Episode #42: Data-Driven Revolution: Unpacking Microfactories and Modern Manufacturing with Errol Rodericks
In this episode of U.S. Manufacturing Today, hosted by Matt Horine, the focus is on the critical role data plays in the reindustrialization of American industry. Matt is joined by Errol Rodericks, Product Marketing Director at Denodo, a leader in data virtualization and logical data management. The discussion covers the rise of microfactories, driven by demand volatility, product variation, sustainability, reshoring, and AI. They explore how modern data infrastructure is essential for operational agility and real-time responsiveness in manufacturing. Key topics include the significance of transient data, challenges with legacy data systems, and the benefits of a logical data management framework. Errol also shares insights on preparing for AI automation and digital twins, emphasizing the need for real-time, trusted data to remain competitive in the evolving manufacturing landscape.
Links
- Errol on LinkedIn
- Denodo Website
- Navigating Trump 2.0
- Veryable Is Revitalizing U.S. Manufacturing
- Sign Up on the Veryable Platform
- Veryable Shop
Timestamps
- 00:00 Introduction to US Manufacturing Today
- 00:16 The Importance of Data in Modern Manufacturing
- 00:36 Guest Introduction: Errol Rodericks
- 01:37 Understanding Microfactories
- 02:50 The Five Forces Driving Microfactory Adoption
- 04:48 Data Challenges in Modern Manufacturing
- 06:53 The Role of AI in Microfactories
- 16:57 Logical Data Management Explained
- 25:50 Real-World Applications of Transient Data
- 32:22 Preparing for the Future of Smart Manufacturing
- 38:07 Conclusion and Key Takeaways
Episode Transcript
Matt Horine: [00:00:00] Welcome back to US Manufacturing today. The podcast powered by Veryable where we talk with the leaders, innovators, and change makers, shaping the future of American industry, along with providing regular updates on the state of manufacturing, the changing landscape policies, and more.
Today we're diving into one of the most overlooked but absolutely mission critical components of reindustrialization, which is data, not dashboards or reports, but the deep digital infrastructure that determines how fast we can build, how quickly we can adapt, and whether the US truly has the capacity to bring production home.
Joining us is someone who lives at the center of that transformation. Errol Rodericks is the product marketing director at denodo, the global leader in data virtualization and logical data management. He works with organizations around the world to modernize their data. Foundations for realtime insight, operational agility, and resilient manufacturing.
As industries shift toward micro factories, regional supply chains, and high mixed production, Earl has a front row seat to a simple but powerful [00:01:00] truth. Modern manufacturing can't scale without modern data infrastructure and today. He's here to help unpack what that really means. We'll explore why microfactors are rising, how legacy data systems are slowing down American manufacturing, what real-time operational data actually looks like on the shop floor and where AI and digital twins are taking us next.
Let's get into it. Errol, welcome to the show.
Errol Rodericks: Thanks for having me, Matt.
Matt Horine: Glad to be. Absolutely. We're very excited to have you on today because I think this is an extremely relevant topic. You see a fast pace of change going on in manufacturing. Buzzwords like ai, things that are happening seemingly all the time.
But, you know, one of the first things I wanted to start off with, why microfactors are exploding across industries, EVs, medical device, aerospace, and why data agility is their lifeline. And first, what is a microfactory for the benefit of our audience?
Errol Rodericks: Basically a microfactory. It's a small, modular, highly automated production node.
It's designed for real time responsiveness instead of [00:02:00] chasing scale, a microfactory typically would optimize for agility, uses what we call transient data, flexible labor. Of course, AI driven orchestration to produce exactly what is needed, exactly when it's needed, and as close as possible to the point of consumption.
So that's in a nutshell, what a microfactory agile, something real type.
Matt Horine: Yeah, that's a very succinct definition and something that people may be familiar with, but just weren't sure what to call it. So that's a really good setting of the stage. You see 'em popping up now with. EVs, med tech, aerospace. Why is the adoption accelerating so quickly?
If we could continue to set the stage around why that seems to be trending now and something that'll be very important in the future.
Errol Rodericks: Yeah, it's interesting. We are seeing it right across the different sub-verticals and manufacturing, if you like, and we are seeing some common reasons for it. Essentially there are, I would say, five forces.
The first is the demand volatility. It's up and down. As you probably know, demand is what takes a lot [00:03:00] of this drive. And then there is product variation. Everything from customized, personalized products to complexity and the products themselves, even the supply chains that sit above and below products here in the case of semiconductors, for example, they have their own supply chains and then they are part of somebody else's supply chain.
The third driving force is sustainability. There's this need to actually become better at sustainability, and sustainability means quite a few things. But by and large it means being aware and being respectful of the environment, but more importantly, being compliant. Then there is support is a big reshoring, active area.
People are increasingly moving their manufacturing closer to where they would like it to be. And as you reshore, micro factories come to the forefront. And finally, probably one of the most significant changes recently is the advent of. In its entirety, [00:04:00] ai artificial intelligence covers a multitude of sin as a whole stack of AI that's being deployed.
And as a result, when you have these five forces, demand volatility, product variation, sustainability, reassuring, and ai, traditional plants can't adapt fast enough. Uh, microfactors, they thrive in environments where you need to change direction quick, whether you're building batteries, and those batteries might well be for electric vehicles or consumer goods, or medical devices or simply packaging.
And I say simply, but that probably no packaging can be quite complex. So that's what's driving it, these five forces, we believe.
Matt Horine: That makes a lot of sense. It's certainly something we've seen play out over the last number of years as manufacturers shift towards this overarching theme of more localized, more modular production.
How does this change and this transformation change the way that leaders need to think about the data? Because I think that's really at the heart of it, right?
Errol Rodericks: Yeah, as you can imagine, we are right in the middle of it. We've seen data [00:05:00] change over the last 20 odd years, and in the past it's been all about centralizing the data you need in advance.
But as we've moved forward, especially in the last few years, you can't rely on centralized data all the time, especially when. You've got the elasticity that is built into micro fact. The flexibility microfactors now are increasingly relying on what we call transient data. So this is data that doesn't stick around for very long, and it also is very hard to centralize, to actually plan in advance and have it central because by the time you centralize it into any kind of data architecture.
It's already out of date. So things like machine signals or quality events, or as you well know, workforce available. Then there are supplier ets, inventory flows. This kind of transient data that's live at the time, realtime. And if you don't have a unified realtime data layer, you can't orchestrate operations and orchestration is at the heart of [00:06:00] Microfactory.
So if you like, we provide the data, nervous systems, we connect everything. Without resorting to massive data movement or copying the data stays where it is, and we can, as a result, give you access to not just the data, but also to the meaning of the data, what it really means, so that you can actually start to leave data where it is.
And often there is data that cannot be moved, should not be moved. Sovereignty being one of them. So without moving massive amounts of data or copying data, that's where logical data management steps in. And normally if you are moving massive amounts of data, like in lake houses or data warehouses, that's normally too slow for Microfactory dynamics.
So that's where data really comes in. And we are seeing a change going into the new year, and AI is now at the forefront and we have to be ready for. AI in a microfactory environment, which is different.
Matt Horine: Yeah, it certainly sounds like it. It's almost to the point where I've heard about this [00:07:00] trend enough that in the next couple of years, will microfactory models even exist without some type of modernized digital infrastructure?
Errol Rodericks: I, I don't believe, I think this is the new, if you like, heartbeat of a microfactory, the new digital nervous system, as we say. And like I said, if you are relying on Microfactory to produce, you need. AI assisting with the orchestration at the highest level, and it's important to understand what that means.
But when you say AI as the orchestrator, AI basically turns micro factories from reactive systems into adaptive ones, for example. Without the data, you can't have responsible AI or trusted ai, and without trusted ai, it's hard to actually understand which line should run a job. Or when you need more labor or which machine is trending toward failure in advance.
Predictive, how to sequence micro batches when to reroute work to another node. So you know, you need AI not just to [00:08:00] predict like it has done in the past. It can act, and that's where what we call gen AI comes in and we provide. AI radio, Gentech, AI ready data when it's needed.
Matt Horine: Some of that framing of the microfactory of what exactly is needed, when exactly is needed.
That makes a lot of sense. Do you think it's something early adopters, are they seeing a competitive advantage right now? Other over competitors who aren't maybe thinking about this in terms of that forward step and kind of the turn of the new year and where this is all headed?
Errol Rodericks: We are definitely seeing manufacturers in different sectors, if you like.
Starting to see an increase or an improvement in time to market. Time to customization personalized products, so that's increasing demand pool. It requires additional labor, which again, as I said you you know about, but it also sees a minimization of things like quality risk. It slows things that would normally slow a machine in rerouting work that doesn't happen.
We have also [00:09:00] seen that micro factories have a strong reliance on being able to handle flexible supply chains because they're seeing supply delay, they're seeing sanctions sometimes limiting some of the supply chain. So being able to. Like again, reroute, find alternative sources. All of that comes into the microfactory environment and the results in terms of comp, competitive stamps.
Is they're quicker, they're faster, they're much more flexible. They're meeting demand, modern demand, way better than we ever could before with the traditional forms of manufacturing, and I say traditional, but this has been around for a little while now.
Matt Horine: Been around a little while, but evolving quickly.
You did highlight an interesting point there about some of the hidden bottlenecks and the potential hidden bottlenecks with the outdated data architectures and traditional centralized systems like MES and ERP and even other IOT systems. They can't keep up with the speed of what's needed and required in a microfactory.
And what are the biggest data challenges manufacturers face [00:10:00] when trying to modernize or scale these operations outside of those normal systems?
Errol Rodericks: So when traditional manufacturers deal with data, that data pulls into three main camps. First of all, there's a core systems that you find in manufacturing systems and yes, uh, definitely ERP and that core data.
Is typically centralized, but it relies on a lot of legacy data that sits around it that's difficult to centralize and to move. The second thing is the transitory data. This, they don't get a look into. It's gone before they can even see it. That definitely impacts the productivity of a microfactory if they don't have access to that transitory real-time data.
And the third thing is data that cannot be moved that they should have access to, depending on what it is they are producing, whether it's sovereignty that's stopping them from moving [00:11:00] certain types of data. And again, like I said, think about pharmaceuticals. Think about not discreet, but continuous manufacturing where there is information that they just can't share.
They don't want to put it into some central place, but they need access to it. At the time of manufacturing, when you have real time data, transitory data, legacy data, that's, and you say you talked about speed and timing, but there's also volume. Some of this data. The data volumes are just increasing exponentially.
Again, trying to pull that into a centralized system, there's no need for it, and it really is restrictive. So when you can access it, where it's when you need it, that then makes it much more usable, much more useful, and much more. Competitive in terms of what they produce. Finally, the other real challenge that they all face if they try to address all of this is trust.
Trusting that data. You cannot rely on data that's not trusted. It's not [00:12:00] compliant, it's not respectful of all regulations. So when you can do that without having to move the data in real time at the time it's happening. At the time those signals are coming in, then it makes sense of all the other intelligence you've built into your microfactory because you know you wanna simulate stuff, and I'm sure we'll talk about digital twins and threads, but to do that properly, you need real time live data.
And that's where they're facing challenges. This is the challenge we squarely deal with. That's with smack bang. In the middle of our sweet spot, and we're seeing a lot of manufacturing manufacturers buying into this. They actually see the need. They trial it out, and within no time, they see the value, so that elasticity in their manufacturing practice.
Is based on trusted elastic data.
Matt Horine: Really great point, because I think the idea around elasticity is that it counters what we could describe as almost like a [00:13:00] latency, right? Latency and impact in realtime decision making on the shop floor. Most manufacturers they may know or they may not know their data architectures holding them back, or is it, do you see the trend as something a completely hidden bottleneck, or is this something that most folks are aware of?
Where do you see most of the market landing on that spectrum?
Errol Rodericks: As we said, there is a spectrum. There are the forward thinking manufacturers who have realized that they're being held by, held back by lack of the right data or the right type of data at the right time. And once they realize that, then they set about trying to fix that.
And that's why we step in. There are the laggards, and I won't say there are too many of them, but there are laggards who are still convinced. I don't know if I'm allowed to say laggards, but anyway, there are a few who, who take time, who take that time in actually getting there, but we have no doubt that they will get there in the end.
And it's either that or extinction. It's a question of survival. Yes, there's a whole spectrum of them. Some [00:14:00] are, and let me say this, the electric vehicle industry is right at the forefront. They're charging ahead. They get it. The types of data that they are interested in and what they do with the data, not just the data during manufacturing, but also the aftermarket data.
It's probably the biggest. Producer of data there is out there, have to think about the data that goes through an electric vehicle in managing it. You're talking volumes of real-time data. So yeah, there is a spectrum. People are starting to realize that there is a bottleneck and it's really, it wasn't until the advent of AI that they started to realize what's possible out there.
The art of what's possible. We didn't know we could do this. We didn't even know this kind of data existed or was accessible in real time, and we didn't know that we could circumvent supply chain routes in this way using this type of real time data. So we are getting that, and as I said, next year will be better than this year.
Matt Horine: Thinking through this in terms of labor, 'cause you mentioned it as part of that elasticity and hopefully [00:15:00] what we can define as infinite flexible capacity because of the amount of data and the ability to adapt. How does the lack of connected data now make it harder to train or onboard new talent in manufacturing, whether that's on the shop floor or in senior director of operations positions, folks who are running supply chains.
Errol Rodericks: That's the other part of this equation. There's two parts to the equation. One is the elastic data. When you look at micro factories, they're starting to break the, what I'd call the old model, fixed shifts and rigid headcount, running short batches, fast changeovers, high variation. That only works if labor availability matches operational demand in real time.
So the old school is starting to struggle with that because if they, once they realize. The benefits of micro factories. Then they realize, of course the benefits of elastic data. Real-time data, logical data, but equally at the same time, they run into lists problem, which [00:16:00] is, but our operations aren't in line with this.
We have the old way of doing things. Now how do we cope with short batches or fast changeovers, high variation, all these things. So essentially you, they're looking for like they've got a data cloud or a elastic data cloud. Need probably a labor cloud if such a thing exists. I believe you could probably tell me more about that.
But the ability to scale people up or down instantly. In the same way we scale compute or data
Matt Horine: major component of that equation. 'cause it's something we see a lot of is these kind of fixed labor models that aren't responsive demand. They certainly don't take into consideration the larger macro environment and it leaves people on the hook when there's a down cycle, which everything's cyclical.
There's ups and downs. If it was just steady 100% demand, yeah, we'd probably be in some type of utopian world where there's just, that is a
Errol Rodericks: utopian world and which, you
Matt Horine: know, not for this podcast to answer, but I don't think anybody who's come on has found that. And that's, we try to manage to the best of, but [00:17:00] turning now to logical data management and virtualized access, how logical data layers unify manufacturing systems without copying or moving data.
Those are big trends right now. For listeners who aren't familiar, what is logical data management and how does it differ from traditional integrations like you would see with we're standing up an ERP or we're switching ERPs, or something to that effect.
Errol Rodericks: Logical Data management actually isn't that new.
It's been around for 20 odd years, but it was really hard to convince people that it's a good thing to do to access data. So let me explain what is Logical Data management is about. Accessing the meaning of data and the data when you need it, without moving the data to any, without necessarily moving the data to any centralized resource.
So accessing data where it is, but in order to do that logical data, key component of logical data management is data trust and the governance that goes into accessing that data. So not only do [00:18:00] we make the information available, but we also make sure that it's. Available in a trusted and governed way. In a compliant way.
So that's the first part of it. The second thing is when you access this data and you know where it is, it leads to the buildup of the semantic layer, the meaning of this data. Because when you're talking about a large number of data sources, realtime data sources or anything like that, you are looking at needing to access or know what these data sources mean in terms of.
A particular use case, whatever that use case might be in manufacturing, whether it's improved visibility or it's predictable ops, or it's compliance or it's supply chain optimization. Knowing whether the data that you are looking at is relevant, it's up to date, trusted, and that's where the semantic layer of logical data management comes in.
And as a result, we build, if you like, for want of a better word. And I don't like using the traditional words in data [00:19:00] management, but let's use this one. It builds up a catalog. It builds up a catalog of what's out there, what it means, and when to use it, when to bring it in, and when to trust it, and how to use it, how to deliver it.
And we are seeing as part of logical data management. The delivery model of data is changing. So what do I mean by the delivery model? How do you make this information available to the people that matter? And they may not be it and data experts, they may be operations people, they may be supply chain people, they may be manufacturing, procurement people, et cetera.
Now we make it available to something new, reasonably new, and we are seen as probably one of the best suppliers of these things called data products. So we productize data, we don't say. Here's a big data fabric. You should access it a deep dive in there and have a good look around and see what you can use.
We deliver curated sets of data called data products, which are curated for certain use cases. So for [00:20:00] example, your buildings. Let's see, what's a digital twin? Everything you need to implement that digital twin, everything. You need the intelligence. Data, the insights that would come probably from an AI stack and even the the agents that you'd need to deploy to automate some of it.
That's all encapsulated in a data product. I remember when I first started at Denodo, somebody said to me, what do you see Denodo as? And I saw the early start of some of our data products for manufacturers, for example. And I said, we're probably gonna be known as data products are us in the future. 'cause you'll be able to pick up one of these data products for something very specific.
So Logical data management brings in the delivery of this data, making it available to people who are not necessarily IT and data trained, but are more from the operations of business side of things to consume this information. So in a nutshell, logical data management, logical access to data, a [00:21:00] governed secure way.
With the meaning of that data embedded in the semantics and in the catalog, and then delivered as data products that can be immediately consumed on a self-service basis.
Matt Horine: Pretty fascinating work. Mostly because go all the way back. You think of data virtualization in the sense of how does it allow manufacturers to access information without physically duplicating it, and then why eliminating that data duplication is so important for modern manufacturing environments.
There's a lot of data out there, as you said, and being able to. Get it into a product is pretty incredible. But what's the danger of that duplication, where it does exist?
Errol Rodericks: Oh, the dangers are many and they mostly to do with the lack of governed access. Okay, so first of all, is the data trusted? Is the information you're providing up to date, is it trusted?
Is it relevant? So the meaning of that data suddenly takes on a new life, and so does the, as I said, the trustworthiness. This is this data current. Is, are we missing [00:22:00] anything? And then how do I use this data? That's where a lot of manufacturers spend a lot of time trying to work out how relevant or what to do with this data.
What do we integrate it into? Which system, which manufacturing system do we integrate it into? But luckily, from a logical data management platform, it's already prompting you, saying you probably want to integrate this into your SAP. Environment or your other manufacturing environment, et cetera, and you probably want to integrate it into this module and this workflow that needs to be automated and orchestrated, and that's why you will use these insights and use perhaps something from your adopted or one that we recommend AI Stack.
Use these tools to make use of this, to give you the dashboard that you need. I'll give you a quick example. One of the key things that they're struggling with right now is compliance. Let's call it holistic compliance, so that's regulatory, compliance, sustainability, ESG, you name it. To do that in a, in [00:23:00] an efficient way, you've gotta contact just about everybody in your manufacturing organization to make sure they're not in breach, whether they're doing maintenance, repair and ops.
Procurement. You could be buying pencils from an unsuitable source and you'd be in breach. So when you have to. Update all of this. Get all this information in. You want to know that this is relevant information. Imagine if you had a data product, that self populated itself, it polled everybody. It updated the framework, the compliance framework automatically, and your auditors would be over the moon about it.
So would you German people, and you'd be doing the right thing as well. So when you have something like that available as a data product, who's not going to want it? Who's not going to say yes? Let's just make sure we implement that. They may want assurances, they may want to test it, which they do. And often in the case of compliance, they test it with internal auditors to start with.
So that's just an example. Now, when it comes to control towers and [00:24:00] command centers and digital twins or digital threads, you'll see the automation that happens and what can go wrong when you get a simulation incomplete, I won't say incorrect, but incomplete. There's something missing and that's, that's when you start to see the difference in.
A factory. Factory that is simulating and therefore predicting and operating much better than another one that's almost there, but not quite there, and missing certain bits, and therefore falling behind on delivery. Falling behind on quality, falling behind on compliance, smart factories falling by the wayside.
They're not so smart when they find that they're missing data. Microfactors not a good quote for data here is that. They are more likely to be smart because they embrace this upfront. So the dangers on are, are multiple. I've just gone through a few. But efficiency, compliance, predictability, visibility, all that is at risk.
Matt Horine: It's a really good framing of it [00:25:00] because I think for one, you mapped out how it looks like for an engineer and operator inside and on the shop floor, right? Can they adapt them into workflows? Can they do them into, do things that impact change?
Errol Rodericks: Does it make their life easier, better? Yeah, absolutely.
Because that's the only viewpoint you want. We don't really want to have just an IT and data viewpoint because that doesn't help the manufacturers themselves, the people who are actually involved in the operations.
Matt Horine: Absolutely. So that's those kinds of improvements and the efficiencies that are, that come out of that seem to be manyfold, but also avoiding the dangers and pitfalls of doing it incorrectly or not having that visibility.
I think you did highlight one really good real world example there, but diving in a little further with transient sensor data and how that enables live responses to quality issues or environmental changes in small batch variability. Can you walk us through a real world example of a microfactory using like a live sensor data, a system to adjust operations in real time?
Errol Rodericks: I'll give you a couple maybe to, let's [00:26:00] take, let's start with life sciences, the pharmaceutical company manufacturing. Drugs or medical devices. So realtime data tells them that there's a disruption to the supply chain. Transient data says, uh, this is geopolitical in, in, in nature. It is a disruptive, and we can tell you from now that it's probably going to disrupt your supply chain real time significantly.
And the people. Downstream from your supply chain are going to feel the impact. So with transient data, first of all, you know that's coming before, before anybody else knows. Uh, it's live data. You've just been told about it or your system's just been told about it. And, and as a result, they, they use AI and some of the predictive algorithms to actually.
Do two things. One, find alternative sources of that su supply, whatever it is that's being supplied. And the second is alternative roots. Alternative roots [00:27:00] to getting that data into getting that the material, sorry, into the factory at the right time. So that's happening right now. I can't say exactly what their names are, but just think big pharmaceutical companies running into this all the time.
Mainly in the US We've seen some of it elsewhere in Europe and in Asia. But it's mainly in the US that they're already dealing with this upfront. So supply chain in life sciences, big time. Then let's go to our offerings in electric vehicles. Okay, so electric vehicle manufacturer needing to produce the best electric vehicles they possibly can are competing on, ah, I can pick a number of things that they're competing on, but let's take battery life as one.
The other is the nature of developments in, in, in the world of batteries being manufactured as part of their manufacturing and them needing live, transient data on what the batches look like that are coming out, what the [00:28:00] productivity is. And so they, again, they're using transient data. So it's not just transient data, but I'm focusing on transient data because that's Duffy, that's the one to overcome.
So we're seeing it in electric vehicles, we're seeing it in life sciences, manufacturing and supply chains. And we are, we are even seeing it in consumer packaged goods where they're needing the latest and the best on health and wellness. Uh, the health and wellness nature of what they're producing. And they want the best because they're focusing on providing the best to their downstream customers.
And again, transient data tells them what the current developments are, what the current research is, and what the current trends are. And again, they get it first. So I don't know if that sort of explains some of this. It, I can go into a lot more detail, but obviously I would then need to start presenting, which I don't want.
Matt Horine: Presentation would certainly be worthy because I think we're covering a lot of things that run through people's minds on a daily basis. You talk about pharmaceuticals or life sciences or those types of things. [00:29:00] Obviously highly technical down to the uh, the humidity change or something to that effect.
If you look at something, maybe something more industrial, maybe like heavy industrial or something to that effect, big concerns around preventing rework and you know, how real realtime quality monitoring can prevent scrap that rework the yield loss. It seems like it's got a ton of applications there as well.
Errol Rodericks: Without a doubt. So on heavy machinery, sub supply chains to other manufacturers, you've got rework, you've got reducing the number of faults if you like, and actually you're not just reducing the number of faults, predicting that you're going to have a problem with the quality of the next batch coming out.
And being able to predict that relies on quite a variety of data down to temperature sensors, uh, that give you early warning that things are not. You know, you, you, you hit the nail on the head there in terms of the quality of the output, the produced goods.
Matt Horine: Yeah, that makes a lot of sense. [00:30:00] And something again that we have folks in shop floor think about every day.
Here's a bigger question. Do you think that micro factories are inherently po better positioned than maybe larger legacy plants to utilize this type of fast moving data? And why would you think one way or the other? Or is it applicable to both?
Errol Rodericks: I would say it's probably applicable to both, but let's just take the benefits that Microfactors have over their larger cousins.
They're just faster moving. They are capable of consuming this data better than most because they're flexible, because they are. They're focused on smaller runs. And more specific variations on products, they will make better use of this data faster than most, faster than their competition. Which is why if there's one thing that a microfactory has to be, it has to be elastic.
And this is the definition of elastic. Like elastic data is real time data. It's transient data, it's live data. And I [00:31:00] think they're in a position to make better use of that. And you may see this trend, I think you'll see this trend start to increase as people realize the economics behind micro factories are no longer restrictive.
And that's a whole different discussion there. But because they're now financially feasible, economically feasible, and they're delivering outstanding results, they are probably our prime target, or will be our prime target going forward. In, in helping with elastic data microfactors, that doesn't mean we won't support the bigger manufacturers, more traditional manufacturers, but I think in time they're going to probably break down what they do into a system of micro factories that become more flexible, that become more in tune with giving customers what they want, when they want it.
So as that trend increases. We'll see micro factories come to the forefront in terms of using this type of data.
Matt Horine: That's something really to think about because I think the [00:32:00] trend toward mass customization, you know what I would call like mass and fast, something that the consumer, anybody really, even if you're an industrial buyer or something like that, expects almost this Amazon like delivery timeframe with customization and data.
Sounds like it's key to unlocking that. I think you also touched on a couple of other. Key points here and what we can say. We're firmly in the AI era at this point, and, but there's a long way to go into the future and preparing for that future ai, digital twins in the next decade. What smart manufacturing will actually mean in that era, should manufacturers be.
What should they be doing today to prepare their data foundations for AI automations or digital twins? Is there something that they can ramp up into, or how would you advise on something like that?
Errol Rodericks: An easy answer would be, make sure you've got a logical data management framework in place. But let's take a step back.
One of our predictions for next year is that in the field of ai, the focus will be on outcomes, not on [00:33:00] actual AI or data for that matter. It'll be on outcomes. So microfactors and smart manufacturers. They're gonna focus on outcomes, and once they are focused on KPIs and outcomes, they will then focus on the right type of data, the right type of data products, and the right AI stack to help them get to those outcomes.
So working backwards from the outcomes. So if they work backwards from the outcomes, the, and they know that curated sets of data and the right types of AI embedded in there. Is what's going to get them to that KPI. They'll also start to realize that the markets in general, not just in manufacturing, but in general, is not going to tolerate AI experimentation anymore.
Yeah. 'cause we've been through a lot of AI experimentation right Across all industries. So as we go into 2026, it'll be quarterly results, three month outcomes. They'll be looking at. What can you do for me in three months, 90 days? [00:34:00] And that's a lot of pressure on the data and IT teams that have to fulfill that.
But once you start with the outcome in mind and you need curated data products to help with your current digital twin, let's call it a predictability model, that needs simulation, that needs data that's real and needs the right AI stack, once you focus on the outcomes and you know that you've gotta deliver this type of outcome.
You'll automatically start building your data infrastructure, your data operations, to suit that kind of operation, focused on outcomes, three month outcomes, no AI experimentation, real results, measurable results before and after being able to do all of that. So setting up for that is key. And in order to sell for that, you do need your logical data framework.
And we are not saying get rid of whatever you've got. Lake houses and data warehouses and business warehouses, they all have a role to play, but what's going to be key to being [00:35:00] able to cope with what's coming down the line and the expectations that are coming down the line. You need this kind of real time, logical data framework.
Matt Horine: Makes a lot of sense. This is something we ask on the show fairly often because we do have experts on, but we look for some consultative free advice is what we would call it. If you were a C-suite of a company that handles manufacturing down to some type of supply chain leader, shop floor leader, if you're feeling behind on something like this, what's the first step you can take or you should take to building your understanding of where you are and how you can project out what it is that you're trying to build?
Errol Rodericks: There's probably two or three things they should be doing. Number one, they should be familiarizing themselves with what's possible. Okay. A lot of shop front operations aren't fully aware of what's possible out there right now. What could change? So parking that, that's a question of being educated on what's possible.
And yeah, we are happy to do it, as are many other organizations, but then you have to actually look at [00:36:00] priorities. Where are your top three five priorities? They need to be addressed immediately and for what reason? So when you say priorities, whether it is improved personalization of products will made to all products with a speed component in that KPI.
So your priorities, where are you going to focus? Remember working backwards from those outcomes to what your data framework should look like, and then see if your data framework measures up to that, which it almost certainly. Won't if you haven't already implemented something that's real time and logic.
So I would say, look at your priorities. Make sure you know what's possible. Look at your priorities. Focus on the outcomes you need now, and there will be many, but like I said, prioritize the top three to five and then more you can expand and say, can my data framework meet these outcomes? And if they can't, then you are, you are in need of help.
Matt Horine: That does make a lot of sense and there are a lot of folks who are trying to navigate [00:37:00] this on their own or trying to make, make sense out of all of it. If our listeners wanted to find out more about this or about what Denodo does, where could they go to find you and. More about implementing this type of system.
Errol Rodericks: You can come to our website, denodo.com. There's a ton of information there. We have special pages that are focused on manufacturing that you can go to and make use of all the resources that are available there. You can contact me anyway you like. I'm on LinkedIn if I'm available through there and through this podcast as well.
I'm sure my details will be made available. I'm very happy to talk to anybody who's cares to speak about the subject 'cause it's a subject I like.
Matt Horine: We could tell. It's certainly something that's fascinating and trying to make sense out of the chaos every day. So this has been extremely informative and we appreciate you coming on US manufacturing today, Earl.
Errol Rodericks: Hey, it's been my pleasure. I was hoping to actually get more also at some stage into what you guys do above and beyond later, but that for another day, isn't it, Matt?
Matt Horine: [00:38:00] Absolutely. We'll certainly stay in touch and happy to do thanks. Thanks again for coming on.
Errol Rodericks: Thank you for having me. Thank you all for listening.
Matt Horine: That wraps up today's conversation with Errol. A few big takeaways, stand out. Microfactors aren't just a trend, they're a structural shift in how products will be built in the US and the reindustrialization movement. Legacy centralized data architectures are one of the biggest hidden bottlenecks slowing American industry today.
Logical data management and virtualization aren't just buzzwords. They're enabling technologies that connect the real time operational picture. And as AI automation and digital twins accelerate. Manufacturers who build agile data layers today will be the ones who compete globally tomorrow
to stay ahead of the curve and to help plan your strategy, please check out our [00:26:00] website at www.veryableops.com and under the resources section titled Trump 2.0, where you can see the framework around upcoming policies and how it will impact you and your business. If you're on socials, give us a follow on LinkedIn, X, formerly Twitter, and Instagram. And if you're enjoying the podcast, please feel free to follow the show on Apple Podcasts, Spotify, or YouTube, and leave us a rating and don't forget to subscribe. Thank you again for joining us and learning more about how you can make your way.