The Intelligent Edge: AI’s Impact on IoT Architecture

As artificial intelligence transforms industries, IoT implementation faces both opportunities and challenges. This episode explores how AI is being integrated into IoT solutions across different architectural layers, from edge processing to cloud analytics.

Nick Earle sits down with Simon Maselli, CEO and founder of Minnovation Technologies, to discuss real-world applications of AI in IoT deployments.

1. The different types of AI and their practical applications in IoT
2. How AI is being implemented across edge, cloud, and application layers
3. Real-world case studies of AI-powered IoT solutions
4. The future of data science and AI in industrial applications
5. Practical advice for companies starting their AI journey in IoT

Subscribe to IoT Leaders

Subscribe now

Subscribe now

Subscribe now

Whitepapers

How Can AI Improve My IoT Deployment?

Learn more

Blogs

Defining AIoT: Artificial Intelligence and the Internet of Things

Learn more

News

Eseye Predicts 2025 Will Redefine IoT Connectivity: MNO Evolution, NB-IoT Decline, and the Rise of AIoT

Learn more

Whitepapers Featured

Eseye’s IoT and Telecom Predictions for 2025

Learn more

Transcript

Intro:

You are listening to IoT Leaders, a podcast from Eseye that shares real IoT stories from the field about digital transformation, lessons learned, success stories, and innovation strategies that work.

Nick Earle:

The subject of this month’s IoT Leaders Podcast is, what else? AI. Everyone’s talking about it, and that’s part of the problem. Everybody is talking about it, and a lot of people are getting very excited.

A lot of people are getting very confused, and people don’t really know, where to start. And so, we decided, that we would have, an AI specific, podcast, and we have gone to Australia for this. There’s a great company called Minnovation, and I’m about to talk to their CEO, Simon Marcelli, CEO and founder, and they’ve done a lot of great work with real case studies on AI. Not only that, but they they’ve got a very simple model, for looking at all of the different components of AI at different levels and looking at what is the architectural taxonomy and what benefits does AI have at what layer of the architectural model, which is something that for us at Eseye, we believe very strongly in.

So, there’s a lot of, synergy here. Simon’s a great guy. He’s in the office late at night in Melbourne, talking to me on the podcast, and I think you’re really going to enjoy, this podcast on, AI, in, in the IoT space, together with some, war stories from my trip to CES, last week. So, with that, let’s get going with Simon Maselli, the CEO of Minnovation.

Enjoy.

Nick Earle:

Hi, Simon. Welcome to the IoT Leaders Podcast.

Simon Maselli:

Nick, great to be here. Great to be here.

Nick Earle:

And I believe you’re in Melbourne. That’s right. So, we’re on the other side of the world to each other right now.

Simon Maselli:

That’s it. Communicating through the power of the Internet. What will we do without the Internet?

Nick Earle:

Well, maybe what we do is that we’ll enhance it with AI, which is going to be the subject of today’s podcast. So, we are a physical. We’re not virtual. We’re real. We’re not avatars.

But we are going to talk about AI.

And before we do that, maybe just an opportunity for you to introduce yourself, to the people who listen and watch, the podcast because you are the CEO and founder of a company called Minnovation.

And so, I looked at your website. It says your additional transformation consulting and delivery, company. So why don’t you give me a quick intro to what you do and, maybe a couple of examples. We’ll talk about case studies later on, but a couple of examples of the type of work that you do with clients.

Simon Maselli:

Yeah. Sure. So, we essentially provide data and analytics, as a service to most, most, local government organizations here in Victoria in Australia, but also to a lot of private organizations. So, this a lot of people who own, lots of physical assets in terms of buildings and infrastructure and roads and, all of those sorts of things. So, the type of work that we typically do is we, connect a lot of their operational technology systems, so their OT technology systems, merge all of their data from multiple sites, and the multiple sites might have different datasets in different formats.

So, we turn that into a really, neat and aggregated dataset that their analytics teams can take away and analyze. And if they don’t have those teams, we also provide those services that, do the analytics, and we even provide a virtual building management, service. So, we actually, make recommendations and organize the contractors and all of those sorts of things on behalf of the of the end user and the customer. And driving a lot of those systems and a lot of those services is essentially AI, which is what we’re talking about today.

Nick Earle:

And can you give me, a couple of, just high-level examples of the types of projects? Because I believe you, I mean, you’re not you work across Southeast Asia, not just, Australia. So, I know you’re involved in a big building construction project in, Thailand, I believe.

Simon Maselli:

Exactly. So, were very, fortunate to win the One Bangkok project. So, that’s probably one of the largest, building construction projects in the world at the moment.

There is, over a million square meters of floor space, thirty-two towers. And what we’re doing is we’re connecting all of the building comfort, data, all of the people occupancy data, a whole heap of different datasets from everything from underground parking to the building, management systems and providing that to the downstream applications. So, they’ve got, other applications that manage things like the tenancies, manage things like their air conditioning and HVAC systems. So, we aggregate that data, pass it through to them, and that’s been a really exciting project and a really large-scale project for us.

But we also work at the other end of the scale. So, some people, particularly local government, have lots of these smaller buildings scattered around the place.

And what’s really important for them is to be able to manage all of those buildings, all of those assets, in a central location.

And, again, those ones are the are the really curly ones, but they’re the interesting ones because they have this, unique need about having disparate datasets and having this data and this equipment that, has a lot of legacy equipment in there that, doesn’t have, modern connectivity with it. You might have on the other end of the spectrum, a really modern, a really good system, and it’s being able to bring all of that data together to give the end user a really great experience, when they’re when they’re analyzing and working with their data.

Nick Earle:

And so, I believe just from a technology point of view, a lot of our listeners are on the on the tech side of it.

I think I’m right in saying that you’re collecting the data from, hundreds of thousands of sensors, but you’re connecting it with LoRa, if I’ve got that right. And then if there’s any backhaul or there’s edge processing and then using cellular to backhaul over the telecom networks. Have I got that right?

Simon Maselli:

Yeah. You’re correct. So, we’ve got two primary methods of collecting data.

So, a lot of the buildings that we enter into now, there’s no infrastructure in place. All the infrastructure is very old, and its legacy equipment. So, in those cases, what we do is we put a in building LoRaWAN system in place.

That aggregates all of the data, and then we transmit that data back over a cellular backhaul usually, to keep this separate from the client’s IT networks, and there’s various reasons for that.

And then in the other instance, there are buildings that are more modern and have, great connectivity, and they’ve got, protocols like BACnet and, RS405 or Modbus, in place. So, in those cases, we put an edge device in, in the system that works or connects to the, to the existing infrastructure. And, also, we send that back via cellular. So cellular in our world is very important for lots of reasons.

Nick Earle:

Alright. Thanks for that background. Let move now to the subject of the podcast, which is AI. And I must admit, I listen to a lot of podcasts. I’m sure you do listeners do.

And it’s kind of unusual now to listen to a podcast which doesn’t have AI in it. This the first time on the IoT leaders that we’ve introduced AI. And one of the reasons is that we felt that there really hasn’t been a clear view in IoT of what role AI will play. In fact, we just, released our, 2025 predictions report on our website.

And, we make three predictions, but one of them is 2025 is the year of AI for IoT.

And we go into a sort of a three-layer model on that. I think we’re going to get into that in terms of the work that that that you do. But before we do that, I think one of the feedback items that we get from people is, it’s like the early days of the Internet or the early days of networking. You hear all these terms. You hear all these acronyms.

People not everybody is following along with exactly what is exactly happening. And in the case of AI, it is moving so quickly. One of the stats that I remember is that the rate of change of AI is between 5-8 times per year. So, we thought Moore’s Law was, was incredible compound growth.

But when something is getting better at 5-8 times per year, you just do the maths and you suddenly realize, oh my god. The implications of this are enormous, and everyone is piling in billions of dollars. So, from a landscape point of view, how would you describe what are the different types of AI that you see out there? Because a lot of terms are thrown around.

It’s not just one thing, is it?

Simon Maselli:

No. It’s not. And you listen to Sam Altman and some of these guys. He says that we’re going to have AGI this year, general intelligence.

Nick Earle:

General intelligence. Yeah.

Simon Maselli:

It’s mind boggling. But look if I think, if we take a few steps backwards, before we go to the to the top end of the AI, what will we call it, stack AI has been around for a long time. And in the sixties, there were these theories around what AI will look like in the future. And there were two schools of thought that were kind of the most common back then. And one was this, terminology called decision trees, which were basically, a series of logic that if this then this sort of thing.

And that model, was one of the prevailing models, and the other was this, neural network model.

And neural networks, the reason that this one’s important is because that’s pretty much what every modern AI has been built on. But the neural network is essentially a whole heap of different layers, where each layer processes a little bit of information. And it’s called a neural network because it kind of works similar to your brain and the neurons in your brain. And sitting on top of the way that this AI works and makes its decisions and makes its outcomes is all of these different families.

And we’ve got these families like machine learning. Yeah? So, again, machine learning is something that’s been around for a long, long time. And, if I remember, I’ll come to a little story about it.

But machine learning again is something that’s come out of the sixties, and essentially it’s just math. And all AI is just math. Yeah. It’s just the way that we process things in and come up with numbers at the end of it.

So, machine learning is about processing these. It’s really great for time series data and the way that it can process these sets of data and come out of outcomes. And it’s good at doing things like finding anomalies and those sorts of things in data, making predictions and what the trend is looking like and finding trends and that sort of thing.

We have another family that’s pretty common in today’s AI world, which is machine vision.

Machine vision is essentially finding different features in a video or an image.

So, what it’s doing is it’s looking for patterns essentially in the image.

It is looking for things like color and contrast and some of those more basic things that sit lower in the neural network. But then when it gets to the top, it’s looking for features like shapes and sizes and those sorts of things. And that’s how that AI finds images and objects in videos. And you would have seen on the Internet, there’s literally thousands of clips of the little car running across the screen with a little box, bounding box around it, and that sort thing.

Nick Earle:

And the box around it. Yeah.

Simon Maselli:

Yeah. And it looks really impressive. Right? But, when it comes down to it, again, it’s just all it’s all numbers.

It’s just finding those features, finding those patterns in the in the data, and that’s what it’s coming out with. And then, probably the other common one is, natural language processing. So that’s kind of what ChatGPT is built on. And what this one is doing is finding patterns in language.

Yeah? And you go into ChatGPT, and you type something, what’s the most common thing at CES? And it’s going to give you back an answer, it’s going to have televisions and great audio.

Nick Earle:

I can tell you having just come back, I can tell you what it is. It’s sore feet. From traipsing up and down the strip. But, yeah, every I went into the hall just as an aside. We’re recording this the week after CES, and I went into the hall. One of the the several halls, there are a hundred and forty thousand people. It’s a small intimate affair.

And it was hard to find anybody who didn’t have the letters AI on their stand. I mean, suddenly, everything was AI and of course, it’s like there’s cloud washing. There’s now a lot of AI washing, which is why this important to lay it out. Suddenly, everything is AI, and we have to separate the truth from the marketing or whatever.

Simon Maselli:

I tell you what, I was there last year. I missed out on this year, but I was there last year, and I went through the Samsung stand. And every single Samsung product was rereleased, the same product as a year before, but it had AI stamping/sticker on it.

And when I was walking through there, I was really frustrated. That’s the word I’ll use on this podcast that, what’s this marketing hype, to pick up on your words? This all, BS.

But I come away from that show and I thought about it, and well actually they are using AI in all of these products because all of these products have machine learning in them. They’re processing data using these algorithms, which until recently weren’t really referred to as AI. But now that this marketing spins there and this marketing buzz is there, why not latch onto it?

So, I agree with you. I think it’s overhyped in that respect, but I don’t think that they’re actually out there blatantly twisting the truth. I think they’re telling the truth.

Nick Earle:

Yeah. I’ve been in the IT industry over forty years, and, Gartner classic hype cycle, is that that we are in the early stages of the hype cycle. But to your point, it is real.

And one of the reasons it’s real is that it is moving so quickly. I mean even yesterday – I don’t know whether you saw it. You’re in Australia and we’re in the UK.

Our prime minister who’s under a lot of pressure, he’s only about eight months into his tenure, but he’s grasped the AI flag and said, well this the economy is not good. The pound is not doing well.

The yields on the bonds are all awful. And but don’t worry. We’re going to put AI across everything, and the NHS is going to be transformed, now with AI. And there were lots of things on the on the TV today.

So, it’s not just the tech companies that are, sticking the AI label on it. And you’ve laid out the landscape there, and so what it shows, it kind of reinforces what I was saying earlier, is that if you’re a layman, but if you are just somebody who is like I am as a CEO of a company, and I’m thinking I’m looking at AI in terms of what can it do to our offering. We have about eight hundred IoT customers worldwide, and we have to do a lot of processing of data, and we make manual choices by looking at data.

So, can we automate some of that? But, also, just in terms of the admin, are there things that we do that we don’t have to do? And the way that you’re describing it, though, is that there’s a lot of different pieces, and there’s a lot of different layers.

We don’t have artificial general intelligence, AGI, although, apparently, it’s coming very, very quickly where the systems are more intelligent than the most intelligent human.

But there’s all these different tools, and there’s going to be more tools. There’s going to be more fragmentation. We know that because that’s what happens with technology. So is there a taxonomy that you can lay out, that where AI instead of the tools, because the tools will always change, is there, some in terms of a network topology, it seems like AI is going to be used at different points. I mean, we talked about edge processing. We’ve got the cloud.

Do you see AI being used in specific points in the network, and that will then lead people into choosing the right tool for the right job?

Simon Maselli:

So, I was talking a little bit about a solution that we’re very good at. That’s video analytics.

So, the way that a typical video analytics solution works for us is we’ve got the edge processing component.

So, at the edge, we’ve got a NVIDIA enabled device, GPU enabled device that does things like detects the objects in the video stream.

It does things like plots the path of where that particular object is going. It identifies counts across a line, all of the basic stuff that you expect a video analytics solution to be able to do for you. Yeah? And actually, there’s cameras out there now. The Bosch cameras are a great example that actually do that on a camera on a really lightweight chip. Yeah? So, these models are really lightweight.

They’re great for deploying out into the field, and they send their data in the device.

Nick Earle:

That’s AI at the edge.

It’s at the edge on the device.

Yeah. Taking the data back. So, it’s really the intelligent edge that we’ve been talking about.

Video is a great example, isn’t it? Because there’s so many data to analyze. It’s so complicated, the patterns, the anomaly detection, and now you’re seeing consumer products, household products that have AI embedded in them.

Simon Maselli:

Nick and then this the second part of this, though, and the bit that I was trying to get to is that we then take things like those paths and put them into a more powerful model in the cloud.

So, we take a path of a person, for example. We take the path of a vehicle.

We look at the angle that they’re going to intersect at. We look at the velocities of both of those paths, and we predict the likelihood and the consequence of that impact potential impact.

And we give the end user a really great likelihood and consequence of all of these near misses taking place on their road, or on their forklift in their warehouses and those sorts of things. So that’s stage two of the AI. And then what we do in stage three is we take that edge data, which is the counts and those sorts of things.

We take the more complex data, which is some of those things like near misses and, gender identification or, it’s face identification where we can identify employees or these sorts of things. And we give it some context and meaning, and we use a large language model to then, turn that complete scenario into a story for the end user. So, what happens is we’ve gone from a simple video feed, which is what it is when you’re looking at it and turning that into data that can be turned into a story that says, in the last month and I’ll give you a typical example.

We give a one particular customer a report that says in the last month, you had fourteen incidents with forklifts. Three of them were, near misses. They were A category which means there’s potential hospitalization or death injury.

Here’s what it looks like. And there’s a whole heap of people within that organization that love this data. The health and safety reps love it. The CEO loves this data.

Even the workers on the shop floor love this data because it’s really, really simple. It’s really, really digest. They don’t have to think about analyzing the data or looking through graphs or looking up their KPI chart for the month. It just lands in their inbox.

It’s got a little story around it that’s a paragraph long. If they want, they can click the report, and it expands the story using language processing or using a GPT engine.

And we’ve done something that historically would have taken, somebody a week or weeks to produce every month in a matter of seconds.

Nick Earle:

And, yeah, that is a good example because not only would it have taken them weeks or months, given the number of things has now exploded, the number of devices, the number of cameras in your example has now exploded, there’s no way you could even attempt, to apply that. You could analyze a few cameras for a few incidents, but if you’ve got five thousand, ten thousand cameras, you’re One Bangkok or that’s going to be in the thirty-eight buildings, I think you said. That you couldn’t even attempt it, therefore, you didn’t do it. As an example of a business process and a productivity improvement that is enabled by IoT, and it still leaves the human. You talked about what you do as a company doing buildings in Bangkok. I believe you’ve also got a pretty big, is it a water monitoring project, that you do as well?

Simon Maselli:

Yes. We do a number of water monitoring solutions, but I think the one you’re referring to is the water pollution solution.

Yeah. So that that’s a really interesting one as well because, the challenge we had with this one is that a typical solution would look like you would install some water monitoring devices.

You would set a limit. And if things went outside of that limit, you would flag an alert, and somebody would go down and have a look. But we had a particular organization that wanted to achieve that that sort of scenario, but their data was a lot more complex in that they were measuring underground waterways. So, these were sort of water pollution tunnels, or wastewater tunnels.

And the challenge was in different temperatures and different flows through that tunnel, the VAC levels changed a lot. So, the pollution levels were changing. So, what we had to do for this particular customer is a multimodal analysis. Even a normal flow, or a normal pollution level could potentially trigger that alarm in high flow.

So, during a rainfall event or there might be other factors that were triggering these alarms. And the end user had to report this data to the EPA. So, they’re reporting all of these false alarms, and the EPA was saying, “Hey. This a real problem. We need to we need to do something about it.”

So, the solution was to do a multimodal analysis of that data. So, we did a multimodal anomaly detection.

So, what we’re looking at is flows through the drain. We’re looking at those VOC levels, and we’re also looking at temperatures.

So, it meant that our alarm point was able to move up and down along with all of those parameters, so those three parameters together. And we reduced the false alarms by a long way. In fact, we’ve reduced it so much that they weren’t getting any alarms and then they were concerned. And all of a sudden, they had an alarm, so they said, oh, we’re going to go down and then investigate this alarm because we haven’t had one for a while. And what do you think they found? They found a major pollution event. They’re able to report it to the EPA, and they were able to get somebody, I don’t want to say in trouble because of our technology but go through the right process to make sure that this pollution event didn’t happen again.

Again, this was a multi-staged process as well. So, we did this processing at the edge.

We passed the data into the cloud. We did a secondary analysis, and we’ve got a patent on the way we do that, but kind of we’ve done a secondary analysis, looking at a bit more detail of the data. And then we turn this into an into a story using the LLM or the GPT, to give the end user this one paragraph email that says, “Hey. You’ve got a pollution event here. You better go and check it out straight away.” And these are the factors that we’ve picked up. Again, all driven by AI.

Nick Earle:

It’s interesting your taxonomy because we didn’t know each other prior to this podcast, but when I talked about our predictions report that we just released, which it’s our number one piece of downloaded collateral every year.

And there’s three, and one of them is on AI, as I said. But it’s exactly the same model. It is the edge.

There’s the application layer. And for us, the slightly different one is the network layer in terms of specific IoT. So, for us, the edge is the intelligence in the edge for the edge to make its own switching decisions because a lot of these devices are, multi-RAT. Just talking about our business, a little bit. So, you might have a LoRaWAN. You might have WiFi. You might have increasingly satellite.

You might have cellular. You might have multiple cellular signals. So, there’s a really complicated switching logic needed to be applied to which is the best connectivity for this device right now. So that’s edge. The video, the one that you say is exactly the same layer as us is the application layer.

And we see a big explosion in video at the edge. So, the amount of data, as yeah. We look I talked in a previous podcast at the cost of data in cellular has come down at about 9-10x in six years.

But, actually, the ARPUs, the average revenue per unit stayed flat. And what that shows is the amount of data that’s being transmitted. Think of vending machines now. It’s basically a big video screen. Reactive video screen, and then there is a bit of telemetry. So, it’s completely flipped. But the processing of the anomaly detection and everything needs that application level artificial intelligence because you’re looking at how someone is interacting with a vending machine or camera security, etcetera. So that is exactly the same.

For us, the third layer is slightly different. It is the network layer. So just in IoT, when the SIM becomes an eSIM and it can connect to any operator once you’ve broken that forty-year lock on the SIM. You then want to be able to say, I want the device needs to pick an operator, but it also needs help in that, can it sense the signal strengths or the latency of networks?

And then you have lightweight M2M, doing the artificial intelligence analysis, so you have long tail personalization of every device makes right connectivity decision based on not just the data that it’s seeing, but the network and from the sensors, but the network state. So, it sorts of sniffs the network, and we did a separate podcast on why for 5G implementation across multiple operators, you would have to have this three-layer model because everyone is implementing 5G in a different way. So, AI is kind of like the enabler of 5G to the edge, but it’s the same taxonomy model that that you’re talking about, and it makes a lot of sense on these AIs.

So, let’s, if we can in the time we’ve got left, just go into what’s changed and sort of conclusions.

I think if we look back, people have been analyzing data. I think you started off by saying since the sixties.

It’s not new. And I remember, about ten years ago, it was all about we’d been doing a podcast ten years ago. It was all about how what you actually need to do is hire data scientists. If your kids are going to university, get them to do data science because that’s where they’re going to earn a lot of money. We’re going to need loads and loads of data scientists, and they’re now the new, powerful people in the IT department.

This changes that model, doesn’t it?

Simon Maselli:

I think it does. I think it definitely does.

I think the role of the data scientist is definitely going to change in the years to come. So today, a data scientist is analyzing data. They’re wrangling different sets of data, putting it all together, trying to make sense of it, trying to find the insights, trying to do some of those pieces of work, trying to find the anomalies, trying to find the patterns.

And, when I talk about that word anomalies and patterns, I’ve used that a few times during our conversation because that’s what AI is really good at and can do it a lot faster than what a human can with a lot less mistakes.

So, I feel as though, particularly as AI matures, the role of a data scientist as we see it today is definitely going to change. Now do I think that data scientists will become obsolete and not a good, career choice?

I think I’m saying no the opposite. I think I’m saying that their role will change, and they will actually sit above the AI. They will be the ones that will be training these models, and they will be the ones that will have to reinforce what the AI is telling us.

And they will be doing some interpreting it.

Right, and let the AI do the grunt work. Yeah. Let the AI do the heavy lifting, if that makes sense of the data, yeah, the data analysis process, and let the data scientists focus a little bit more on the outcomes.

I think AI is going to become something a little bit like IoT was a few years ago. Yeah? So, if we think about IoT, I remember when I first started this company there were people coming to me saying, I’ve heard about this thing, IoT. We need to do IoT in our business.

And you sort of think about IoT then was “We need to find, something, here is this technology. Let’s try and find, something we can solve with it or try and find a solution. Yeah.”

And AI, I feel is at that space now. Yeah. We’ve got this great capability but the use cases haven’t really become as common as what the solution is, if that makes sense, or as commoditized what the solution is. But I feel as though once people start to figure out that here are the use cases that AI works really well for, that’s when we’re going to see an explosion in productivity.

And if you hear, the way Microsoft talks about AI at the moment, they’re talking about these agents. Yeah. So, AI will not become this great big Skynet.

AI will become all these agents that are that are doing their piece of work really, really well and enhancing what the what the human can do? Do all of that grunt work. Give it to the human to then make that decision for the things that, that are important, and maybe it’s community benefit. Maybe it’s, some other sorts of, solutions that AI isn’t really, really good at.

Nick Earle:

And you could argue that that that that that’s exactly if you go back to all technology. I was at CES and looking at Caterpillar, the big digger company. They don’t describe themselves as that, but they had this massive, great, big, open cast mine. I don’t know how the hell they got it into the hole.

This thing was absolutely enormous. You needed a ladder to get into the cab. But actually, when steam shovels were invented, then the work just changed. And people instead of doing the grunt work did different types of work in construction.

So, the work moved upwards. All the spreadsheets took the grunt work and the basic analysis out, and then people started analyzing the data and what you’ve just described. With AI, now we’re seeing incredible models. The difference is the amount of data that can be analyzed.

You’ve got that protein folding example where you can now predict protein structures in minutes that would have taken years.

So, we’re not talking about 1x, 2x, 3x better. We’re talking a 1000 or 10,000x better. And I think that’s one of the big differences is that now you can really start looking at examples like your video example where you couldn’t even think. It’s not just that you’re going to take an existing process and make it more efficient. There are things you can now take on that were unthinkable, which will create new opportunities, new growth, new competitiveness, new revenue streams for companies.

Simon Maselli:

Absolutely. And look, I think the driver of this has been a convergence of technologies.

Yeah? So, it’s only been the last few years that some of the hardware that we’ve got available to us has become mainstream. Yeah? So, the NVIDIA processing units, the Jetson units that drives pretty much every video analytics solution across the planet has come that far down in cost that it’s a commodity.

People can go out and buy one of these GPU processors for less than they pay for a Mac computer? So, yeah, this technology cost coming down and the computational power that’s available to us as well as these new methods in terms of the software that’s driving all of these things, as well as all of these other enabling technologies like the speed of cellular now and the speed of networks and these large data centers and these sorts of things. This what’s enabling this next wave of innovation that’s taking over.

And I was sort of saying that industry 4.0. So, we’ve had, industry 1, 2, 3. Now we’re at the edge of industry 4.0. So, we’ve had this industrial revolution. With industry 3, everything become automated, and machines become automated.

They’re able to do their own thing through robotics. And now industry 4.0 – the promise of industry 4.0 is that these systems become interconnected. Yeah? So, they’re not just operating on their own doing their own thing.

They’re looking at the entire process and giving you another level of efficiency again. And a company that are actually an investor in us. They’ve moved all of their operations out of their Australian mines. So, what they’ve done is they’ve put together a Malaysian command control center.

So, all of their Australian command control centers, their SCADA systems feedback to Malaysia, and that’s where they’ve got their operational personnel sitting there and monitoring all of their operations across Australia and across the world, by the way.

So, this this industry 4.0 at its early stages. But if you then bring AI into that fold and you take those operators away from having to sit there and watch a screen and get AI to start making a lot of those decisions, all of a sudden we’re starting to live in a very different world to what we live in today. And you apply that to any other industry, whether its driverless cars all talking back to some sort of central command control center. So, they’re all coordinated. The buildings that we do at One Bangkok, yeah, or across the local governments, all of those buildings doing all of their things and making their own decisions, on their own, but being interconnected and having complete awareness of one another and the impact of their decisions against another.

It really changes the world in which we which we live in, I guess.

Nick Earle:

It does change, and, actually, it has changed. And just a final comment on my trip to CES. I went in one of the holes, which is the one that has the big pieces of equipment. I talked about a big Caterpillar digger in there.

But the other thing they had on the Caterpillar stand was a little virtual reality headset, joystick, and chair. And actually, it was exactly what you just said. It was the ability to control Caterpillar products that were hundreds of miles away from an office environment. It’s a remote control, system capturing all of the data, across your fleet of construction equipment being controlled virtually, by one person sitting in front of a computer screen.

And they were demoing that at CES.

And then all the data and all the analytics and all of the video and you’re getting real time video back from the device. So, absolutely, the companies that are making the products are absolutely moving. I think industry 4.0, which has long been touted for many years, but not really realized, I think where AI could unlock it.

Just to finish, I guess in a sense it’s a rhetorical question. I think you’re going to say absolutely, but I do believe that what you’ve been talking about really does underline this.

Advice would be to people, you need a partner to do this. Right? The landscape’s changing.

There are so many different opportunities. There are so many different tools. There are so many areas where it’s not applicable. You need to look at the three-layered taxonomy, the architecture that we talked about.

I guess your overall conclusion would be, and it’s the business that you’re in, is that you need a partner to help you navigate through this landscape because it’s changing so rapidly.

Simon Maselli:

Yeah. Let me give you a bit of a long-winded answer. I’m not sure how much time we’ve got, but if you think about the AI maturity curve or even the IoT maturity curve. Yeah?

Every organization needs to step through that curve before they can realize the benefits. And the curve, for those who don’t know, sort of, starts at the bottom with some very basic early steps into whatever we’re talking about with low return on investment. And then when we get to the top end, it’s more advanced solutions, but there are higher returns on investment and those promises. We’ll just talk about industry 4.0.

And any organization in the past, let’s say IoT, would have been able to go through that journey on their own, make those mistakes, make those successes, get the organization to change, and adopt those new technologies, and walk through the landscape and eventually get to the to the higher stages of maturity so they get those promised returns on investment.

I think the challenge that we have with AI and all emerging technologies today is that the goalpost moves so quickly. So these higher-level returns are changing so rapidly that the speed that it takes an organization to learn and go through that maturity is slower than what they can get to the front end.

Nick Earle:

Like they’re chasing a bouncing ball down the stairs, and they can never catch it.

Simon Maselli:

Exactly. So the only way or the, most efficient way, rather than spending lots of money and putting lots of resource into it, is to partner with somebody that’s been there, walked the journey, knows what the end goal looks like so they can accelerate your progress through that maturity curve as quickly as possible and get you to that return on investment as quickly as possible. Because that’s what it’s all about for business. Right? Whether you’re in a local government that’s, spending rate payers’ money or you’re in a mining company or whatever it turns out to be, Unfortunately, money makes the world go round, and we need to get that return on investment to really change things. So, that’s what the goal that we’re all chasing is.

Nick Earle:

But that’s the business that you’re running, and that’s what you do. And your case studies are very, very relevant.

We did want to get into the what’s real, what’s not, and the architecture and the use cases. I think you’ve done a great job. Simon. I really appreciate it. I know it’s late for you.

You stayed in the office late in Melbourne for us, to allow me to record it during the office hours here, so I really appreciate that. And thank you, thank you so much for being our guest on the IoT Leaders Podcast. And if people do want to find out more about your company, it’s Minnovation, isn’t it? So, it’s like innovation with an M.

Simon Maselli:

Funny and funny is something. That’s how the name come about. So, we’re registered as a company called Metamorph.

And when I used to say Metamorph, the people used to say, how do how do I spell that? And I’d spend five minutes trying to try to spell it out with people. So, I thought I’ve got to change that. I’ve got to make that easier for people to find and spell and all those sorts of things.

So, I took the M out of Metamorph and thought, oh, we do some innovation stuff. Let’s make it Minnovation. Well, that actually created another headache or a whole another set of headaches because people were saying, is it m space innovation? Is it Minnovation?

Is it m dash innovate? I don’t know if the grass is always greener on the other side of the fence.

Nick Earle:

But I know there’s a few Minnovation out there because when I googled it, I found three or four and I had to so let’s give the full URL. What is the full URL, so they don’t get the wrong Minnovation if they look you up?

Simon Maselli:

Minnovation.com.au. We’re an Australian company.

Although we work, work through Southeast Asia, we’re based in Australia. So, look for that AU on the end.

Nick Earle:

Excellent. Alright, Simon. Thanks very much for educating everyone. Congratulations on your use cases. And on the basis that it’s changing 5-8x faster, every year, we might need to do another one in a year’s time because all of this will be completely out of date.

Simon Maselli:

I’ve enjoyed the conversation. Yeah. It’s great. Great to hear about your experience at CES as well. I love the place.

Nick Earle:

I’m glad it’s once a year.

Alright. I’ll talk to you soon. Take care. Thanks again. Bye.

Outro:

You’ve been listening to IoT Leaders, featuring top digitization leadership on the front lines of IoT.

We hope today’s episode has sparked new ideas and inspired your IoT and digital transformation plans. Thanks for listening. Until next time.