,
with some of this new, these new models

and these new systems maybe can really
help us in that area, which is something

that would be really awesome I think.

Denis: Yeah, absolutely.

I agree.

What I personally predict, I think
there's a lot of things that will change

or will be created, but I don't see the
existing systems going anywhere soon.

Especially because they're
also used for control.

Right.

And people are very hesitant
about changing something that

actually steers the process.

Lonnie: Yeah, I haven't really
like, seen a lot of success with The

whole, like, there's a factory 4.0

or 3.0,

I can't remember which, what number it is.

Anyway where it's a factory of
the future where everything's

like, oh, nice and organized.

You can just plug and play things
and everything's discovery and

self discovery and all this is,
that's never gone anywhere because

people won't agree on those things.

And you have all those o old systems that
nobody's just gonna go replace just to

get to the new factory and all that stuff.

So, so yeah the old stuff will, is
still here it's not going anywhere.

Maybe in a hundred years.

We will have, even then
you're still gonna find stuff.

So you gotta like deal with that, right?

It is just like a reality and you gotta
get over the fact that, hey, we're

not gonna be able to standardize this.

Not that people shouldn't try to
standardize new systems and do things.

I mean, I'm not against that, but
I'm just saying don't be delusional

thinking that we're eventually gonna
get to this perfect place where you

can just hook up to, you can plug into
a system and it's all self discovery

and everything can organize itself.

Those systems aren't there.

They're not designed to do that.

They'll never do that.

So it's gonna have to be a
layer or something in place.

It's gonna be able to kinda like take
all that and then do that part, you know?

And so that's what I think is exciting.

I think we're gonna get that part with ai.

I think we're gonna get there, but we're
not gonna get there in the way that

people have been trying to do it in the
past, which is from the ground up, right?

Denis: Yeah, me too.

And I'm also happy to hear that
I'm not out of a job just yet.

Lonnie this, this was great.

I think we really mapped out
the evolution of, let's say,

infrastructure in OT and how it
evolved and the past and the future.

Would you say we have missed anything?

Is there anything you would like to
add as a remark to the discussion?

Lonnie: We kind of gave a good overview
of the, of, of a little bit of history

and why we are where we are today.

And when you get into any company,
there's people that are in there that are

thinking about these things and trying
to figure 'em out, and I think that's.

You know, they're, what they're thinking
and what they're feeling about the

world and, and all the technology and
everything like that is, is, really

important to figure out, okay, what
are the things that matter and what

are the things that don't matter?

And uh, so, uh, that's kind of what I, who
I, what I try to pay attention to is what,

what uh, the actual people that are trying
to do the work, what are they actually,

uh, trying to do What are the problems?

what are their struggles?

And so.

It is definitely like a, an exciting
time because of all the new technologies

and all the capabilities and the costs
are coming down and the ability to

do things uh, now that we couldn't
do even a few years ago are coming.

So it's just, it's gonna be figuring
out, okay, who can do what, but

our industry's slow adopting, so,
you know, it won't happen overnight

even though we feel like it is.

Uh, but uh, you know, but it will
eventually happen in, in some way.

Denis: I agree and I believe a nice
message of hope that I've heard in

your thought was that if you are
like a company listening and you want

to start doing this whole AI stuff
or getting more modern, I think the

nice thing is that you don't have to
throw away everything you've built.

You can build on top of it in a smart way.

Lonnie: Yeah.

Yeah.

And and I would listening to this as
they're thinking about these things

is really like kind of going back to.

Consider what problems
you're trying to solve.

Like if you can start out at the beginning
of a project or figuring these things out,

what are you really wanting to accomplish
here as specifically as you can get.

That'll help I think improve your
chances of success around these systems,

because where I've seen companies
kind of fail or spend a lot of money

and not get much results, which I
guess also is considered failure.

they don't really have a good
plan They They we'll stuff and

then we'll hope for something.

and typically hasn't worked out.

So I would just say like, yeah,
definitely try to get as specific

as you can use cases, what you want
to do with it, what you wanna solve.

I know sometimes that can be hard.

Especially when you're talking to like
CEOs and things like that, where they're

you know, wanting to kind of it'll help
a lot to figure out where, how these

systems can be put together and what
tools you should be using and what

technologies you should be looking at.

Denis: Yeah, I think those are
some great tips to end the podcast.

So, Lonnie, what have you been, what
have you been up to the last two days?

And where can people find more information
about your work, what you're currently up

to, and how can we be in touch with you?

Lonnie: Yeah.

I think the best way if you're
interested in talking to me about

any of these types of things is
my email which is Lonnie Bowling,

L-O-N-N-I-E-B-O-W-L-I-N-G, at diemus.com,

D-I-E-M-U s.com.

Um, You can also go to my website, but
my website's not really updated very well

right now, so, that's at diamonds.com.

But anyway, yeah, if you can contact me
directly through email and and as far

as what I'm up to, um, you know, I am
uh, I'm kind of at a, I just, I just got

off a, a, sabbatical, um, for about, I.

I mean, maybe six months after
a startup that I did for a

year that didn't work out.

So I'm kind of at a little bit of a,
like figuring out what's next too.

So I'm interested in a lot of these things
that we're talking about, figuring out

the role that I can play that can help our
industry and help people in our industry.

But I haven't.

But I don't have the answers either.

But I know one thing is like, uh, you
know, I definitely can talk a lot about,

you know, the problems that we have
and, and, and what doesn't work and,

and maybe, you know, get involved in,
you know, coming up with some of these

interesting solutions in the future.

So that's what I'm doing.

So anyway, mainly consulting.

Denis: Well that sounds great.

I'm sure we'll hear you again
perhaps even on this podcast.

I don't think we quite
finished our discussion.

Lonnie: Yeah, it's been a long way too.

Yeah, it's a big topic, so, but
Denis, thanks for having me on.

I really appreciate it.

Denis: Yeah.

Great.

And thank you to the listener
for listening to new episodes

of the Industrial Data Podcast.

So we'll see you in the next one.

Thank you.

Bye-bye.