Beyond Dashboards – How Decision Intelligence Optimizes Industrial Asset Risk

We don't have a data problem. We have a decision problem.

How can we parse the data in a way that allows us to

have the context that we need to make an

operational decision now when every

minute or every hour has cost and increased

risk associated with it? So it's not the data side, it's

how do we boil all of that data down and filter it into

the key decisions that we need to make today.

Industrial companies don't have a data problem, they have a decision problem.

Andy Pruitt explains how AI can turn mountains of sensor data

into smarter, faster decisions about physical asset risk.

Welcome to Data Driven.

Hello and welcome back to Data Driven, the podcast where we explore the

emerging industry of data science, artificial intelligence,

and without data engineering, all of it is for nothing.

My favoritest data engineer in the world is unable to make it today.

However, I do bring you a different Andy. Today's guest is

Andy Pruitt, who is a

co-founder and CEO at Lumiscent. Lumiscent. Let

me say that again. As the co-founder and CEO of Lumiscent,

And he drives the vision and execution of innovative

solutions that optimize asset performance, reduce risks,

and enhance customer financial outcomes. Welcome to the show, Andi.

Hey, thanks for having us, Frank, and welcome to the Lumisphere.

Nice, nice. I like it. I like it. Next thing you'll have an attraction in

Vegas. Yeah, yeah, yeah. And it's good that— it's

never good when Andi Leonard isn't here, but at least I don't have to get

confused between the 2 Andis. But so tell us about your company.

What is it that you do? And in the virtual green room, we did mention

IoT. And I know that IoT is a—

you don't hear that term as much anymore. You hear physical AI, you hear

IoT seems to be falling out of favor within

the industry, probably because the S in IoT stands for

security. Yeah. Which is a joke that I thought everyone had

heard. But apparently yesterday I said it to somebody and She laughed,

so here we go. We're living in a post-widget

era, Frank. We want the proper insights and outcomes from

whatever devices we have on the shop floor. Lumasent is the

intelligence layer for physical operations, so

specifically decision intelligence for physical asset risk.

So who is your primary

customer base? Is it industry? Is it— you deal with the

SCADA drivers or the software layer on top of that, or you feeding the data

into a pipeline? Like, where do you sit on that stack?

So if you think about traditional operation

with the various levels of software, whether it's

ERP, SCADA, manufacturing execution, building

control systems, all the way up the stack, we are the

first kind of platform that is solely focused on

the risk layer. And so that's our core kind

of focus area. And reason being is that over half the

industry doesn't have dedicated asset risk managers. So how

can we create visibility from an asset risk

standpoint into that stack in an

unencumbered way that allows you the information that you need

to make decisions around physical asset risk? So

how do we find gold in the data that we're generating? by

consolidating it, analyzing it, and then providing direction,

next step direction to the people that are using it.

Interesting. What is physical asset risk? What do you mean?

That's a great question. Yeah. Physical asset risk could

be any failure that you have. And let's

say you're in a food plant, single points of failure

sometimes aren't always abundantly clear. So if you're in a food

plant, like your makeup air system, so if you think about a big food

production facility, usually on the roof they got this big jet motor that's

creating negative air in the facility, because if that's offline, then we may

suck a whole bunch of dust and debris into the facilities and contaminate food.

So that single point of failure is a risk asset.

If it fails during operations,

likely that's a business interruption claim. So likely they there's going to be

an insurance claim that is associated with that failure.

So we're really there to help manage those single points of

failure in your organization that leave you exposed.

So it's looking at that specific activity for those

specific assets and pulling that information

out so that you can make decisions today on the things that

are going to cause failure in your environment that may cause an

undue exposure. Does that make sense? Oh, that makes a lot of sense. So

it's that type of asset risk because that could— it was, it was

interesting because asset risk could mean a lot of different things to a lot

of different people, right? I might— started my career in finance, so when I heard

that, I was like, doesn't jive with the physical AI aspect of this.

But what you said makes a lot of sense. I totally appreciate it. When I

do say asset risk, a lot of people immediately go to

financial risk. The 2 things are related, right?

But we're talking about the physical

machinery, the physical operations that when

they're impacted, they cause a downstream financial

risk. And so one of the things that we highlight

is 6 different dimensions of risk. So when we're

classifying those assets, what could happen if they

fail? Is this an environmental— Right. Is it a strategic

risk? It is a supply chain risk.

So it's quantifying the downstream impact of that

individual asset failure. And the dollars associated with

it, it is financial risk directly related,

just outside of the banking industry. And I'll give you an example. You can have

risk exposures and single point of failures that could be half a billion

dollars or more. So there are, are big

exposures here that we're talking about. It's not small dollar with single points

of failure. And I would imagine that if you can quantify that to

a real number, so what, you can imagine somebody, maybe

not in the food industry, but, oh, the HVAC system goes down, right? What does

that mean? That, so what is that? So what, you call somebody and they fix

it? But if you can quantify that, no, the downstream effect of this is,

in the extreme case, probably half a billion dollars, now you gotta get the board's

attention. When you think about how we've designed

our industrial systems to really absorb

failure instead of building risk as

infrastructure and then using data to drive those

decisions around risk. We need

to shift our thinking around it because things may not be quite

as intuitive as we think they are, Frank. We may not think about a

single leak depending on where it happens taking down

the entire operation, right? We think about those things as, hey, They're going

to happen eventually. We've got to deal with them. We can design systems not only

to absorb the failure, but to alert us early on those things because

we're on an industrial site. You have leaks that come up all the

time. Where are the important ones? The one that's gushing 1,000

liters a second or a minute. So it's really

understanding the context of where the risk exists

and then highlighting when it bubbles to the top. What do I need to

do right now? Because you think about the amount of information that's

flying at you at an industrial site, it's hard to

parse. Think about today, all the social

media that you get, all the bombardment of information.

How do you surf through all that stuff to find

the information that you really need or you're really discovering? The same thing

happens in big business, right? Our customers are heavy industry, so it's

Mining, oil and gas, forest products, heavy industry manufacturing,

and they're being asked to make decisions faster

with less information. So how do we, one, give them

better information and better data to drive the decisions they have to make,

but contextualize it in a fashion where it's really understood?

So it's intelligence with integrity, but it's also

driving contextual awareness of the decisions that have to be made

around asset risk. That makes a lot of sense because all these industrial

organizations and entities are probably— there's sensors everywhere now,

everything, but they're probably just spewing tons and tons

of data. So we are— sorry, Frank, go ahead.

No, go ahead, go ahead. We have plenty of data, but we don't make use

of that. What do you make of that? You're bang on.

We don't have a data problem. We have a decision problem.

How can we parse the data in a way that allows us to

have the context that we need to make an

operational decision now, when every

minute or every hour has cost and increased

risk associated with it? So it's not the data side, it's

how do we boil all of that data down and filter it into

the key decisions that we need to make today,

that we need to make now.

Interesting. Volume of data. This goes back

years ago, but I was trying to highlight

the sheer volume of data we were trying to deal with. And this

is a mining example. So I was giving a presentation at the

time. It was when 1 gigabyte thumb drives were brand

new. And you and I are old enough where we remember That was a big

deal. It was a big deal. And I came out on

stage and I pulled out my thumb drive and I said, this holds 1

gigabyte of data. How much data

do you think we have as an organization? And this is

advising a major corporation. So we have enough of these

thumb drives to fill a ship the size of the

Titanic. Wow. And that's

all data, corporate financial, geologic, health, safety,

environmental risk, asset, and its compounded

growth is 10% a year. So

how do we actually parse through that volume of data to

find the specific information that we need,

turn it into knowledge that allows us to execute on it?

That's kind of a needle in a haystack problem. But I think the era that

we're living in now, where we have the right tools,

as long as it has the right harness around it, I think we have a

way to parse that data. And if we have specific subject

matter expertise, then those things can be more

agentically driven. That makes a lot of sense. And I,

when you say subject matter expertise, I would imagine

a few years ago would have been human only, but now I think we're talking

about some kind of hybrid of human and some kind of AI subject matter

expertise. I think you have to have hybrid,

right? I think you have to have a hybrid solution. You look at the

which is what, 3 years away, 4 years away? Scarily,

yes, 3 years away. That means

3 quarters of the people that I came up with in

industry will be retired. They're gone. So no longer is

it a 10-year time horizon. We're within 3 years, so

it's happening every day. And when that knowledge

walks out, it is impossible to replace

because there's just not enough people entering the industry, whether it's a risk manager,

whether it's maintenance managers, whether it's trades, your

millwrights, your electricians, your, your Red

Seal-type carpenters, right? So we have to

provide tools. that enable them to make better decisions when they

don't necessarily have the longevity or operating context

to make those decisions because they just haven't been in industry that long.

Wow. Yeah, no, I think that there's some kind of— people talk about the

job apocalypse, but I also think there's a retirement apocalypse, and

particularly in the trades, I would say.

Absolutely, absolutely. And those senior people too.

Sometimes they'll come back, do contract work, but they don't want to deal with any

of the corporate stuff, right?

Hey, I can help you to achieve this objective, but I only want to

do that. And then I want to go back to being retired. Yeah, I imagine

that there's going to be a lot of people that are going to be called

out of retirement and things like that. And you're seeing that also in the tech

sphere too, where companies are offering early retirement to people who

literally built the company, right? Yeah. And I understand they have

reasons, they have their financial reasons, right or wrong, but I

just think that they're throwing out the baby with the bathwater when it comes with

the, the institutional knowledge to do it that quickly and that

rapidly. I, I just, I think we're gonna see the long-term

consequences of that in short order. I,

I totally agree. And that's what LumaSint is here to help

solve for, is to help you make those decisions, capture the

information, make those decisions so that you don't lose

step when you lose resources. So you can maintain some

consistency and contextual awareness. So it really is

driving that decision intelligence that comes from a place of authority.

And the other way I like to think about it too is this is

not— what we focus on too is not general

purpose. It is specific. Right. So our

expertise is a wedge. So if you think about Everyone talks about like

council of agents doing things. Right. We're really focused on

that asset risk in heavy industry. That's our focus area.

And we're deep, deep on that subject. So

we want to make sure that people understand that when we're

looking at risk, it can't be good enough, right?

It needs to be contextual. It needs to be

validated and it needs to be validatable. We like

to say intelligence with integrity. In that we can show you the math

of how a decision was arrived at. So that's a— that

was a good segue into my next question is a sensor

detects an anomaly, maybe a family of sensors that are related. What happens

between signal and making a

decision? Because I would imagine that capturing the data, capturing the signal,

that used to be a technical hurdle. I think that's— those days are gone. But

now I think the last mile problem is very real. You have all this data,

You have all this. Okay, now what? How do you make a decision? And

then obviously different time horizons and things like that. How do you—

what happens after that? Yeah, the

signal through capture needs

to take the next step. So we've got the first mile out of

the way, to your point. I think we've got the sensor stuff figured out.

I think we have most of the infrastructure figured out. We know how to collect

the data. We know how to consolidate it. We even know how to put

together good dashboards, right? Lots of blinking lights.

But I've also encountered over the last half a dozen years

that people are blind to dashboards now. So we've gotten to this

level where the dashboards has gotten us so far, we've

got to take the next step, which is what are the decisions that

have to be made based on the data that we have? Right. And so

it's no longer signal, gather the

data, turn it into some level of a

base of knowledge, and then turn those into actionable

outcomes. You really need to do that as a whole stack

in order to go beyond the dashboard. So it,

it's really driving that decision intelligence by being

smart about the data that you look at, the data that you collect,

how you parse it, and how it drives those decisions.

So it really is beyond the signal because sometimes it's too

much, right? Let's take a one-sensor example. You can have a

high-fidelity sensor on your shop floor that generates

50 gigs worth of data in a year. Right. It's delivering

more if it's streaming. What's important in that data

was the anomaly that I was able to detect and the resource

that I was able to get to that anomaly to address it before it turned

into a larger issue. So I'm even

like thinking about this is not predictive because that's where we always go.

Isn't— aren't you talking about predictive maintenance? No, I'm talking about

prioritizing the risk that's coming through a signal

so you know what to do today. Wow. The predictive stuff is

great. There are people that, that are really awesome in that

space that are telling you 6 months from now this is going to

happen. We're talking about the operational block and tackle of risk.

What do I need to do today? What do I need to do tomorrow? What

do I need to do on Friday? This seems like the tactical version of preventative

maintenance. It seems like your heads-up

display that's really going to help to drive better

decisions around physical asset risk.

Interesting. Interesting. So

Not all anomalies are created equal, right? So obviously

there has to be some kind of way to determine, because we're

beyond, you know, you're working beyond, way beyond kind of traditional

predictive maintenance, right? Not all anomalies are created equal.

Not all breakdowns are created equal. How do you

address that? Obviously there's gotta be some kind of, you mentioned blocking and tackling,

obviously you wanna block the big— Yeah. You don't wanna block the biggest guy on

the other team, but you wanna block the, I'm gonna use an American football analogy,

sorry for my European friends, and global listeners,

but you want to block the guy who is about to tackle your quarterback,

not the biggest guy, right? And I think that

traditional preventive maintenance blocks the most

obvious threat, but, or the, the biggest guy on the other team.

You want to protect your quarterback, right? That's really the, what you're trying to do.

Absolutely. Those are the people who win games. Absolutely.

I love your analogy and the framing of it. Because

that's it, is that, hey, if my predictive maintenance strategy

is focused solely inside plant on rotating equipment, but

I haven't looked at the sectional valves that supply my fresh

water for my system, and that goes down and my

plant catches on fire, what predictive maintenance system

actually helped me drive production when my water's off? So

it's making sure that you know where that little guy is going to come

out of nowhere and take out your quarterback, even

with all the big guys. So that's where our focus area is,

that there's all these pieces that are inside and around plants

that will catch you off guard. And normally we don't figure

that out during the engineering of the plant because we're not

designing risk as infrastructure. We're designing it

for production outcomes. So yes, we design it to be hardy, we design it for

the output, But as we build and

operate, new context comes in, new

risks that didn't necessarily occur when you actually built the operation.

I'll give you a really good example about this. Let's say

you're playing football and all of a sudden they put an autonomous robot on the

field. Right. Changes the game, right? So we had a

customer that came to us and said, look, we've started to put all

these autonomous robots on the factory floor. And we didn't

realize that they're heavy. And so they're using the

same infrastructure on our walls and our ceiling where our lighting is

hung, where our fire control systems are hung. It's

causing a high level of vibration. We need all that stuff

to work. How do we deal with that? We don't even have

mass contextual awareness about robots coming into the factory floor, and now we

have a risk that's being introduced almost in real time. That

is impacting things like health, life, safety systems. So

we have to be aware that game of football is going to change

and we're going to have new dimensions that are added all the time.

And it's not slow, Frank. I'm not talking about 2 years, this is going to

change. It's next week, it's changing. It may be next

month. There's a new operating context.

No, and factories are not built for— factories are not like

software, right? These things are built for Decades of run.

And I don't have a lot of experience in manufacturing, but as a car guy,

I kind of understand. If they change up the model of the Cadillac Escalade, they

have to take down the assembly line and re-engineer it.

But you're right. But these factories have probably

been going part running, some of them probably

the better, these buildings, these infrastructures, at least a century old.

Yes, some of them are. We have customers that do have facilities that are

100 years old. But it's also the new stuff that's being

built as well, is that— Right. We're always trying to drive optimization,

and that's a nice way of saying we're trying to optimize for cost.

And so we don't end up getting all of the

risk infrastructure that we necessarily need as part of the project. So

how do we enable those customers by having something that's really

quick to set up, really quick to deploy, really

quick to gather information, and really quick to drive decision

intelligence? I think that's very important to our customers is

just speed. Is this is happening to us, how quickly can you

help us address this challenge that's coming up? In the

field, whoever adopts this faster is going to get better

profits, better, better output, better, ultimately better profits, and

potentially put you out of business. There's definitely a real,

there's definitely a real motivator there. I think another analogy would be my wife and

I are house hunting. And we saw this beautiful

the coaxial cable is kind of stapled to the outside of the wall.

And so, and I was, we were looking at the

aesthetic problems of that. And it was just like, pretty sure they didn't have

Cat5 in mind or co— you know, co— cable

TV in mind when they built the house. Right. So you kind of have to—

There's— in that example, it's just mostly

aesthetics in terms of the cables showing. But if you look at newer construction,

there's fiber optic cable inside the house in some extreme cases and things like

that. Like, how far can you push old infrastructure?

How far can you push it? I think we're at a breaking point. So you

look at how far can you push old infrastructure? We've got something

like 420,000 miles of unmonitored,

unmanaged, like fresh and wastewater treatment systems across North

America. So how far can you push it? Only so far until it breaks,

and then there's an incredible amount of pain, right? We

have, I think, the benefit of having

designed, engineered, built, and operated

feats of amazing infrastructure, whether it's our

railroads, whether it was our bridges system, whether it is water

treatment or wastewater treatment. I'm from Alaska originally, so I always

go back the construction of the Trans-Alaska Pipeline. Right. And man,

peak construction, that thing was a global—

absolute global— one of the largest global projects in the world.

Billions and billions of dollars being spent on its construction. And it's

lasted for 40-some years. We don't build infrastructure

like that anymore. And the infrastructure that we did build to that

level, to that capability, is aging. And the vast

majority of it, again, is not measured. It's not monitored.

It's not managed. And our big goal is to be able to

replace it. But the cost of it is just amazing.

I think in the US, the infrastructure backlog is

trillions and trillions of dollars of what needs just to replace the

20% of either poor or

dangerously poor deteriorated assets.

Yeah, no, that's a sobering thought.

What are the major points of failure in industrial AI,

right? Are these bad? Because it sounds like what you do

is a very comprehensive approach. You don't just look at the building, you look at,

oh, this pipe fails this. There has to be some kind of

human and expertise that you have to go in. And do you

offer the services or do you offer the software or do you offer both? That

would be my question. With our customers, we've

got a program that's a luminaries program.

So we partner with the businesses that want that extra

help. Because what we found is that customers just don't want to

buy software anymore. They want you to partner with them

deeply. And I don't mean that like marketing partner. They want you to

be helping to own the results that you're

trying to drive. So it's not necessarily

traditional consulting because when we're there for the life of the asset,

we're there to help you throughout the life of the asset. And so when we're

talking with you, it's about optimizing everything from maintenance strategy to risk

strategy. So we're just looking at things slightly

different. So yeah, we help customers with the software, but I think our

human touch is something that's really important in today's kind

of AI age. and the expertise we bring to the table,

right? To find a company that actually has depth and expertise that

can deeply partner to help walk you through that

journey is important. And you said fairly comprehensive. We're

comprehensive in our focus area. So we're not building

control, we're not a process control system. We're

looking at it again strictly from that risk lens for those

60% of companies that don't have a dedicated asset risk manager.

That's our piece of the pie where we focus.

Interesting. Yeah, it seems like there's a lot of places where somebody

can get distracted, right? In terms of like,

because there's a lot of moving parts here, right? And I think that's an understatement.

What, how do you start? It seems to me, for me, I'm looking at this

as somebody who's not in industrial control and not in this industry.

I look at these problems and I'm like, where do you start? It's not

just blueprints. You actually have to walk around and kick the dirt around, right? And

like, where do you start? Where does one start? I know that's a

kind of a small question with a big answer. Frank,

that's another great question. Where do you start? Where do we start?

So all of our customers have an insurance partner, and

that insurance partner has— most of them will

have what's called risk engineer. Oh, okay. And that

risk engineer is a subject matter expert that's

coming in to define what are the things that need to be

monitored, what are the things that need to be managed. And

we're in to help after that. So we'll partner with

those insurers, we'll partner with those risk engineers to make

sure that once they've done their assessment and the report, that's

translated into the system. So it's translated into what needs

to be measured, monitored, and managed. So we meet

the customer where they're at in terms of defining

single points of failure. Usually they already know, right? They've got the data

inside their environment, pulling out the things that are important and working with

the risk engineer to be able to make sure that they're— everything that's

been identified is monitored correctly.

So that's where we start. And that

risk engineer is probably not coming in cold, right? That risk engineer

probably, you know, has worked—

Long-term relationships. One of the reasons why we have

an FM-approved product, and Factory Mutual

is an engineering-based risk

leader, but insurance company, and we make

FM-approved products. FM approvals is their approvals

division. So they focus on making sure the products that

hold the FM logo have been tested to their most rigorous

standards. FM, when they come in— and this is not

us, right? We're the, we're the help block and tackle afterwards. You've

decided what field looks— let's help you play the game. FM

recommendations for their customers. Right? Wow. So

48,000 things that we need to do to mitigate

risks across all their customer base. That's one insurance company

that covered almost $1 trillion in risk,

those 48,000 recommendations. So now how do we take those

48,000 recommendations and translate them into

operational activity? What do we need to do to effectively manage them

within the context of risk? So we

partner deeply with our insurance partners who are very

data-driven, Frank, very data-driven. But we're looking at the

world slightly different in that we have engineered

risk and we have actuarial risk, right?

What's the difference? That's a great question.

Engineered risk is based on, I've

looked at the asset I understand

how it's going to operate. I've written recommendations

to mitigate everything from fire, flood, earthquake, tornado,

whatever it is, but I'm making those recommendations

based on a very long history that I've developed.

Actuarial risk is what financial losses

have I had and what financial losses have I had that can

predict a loss in the future? So one is a

financial activity and one is an

engineering and field activity. And so insurance

companies are split behind 2 lines:

actuarial-based risk or engineering-based risk.

We're solidly in the— we're there to help with your

engineered-based risk outcomes, even though it has big

implications for the actuarial risk outcomes as well.

But we have to follow the physical operations. So that's where we're focused. is

on that engineered risk side.

Interesting. Interesting. It's such a fascinating world I didn't really

think about. You think about

factories and you think about, as long as you have an eyewash station and things

like this, and you don't really think about, at least from my point of view,

right? This is a whole world that's completely new to me. It's fascinating. And

I would imagine those 48,000 items that have been

highlighted are probably triaged or would be

dependent on whatever the individual factory would be, not even the industry,

not even the customer, right? The individual facility probably has to

get their own triage of X, Y, and Z.

Yeah. And, and our job for those customers

is to take that information and illuminate it.

What are the important things? Let's put them in, let's

manage them correctly, and let's do it quickly

and cost-effectively because the risks that we're trying to manage

have pretty big implications to the operation. Otherwise they wouldn't be

on an engineered report. As something, let's take a little bit of a

dogleg and this is definitely data-driven. I think you'll

find it interesting. I love you're using our podcast name throughout the whole thing. I

love that. That'll definitely help with the AI optimization.

Yeah, I just need to get to say Lumicent 10 more times and it's like,

no, actually— There you go. You look at what's happened in

terms of what are called natural catastrophes, right? Fire,

floods, earthquakes, hurricanes, stuff like that. In that

you have insurers that have pulled out of entire markets, said,

yeah, too much risk there, we're not going to write there anymore,

whether it's floods in Florida, whether it's wildfires in

California, it's the insurance company. Commerce protects

itself, right, where it's protected. So in order

to get back, get insurance companies

back to the table, new

ways of monitoring, measuring, and managing

risk outcomes had to be created. And one of them is parametrics.

Parametrics is something that Normally when I say that, if you're not in the

insurance industry, you have no idea what it is. Parametrics is almost

a product that was brought out to help mitigate

risk for the consumer, whether it's the business and the insurance

company. And a good example, and this is again, we make

devices that help enable this type of stuff. We're not a

parametric product, but for example, you've

had, you have no insurance, you run a distribution

facility in Florida. And you've been flooded so many times you can't get insurance.

And then the insurance companies pulled out of the market. They came back in, but

they said, we're going to give you a parametric product. Oh, what's that? We're going

to pre-agree what your losses are

based on the level of water. Hmm. So they go in and

put a sensor on and say, look, water gets to 5 meters,

we're going to pay out. We'll pay out within 72 hours. Gets to 10

meters, We'll pay out, we'll pay out in 72 hours. So

then you've pre-adjusted, not post-adjustment. You haven't

had a loss and then had to have an underwriter come in and figure out

what your losses are and then end up in court. You're pre-agreeing what that

looks like, and you have a sensor that's actually the arbitrator

of the risk event. So

that's one component. You also have the—

we're looking at satellite imagery. We're looking at roofing systems,

we're looking at positioning in terms of where the facility's at. All

of those risk components that are around it from the

insurance side are tools the insurer is

using to make your policy go up. And they're not intentionally trying to make your

policy go up, they're trying to manage their risk. Right. Our job on behalf of

our customers is to mitigate the risk as much as

possible so that they can prove to the insurer that

they've taken every step possible to be able to

mitigate the risks that have been identified.

So insurers are using these tools to protect their risk.

We've come up with a set of tools to help the customer

protect their end so they don't have to end up with constant

pressure on policy premiums. Oh,

that makes a lot of sense. So

Given that there's a lot of dollars attached to this

and there's an IT element, there's an

engineering element, who runs this? Who

should— I guess there's the should— who should own this and who

actually owns it, right? This is clearly, if you're a big manufacturing

outfit, even a moderate one, like the C-suite, this has to

bubble up to the C-suite, doesn't it? It does.

We know that about 57% of industry doesn't have dedicated asset

risk managers. So it ends up being— Really? Yeah, it's

hard to believe, but they don't. And we don't make enough of them either. We

don't have enough graduates. It ends up falling usually

to the CFO, and the CFO usually manages

it as an insurance policy.

So we're managing the financial outcome. So it's back to actuarial risk.

versus engineered risk. So what we're trying to do, what we

are doing, is bringing decision intelligence

to the table for that C-suite that gives them

the operational knowledge to make better actuarial

decisions around the insurance that they have to buy

because there's nobody dedicated. And I, Frank, I always come at

it from the maintenance side too, because at my heart, I'm an operations

person. I, I'm a maintenance guy. I'm not a, a risk guy. But

what I saw in my experience was that

we don't have enough on the risk side, and our maintenance brothers and

sisters end up being the people that are the de facto asset

risk manager. So how do we put better tools for

operational to understand what they need to do to

manage risk? And how do we put tools in the place of

plant managers and CFOs to be able to

drive those decisions more effectively.

Interesting. You mentioned that we don't graduate enough of this. So this is an

actual degree you can get? I say this as a parent of a high school

junior, because this seems like a growth industry and a stable thing. Is this an

MBA? Is this a graduate level? Is— are there undergrad programs? Like,

where would Risk engineering programs. So

there are— Really? Absolutely. Absolutely. So

you could go in specifically to risk engineering. And

normally those people migrate into places like maybe they work for

Tyco or Simplex or Siemens, right? Health,

life safety control systems. You'll find risk

engineers in your EPCMs, your engineering procurement construction

management companies. So it's, there's a whole

body around risk engineering that is invisible.

If you've never worked in an industrial operation that has a dedicated risk engineer,

which is again, 60%, you probably not know that

the discipline even exists. That is

fascinating to me that there's this whole discipline that you can get a

degree in and had never heard of it, right? That's interesting.

And it sounds like a growth market. So interesting to get a

degree in, I would imagine. It's a growth market among

so many other of the STEM that we need to invest more in,

whether it's metallurgical engineers, right? I think China is

graduating more metallurgical engineers a

year than we have operating

in the practice in North America. Yeah, I'm not surprised. For

one, they have a 4-to-1 roughly Yes. Population

advantage, right? So all things being equal, it would be a 4 to 1, but

it's not. There's more to it than that. But how

do you calculate— because it's always

the thing that you never see coming, right? How do you prevent a failure that

never happens? Because that, as someone who's been in data

science and questions like that, one of the questions I'd get

that would always amuse me and bother me philosophically,

later was, how do we predict when XYZ

is going to happen? And I said, when's the last time that happened? How often

does it happen? It never happened. Like, how do you—

and I say, this is predictive analytics. It's not predicting the future, right? Yeah.

So how do you— is that

possible to do that, right? Is that still impossible? Philosophically,

I would say it's probably impossible to do. You can probably approximate it.

But what is a better answer? Is there a better

answer than that?

I've always had a challenge with this too on the predictive side

is that, okay, so you say you're going to give me this

predictive analytic model and it's going to be a crystal ball and it's going to

tell me that I'm driving down

whatever Highway 1, Route 66 and

at mile marker 328, my transmission's gonna

go out. How do you know that, right? What gives you the indi— like,

what tells you that? What dataset allows you to do that?

That always frustrated me because I just can't really

get aggregated. I could say maybe if I look at a

million cars and I look at the same

operating context with the same car on the same road, in the same

context, maybe I can derive something out of that.

I think a better way to look at it is if we have inputs on

the car that I'm driving, engine temperature, the type of

fuel I put into it, I can have a

probability of certainty, which is one

level, and I can have a

certainty around the context. So I can have

multidimensional Hey, I just detected

an issue with the engine. It may be one of

5 things. Can you give me some more information? If you give me some

more information on one of those 5 potential things, maybe I'm

low on oil, maybe I'm low on gas, but if you give me more

context, I can get a greater certainty of what the

problem is. And I can actually adjust my probability

of failure based on those inputs. So to me,

it's much the whole context around, yeah, we've

got predictive that's based on all of our past history, but we want to look

forward. That's saying, hey, my mean time between failure was

this, so we know that about every 60,000 hours we're going to have an

operating failure in the engine. I think a better way to look at it

contextually for what we have to do today is, what's my probability

that this is going to happen? the likelihood. What's my

certainty? What information and context do I have around that probability

that actually allows me to make that decision? And that's again decision

intelligence. That's really highlighting what information do

I need to give a better and more accurate

insight into what needs to be done. So I want to take kind of

the crystal ball out of the equation and really talk about

Hey, there's probabilities and certainties that we need to talk about and

we need to actually put into the equation so that we deal with the

reality of what we're dealing with today, not what may

happen. Right.

Fascinating. I'm just— not every day I find out that a

whole new field exists, so I'm still processing that.

How did you get into this? I'm sorry, go ahead. Go ahead. Were

you Googling it? I did. Yeah. I was like, there's a number of

schools. You weren't kidding. There was, there was, I was just like, wow, that's— and

I would imagine that you probably have to be pretty good at math. You probably

have to be like a, like a nerdy MBA, almost an

engineer, right? That would be kind of the way I would describe it.

Though my, my oldest is very good at math and he,

you know, he builds robotics and he's in a— so I'm like looking,

I'm like, right now he wants to be a mechanical engineer and I certainly want

to encourage that, but if he decides he doesn't like it, I'm definitely going to

float this in his direction of— because it's an

interesting— as a dad, you're always selfishly looking out for your kids, right?

But how did you get into this, right? You mentioned you grew up in Alaska,

and Alaska seems like an interesting place to grow up

because obviously you're not that far from the wilderness, but also it

takes just a lot of good engineering just to survive year-round up

there, right? So How'd you get

into this? I come by

the industrial operation space naturally. So I

started my career as I was growing up and through university in

oil and gas. And so it's wired into me that we're

always planning for the worst-case scenario. So you don't go

outside of 40 below zero to do something without

planning Planning ahead.

Yeah. I mean, the couple of times that I did

that were not good, right? A couple of times I went into the outback

hiking without a gun, for example, and got pinned down by a

moose who a bear had just eaten her calf. So

if you think about risk engineering, it's part of planning, but

it's also about building resiliency. How do you

deal with that issue? And then how you respond to it, how you manage

it, is actually the determination of if you survive.

And so that kind of underlying thought process, I didn't

even think, I didn't even know risk engineering was a thing until 10

years into my corporate career. But what I found is that

it's built into me naturally because where I come from, what I was exposed to,

and from who I learned from, right? From the people I learned from, they

went, look, we're running a multi-billion dollar operation

above the Arctic Circle. that's housing 1,000 people. This is oil and

gas. We have to plan, schedule,

execute, measure, monitor,

manage seamlessly. Our tolerances for

risk are really low, right? Really low.

I was in an event that actually forced me further into this

area, and that was an event where we actually lost

power at 40 below zero. So it was an

entire site in Alaska where the

80 kV high voltage power line was dug

up by an excavator, right? And here you are, you got, you have

600 people out in the middle of nowhere. It's 40

below zero. And I had a, at the time, a

just fantastic general manager that I learned a bunch from,

you know, called all the leadership into a room and said, I need you to

go look at ventilation systems. I need you to go to wastewater, and I need

you to go to freshwater treatment. I need you to go to living facilities. He

knew what to do to help manage the outcome because

he knew we got about 2 hours before stuff starts freezing up. And if

it freezes at 40 below zero in February in

Alaska, we're down. We're done, right? We're done until the spring

because we're not going to get this stuff unthawed, everything that we need to get

unthawed. So my experience in that was, wow,

this is amazing. You have like a Captain Picard that knows

what to do. Oh, and then I made—

I come from oil and gas when this incident, so I came with my mental

model of oil and gas because this is when I started oil and gas

post-Valdez oil spill. So we had all these

emergency management and drills that we went through and

very structured. What I saw in that organization was we

had personality-driven risk leadership. We had people that had longevity

in industry that knew the decisions that they had to make and how they had

to make them. 25 years later, we don't have

enough of those people. So the incidents that I was in are like, I don't

know what to do. Do you know what to do? I've got a piece of

it. You've got a piece of it. But a single person that could actually

quarterback all those things, that had the risk model in their head to go, We

got an hour here, we got 3 hours over there. That was very

personality-driven, not process-driven. And so now

fast forward today, that's what we need to do with Lumicent.

That's what we are doing with Lumicent is really

being that risk manager in a box for physical asset

risk, not full enterprise risk management, but physical asset

risk to be able to look at those things, make a determination of what's

important, and direct your people to those things. I, my

experience over the last 25 years is that our complexity has

increased enormously. Our volume of data

to be able to sift through, turn into knowledge, has increased enormously.

But our ability to actually make decisions founded

in data and be able to execute on them has actually decreased.

And that's mostly due to we don't have enough people in industry with the

longevity to be able to help make those decisions.

Interesting. Wow. It's a, it's a whole, I think it's a

testament to how good the people were at their jobs that we didn't know this

existed. You never had

to guess whether or not if you turned on the light switch, the light would

come on. You never had to guess whether or not if you flush the toilet,

whether it was going to flush because the infrastructure was there and supported

and built by people that knew how to manage it. And you only know

when it's not there. Yeah, I'm sorry. You only know when it's not there. Yeah.

Wow. Wow. We're coming close to

time and I want to be— this is fascinating. Not every day a whole new

world opens up. And just for your information, I'm going to

fold this to my kid. I live near Johns Hopkins University and they have a

master's degree in risk. So maybe I can talk him into staying closer to home.

But But what, what can people do to

find out more about you and your company and how to reach out

to you? www.lumicent.com,

L-U-M-I-C-E-N-T.com. Reach out to me on

LinkedIn and I'm available, right? I just love

talking to customers about this subject. And yeah, you reach out

directly through the website, me or anyone on our team, but I'm happy to

talk to customers anytime I get the opportunity. I'm

passionate. About helping customers solve the challenges

that they have around physical asset and decision intelligence.

Awesome. That's amazing. And we'll make sure all the pertinent links are in the

show notes. And with that, we'll go to the outro.

That was great, man. I really— that was amazing. I'm just floored

that there's this whole— Thanks again to Andy Pruitt for joining us and showing

how AI and decision intelligence can help organizations turn

mountains of industrial data into smarter decisions about asset

risk. Subscribe, share the episode, and join us next time on

Data Driven.