Can AI Fix IoT Adoption?

Transcript Ep 63 Afzal Mangal Intro: You’re tuned in to IoT and AI Leaders, your go-to show for insights, predictions, and big ideas on how IoT is reshaping the world of AI. Nick Earle: In this episode of IoT and AI Leaders, I’m going to be interviewing Afzal Mangal. Afzal will introduce himself, but essentially, he’s got a deep and long career in IoT, particularly coming from the product side. He has a Cisco networking background, but he’s also a thought leader in this space. He’s written a really good book about IoT and the issues around IoT adoption, which is called IoT: The Hype No One Knows About. What you’ll hear, what we covered, is the issues to do with IoT adoption, and how the technical integration challenges are holding people back, as well as the ignorance, in his words, of the fact that people do not realize it is all about the device. And so, we talk about that, but then what we do is we pivot into: does AI, and can AI, play a role in solving these issues? He has some very interesting views on what I call “turning the model upside down”, which is if we started with AI and then looked at IoT as a way of capturing data to feed AI, we actually could get the return on IoT that we’ve been after all these years, and much higher adoption across a much broader business and consumer audience. So, it’s quite a thought-provoking interview. We go back and forth. We don’t agree on everything during it, but it is a good debate about where IoT is going and the role of AI, and whether we should even be talking to the AI consultants rather than the IoT consultants in order to get products to market. So, I think you’ll really enjoy it. With that, I’ll hand you over to my podcast interview with Afzal Mangal on IoT and AI. Enjoy. Nick Earle: Welcome to the IoT and AI Leaders with me, your host, Nick Earle. This is the second episode that we’ve got in our new format, and for those of you watching on video, I’m sure you’ve noticed we are in a new location. We’ve now got an IoT and AI Leaders podcast studio, complete with lots of things in the picture, including a picture of our dog, Ollie. So, in case you’re wondering why the dog is in the picture, my guest today is Afzal Mangal. Afzal is based over in Holland. He is an expert on IoT. In fact, his LinkedIn profile, if I can quote from it, is “a builder of technical products, markets, and movements”, and then he says, somewhat modestly, “knows stuff about IoT”. So we’ve definitely got an IoT knowledgeable person here. But more importantly, he’s written a very, very good book that he’s just sold out of the second print run, and it’s called IoT: The Hype No One Knows About. So let’s get into this. Afzal, welcome to the podcast. Afzal Mangal: Thanks, and it’s an honour to be the first guest in your new studio. I really love it. I also like the IoT and AI Leaders. What is it? It’s a lighting board. Nick Earle: It’s a light box, yeah. Afzal Mangal: Yeah, yeah. Nick Earle: People who are listening to this won’t know what we’re talking about, but when you watch podcasts, you occasionally get a light box in the background, which you can order, and it just arrived, so we managed to plug it in in time. Great. So let’s get into this. First of all, I said you were based in Holland, but give me a little bit about your background, because the book goes into a lot of detail. We’ll get into some of that in the time we’ve got, but what is your background? Afzal Mangal: My background? You mean, from which perspective? Nick Earle: Well, from, let’s say, the last 20–30 years, your experience in IoT. Afzal Mangal: My experience in IoT. So I’m in telco tech since 2005. I started very young while I was still studying as an engineer, troubleshooting Cisco networks, designing Cisco networks for banks and healthcare. I’ve also been part of two startups. We were building their own cloud PBX infrastructure. You know the time when we were switching from those physical big PBXs? Nick Earle: Well, I actually was the other side of that coin. I just joined Cisco on the whole physical switches to the digital, IP-based. So I’m very familiar with that. You were on the implementation side. I was on the Cisco side. Afzal Mangal: Ah, yeah, okay, cool. So, you know everything about Cisco and probably maybe you also even have your CCNA yourself. Nick Earle: No, I wasn’t one of those. I went the sales, marketing, management route. So please don’t ask me about protocols on Cisco. I appreciate that. Afzal Mangal: But for me, it’s also a long time ago. And maybe I also have to tell you that I have two backgrounds. I studied network engineering but also marketing and communications. Nick Earle: Okay. Afzal Mangal: And it was in 2017. I had seen both corporates, large telcos, and startups building their own cloud PBX products. In 2017, I thought that now I want something where I can combine both things: network engineering and marketing communications. And I don’t want to work at a big telco anymore, delivering or maintaining traditional products. So I got an offer from T-Mobile Netherlands at that time. IoT was something new for them. They never did anything in IoT, because in the Netherlands, the M2M market, because we used to call IoT M2M, the M2M market was mainly owned by Vodafone and KPN in the Netherlands. So for T-Mobile, it was something new. They just bought and installed a network from Huawei, brand new 5G network, 2017. In the meantime, it’s Ericsson now, but in 2017 it was still Huawei. So they had the opportunity to be the first operator in the world to have a live narrowband IoT network, because it was just a software release. But they didn’t have a clue what to do with it. It was a fun time because it felt like giving a SIM card to a hardware developer for testing a new network felt like giving candy to toddlers. And because we were the only ones in the world, people from all over the world, they came to Holland, to Schiphol Airport, stayed there for just three hours to get a SIM card from me, send the “hello world”, and fly back to the country to tell their boss that it works. But I quickly saw that this didn’t fit into the standard B2B business of a telco, because the B2B platforms for SIM management were very old school. So it was very expensive and time-consuming to onboard a customer. You had to create contracts manually, upload them manually in the platform, onboard the customer, train the customer on how to use the SIM management platform, while most of our users were hardware developers ordering one SIM card, and we never knew if they would come back to order ten, let alone thousands. So I saw the need for an e-commerce platform where users were able to onboard themselves, so we didn’t have to worry anymore about people ordering small numbers. Something that I also learned quickly was that we’ve all been naive, because at the beginning, and not only at the beginning of narrowband IoT, but also the other low-power wide area technologies — LoRaWAN, for example, Sigfox — because we believed that these low-power wide area technologies would enable mass scale of IoT, right? And we said to each other, if this works, we are going to roll out hundreds of thousands of sensors in infrastructure, in building and construction, in healthcare. I even had system integrators in railways, for example, who came to me and said, “Hey, this is amazing. Now we finally can measure the temperature of railways remotely, which is very important during heat waves. We are going to order tens of thousands of devices.” But what we saw was that even though the pilots were successful, the people inside the organization, they were not ready for the change, because their jobs were about to become obsolete. Their tasks were about to change. And then I realized, okay, this needs a different approach. We need to focus on education, which is something that we collectively failed to do in the last decade, I think, all of us in the IoT industry. So yeah, currently I’m trying to solve the problem. I thought, you know, I’ve been complaining about this problem in two books, on LinkedIn, on stage, in podcasts very often, that we need to focus on market development, creating our future pipeline, getting people used to connecting the physical with the digital world. So currently I’m doing that in an alliance called Hello Things. But I’m talking so much right now, and it should be a podcast, so… Nick Earle: No, but it’s very relevant, because it explains why the book. I wanted just to double-click on the book. There’s a great foreword, which you didn’t write, it’s written by somebody else who’s a CEO of a company called The Things Industries. And your business is, as you say, called Hello Things. What the foreword essentially does is it talks about two types of ways of approaching IoT projects. And we’re going to start this as the path on the way to talking about the role that AI will play in all of this. But to start off first, you talked about in the foreword at least, it talks about 2015 as being the time things really started to take off. A lot of venture capital money came in. A lot of companies were founded. This is something we’ve talked a lot on the pod about, in its previous instantiation, just pure IoT Leaders, and about the fact that people would build silo technology. I mean, the whole industry is essentially silos. I’ve used the analogy previously of it’s like cars where you buy all the pieces separately and you have to assemble it on the path outside your house. And the operators — there’s 800 of them — you mentioned the operators and frustration, proprietary SIMs. Even the other companies, such as MVNOs, they were sort of stapling IMSIs together, but it really didn’t work, and there were a lot of claims in the industry. There was a great quote: the book talks about it doesn’t matter how much money you spend at it; you’re not going to make separate technologies easier simply by throwing money at it. And we all know that tech complexity at the components level is one of the big factors that held back IoT adoption. So, I just wanted to drill down. You cite multiple areas, but I want to drill down on two. Let’s start off first with the device, and then you go into a very interesting area on awareness. You talked about culture change and the potential threat. But first of all, on the device management level, you talk about the device really matters. What is your view on the importance of the device? Because, as you alluded to, on the consumer side of mobile, you don’t think about the device. Somebody else packages the device, somebody else tests it. You put the SIM in and off you go. You may have to buy another SIM when you arrive at Schiphol, but essentially the device works. IoT side, we know from our research that 95% of all IoT projects that get into difficulty — which is the majority of IoT projects — fail because ultimately when people look back, they said, “I didn’t pay enough attention to the device.” And in particular, the firmware, for example, in the device. Could you comment on that, with your perspective and history? Afzal Mangal: The thing is that most of the IoT projects fail. And where it hurts most is the device, because that’s the most expensive part of the solution in the development phase. But I think it’s one thing that we simply have to accept, because every IoT use case out there is innovation, and we know how it works with innovation. If there were many more people like Mark Zuckerberg who try to create a social media platform, thousands of them failed and one succeeded 15 years ago. The same applies to IoT. It’s simply something that we have to accept. We even need more devices to succeed with IoT. We need more people to fail. We need more device makers to fail. We need more devices to fail. But the thing with devices, I see a lot of stage talks these days of people claiming that IoT is not about the device, it is about… and then they start to promote their own service, which is not a device. But I think it really hurts if we downplay the position of the device. There is no debugger to find out if you position the antenna at the best place. You can only try it out, learn and fail. While you’re experimenting to create the best software, you’ve lost some time. But when you experimented and failed with hardware, you’ve also lost expensive components that you have to replace. The other thing is that if you look at the full stack of an IoT solution, everything is directly interchangeable with something else. Your SIM doesn’t work; you can put another SIM in it. Your software dashboard, where you look at the IoT device data, you don’t like it, you can integrate it into another platform or send the data to another platform. But if there is something wrong with the device, you have a big problem. Nick Earle: Let me jump in on that, because it is a subject, and regular listeners… If you are a first-time listener, let me just explain that this pod’s been going for five years. We just had the anniversary. We have talked about the device many, many times. What we have found as a company that I’m the chairman of, Eseye, is that we won’t take a project on unless we can do an audit and a remote check on the firmware in the device. And frankly, that means that a lot of people say, “Oh, nobody else asks me to do that. I just want a SIM, and I want a data plan, I want pricing.” And we say no. Unless we can do a full check on your firmware, we’re not going to do it. Because we sell global connectivity, 100% global connectivity, agnostic to the operator. The stat that’s interesting is that it takes them about two years to fail. You talk about time. It takes them about two years to fail, and they always come back. And when they come back, they say, you know what… they phrase it differently, but essentially, they’re saying, “You were right.” Sometimes it’s a different person that comes back, but you were right, and I should have looked at the device. Because it’s the dance between the SIM, the modem, and the processor. The reason technically that’s so important is that if you are going to get rid of the silos and do agnostic operator IoT with eSIMs and new standards like SGP.32, you have to change the IMSI in the device, which means you have to integrate in a different way. The bottom line, without getting too much into it, is the firmware in the device, which most people thought we’d left behind, because in the mobile phone you never think about your firmware. The firmware in the device is the biggest single determinant on the success of the IoT project. So, we have about 700 customers. Every single one of the devices that those customers have implemented, we’ve had our hands on the firmware, and if necessary, made suggestions or done the work ourselves to do it. So, I think that’s a good example of why — it’s a very practical proof — of why the device matters. But it also raises another issue, which takes us on to the second area, which is to do with change management. When we say, from our perspective, when Eseye, we say we need to look at your device, people say nobody else says that. “I’m just going to start my project.” And I talked about what then happens. But there’s an awareness issue here. Typically, it’s product people trying to go quickly who don’t know about communications and firmware and hardware. But secondly, there’s also an awareness at the business level as well. So, it’s not just a technical issue, is it? One of the reasons that IoT adoption has been held back is to do with cultural issues and awareness issues. I know you also cover that in the book, so maybe you could just give your view on that. Afzal Mangal: Yeah. Before going to that, I just want to say, and also ask, if I may — because you’re actually the interviewer here — but it’s kind of unique that you are getting into the firmware as a company whose core service is connectivity, and I really like it. But how do they respond? Nick Earle: That was perfectly fine to ask me questions. It makes it a real podcast. So we train all of our biz dev and our first response back to prospects to say, this is the way we work. We want to look at your device. If people say, “No, I want pricing”, we say, “Thank you, good luck.” We don’t say “see you later”, but we know we will. If they say why, we explain. It does mean a delay in the project, so it’s going slow to go fast, because it depends on how deep you’ve got to go into the device. We can remotely do some scenarios and some tests — try and break the device. Will it switch? Let me give you a real example. We’re the platform for AT&T for their new business globally. And AT&T customers have had typically just an AT&T SIM. Well, if you’re now going to do global, the firmware in those devices has to be optimized. So we’ve got a set of tools and an in-device application. I think this is the real answer to your question. We have an in-device application that sits in the firmware, runs on the processor, which actually is like the intelligent switch. In the past, the switch was in the MNO. The MNO said, “You connect to my network”, like Vodafone, “connect to my network”, or I will choose the roaming agreement for you. So that’s essentially… the MVNOs said, no, the switch is abstracted into the MVNO cloud platform, which is we will do the switching, but just between one or two MVNOs. Our view is that the switch has to be resident in the device. The intelligence has to be from the device back. So our switching software works like a client-server model where the device calls the shots, and then the orchestration engine — eSIM orchestration — is done in the cloud. So that’s the way it works. Some people say, absolutely, I get it. Some people say, that sounds complicated, I’ll go somewhere else. It’s a strange way of selling, because what it says is unless you agree with the way I do things, I want you to go somewhere else. It takes a lot of training for the sales force. Afzal Mangal: That’s why I was curious, because customers are usually stubborn, and if your customer is a developer, then they’re even more stubborn. Nick Earle: And you know what? We pride ourselves on qualifying out. Qualifying out is one of the biggest victories, because it means that we’re not going to waste their time and ours. So you’ve talked in the book about awareness and cultural change. What’s your experience of IoT in general of those two subjects? And then after that, I want to get into some very interesting thoughts I know you’ve got around the role that AI can play in addressing effective IoT implementations. We’ll come back to that. But in your experience of IoT projects, and you’re coming from the device side — you’re very technical — but also sales and marketing, it’s a good combo. What have you seen in case studies around the role that awareness and other issues have made, like culture change resistance, in holding back IoT projects? Afzal Mangal: It’s easy, but it’s also crazy what I’m going to say right now. I can’t believe it myself because I’m in IoT for more or less ten years now. How is it possible that ten years ago many people in this IoT industry were talking about and working on, for example, smart waste management solutions, the fill-level sensors in waste bins. Nick Earle: Oh yeah, that was going to be ubiquitous everywhere. Every waste bin would have a little sensor in it, et cetera, et cetera. Afzal Mangal: And it is happening, but slowly. The thing that I can’t believe myself when I say it is that most of the waste management departments in the municipality offices all over the world still don’t know about the solution. That’s crazy. Why are there so many powerful companies in IoT with huge brands, with a lot of reach — AWS, Vodafone, Deutsche Telekom, IBM, Ericsson, Nokia — they have all been involved in these solutions, and this is just one of many examples. But still after ten years, or maybe after fifteen — because when I say ten, it’s just because I’m in this for ten years — why is it that most of our target users, most of our customers, are still not aware about solutions that already exist and are already proven? I know you said that we will get to AI, and we are going to talk about AI from a technology perspective, but let’s look at it for a moment from another perspective: the adoption perspective. I always use my mom as an example. Why does she know exactly what AI is and what it can do? Because she knows that she can use an LLM if she needs to write a difficult letter. When she had a car accident and she was dealing with lawyers, she knew that she could use ChatGPT for that. She knew that she was using AI, that there is AI under the hood of ChatGPT. She knew that when she had to design a birthday invitation for my daughter, she could use GenAI for that. Yeah, my mom. But while she is using her smartphone to manage her climate system in the living room, she doesn’t know that she is using IoT. While she is using her phone to manage the lighting in the living room, she doesn’t know. She has that because my brother took care of it, that everything in the living room is smart, but she doesn’t know that it is IoT under the hood. While she does know when she uses AI solutions that there is AI under the hood. I use my mom because I want to talk about the mainstream. She’s not into tech at all. She works in healthcare. The mainstream is still not used to the idea of connecting the physical world to the digital world, even though they’re already using it. They’re still not used to it. So she’s not triggered to think about more ways of using this concept of IoT in her life or at her work. With AI, it’s completely different. She already expects that one day her boss will come and say, “Hey, this job, what you’re doing right now behind your desk in healthcare, like planning appointments with patients, we are going to do this in a smarter way. We are going to use AI automation for that.” She already expects this. So she’s already preparing for that. Like, okay, then I’m going to learn new things and I’m going to do different things. But she doesn’t see any IoT solution coming in the business. So for that, she’s unprepared, and the moment when it comes, she’s going to resist. Nick Earle: Let’s take a step back. That’s interesting. We all know that IoT adoption has been slow. You talked about 2015. In 2011, I was at Cisco, as I said, but Cisco, IBM, many others, including Ericsson, put out this big report that there would be 50 billion things connected by 2020. Life was going to be great. And actually we got to about ten or eleven. So complete failure. We still have a lot of issues, despite standards like SGP.32 and new capabilities, and we’ve learned a lot about devices, but not everyone has swallowed the Kool-Aid, if you like, on why the device is important, as you pointed out. So although IoT adoption by the numbers is growing, it’s still got a lot of issues. What you’ve said, which I found interesting, was: let’s contrast that with AI. Because for IoT to be useful, it’s the businesspeople — it’s your mom and your mom’s bin — it’s the lamppost in the street. And ChatGPT, I was just reading, they’re just approaching a billion users, and it’s been three years. The reason we expanded this podcast from IoT to IoT and AI is that we’re basically saying that the AI models — LLMs, which feed the agents — have sucked up all of the written data, the visual data, the audio data. They’ve sucked up 90% of it, so they’ve trained the models. We won’t get into whether they legally sucked all that up, but they have got it, they captured it. But there’s 50 times more potential — 50 to 100 times more potential — data in things than there is in data that’s stored on the internet. And the benefits of being able to get all the insights, and it’s called sentient IoT from things, is absolutely enormous. So here we are having failed in the first wave of IoT because we haven’t got the adoption. We’re getting better each year, but we haven’t got the adoption. And now we’re saying, oh, and there’s going to be something that’s 50 times bigger. Let’s go again. What I think you are saying is the integration between AI and IoT will actually help to address those issues, because people will say, “Oh, AI is collecting data from the things,” as opposed to there’s an IoT project and there’s an AI project and they’re separate things. So I think you’re optimistic, and I just want to check that’s what you’re saying. You’re optimistic that the convergence of IoT and AI over the next few years is actually going to help address some of the core problems that you’ve seen in IoT adoption. Afzal Mangal: Yeah, and I really love the idea. I really hope that’s going to happen. I just wonder… most of the companies… for example, I know a company that is into sound sensors with IoT. We’ve discussed that one before, but that was from the adoption perspective. Now we are talking about IoT combining with AI. So in the existing version, it just measures the noise level, and it sends you an alert when the level is too high. When it reaches a certain level that we don’t like, it creates an alert. If you include AI into the same device, into the same IoT sound device, the alert will contain more information. It will tell you that, “Hey, there is too much noise there and people are having a fight,” or “someone is in danger because I heard someone screaming,” or “there is too much noise, but people were just having fun and laughing.” So there is a big difference. And then, in the third situation, nothing is wrong, so you don’t have to act upon that alert. This is a great example of using AI and IoT solution. I just wonder, this company that offers these IoT sound sensors, do they still have time and money for that? Because this first version of their solution without AI, they’ve been working on that for the last five years, and every year they try to survive with one or two customers. And now bringing AI into the same solution… Nick Earle: …is… Afzal Mangal: …that the salvation? Time and money? So I’m sure that if they can add AI to the solution and put it in their marketing stories, they will create more awareness, attract more users, attract more customers. But is there still time? Nick Earle: And is there money? I totally agree with that. The IoT landscape is full of a lot of companies that are losing a lot of money, and significant consolidation. About four or five pods ago, we had somebody on, I think they were based in Turkey, and they were in healthcare. So I’m going to give you an example back. We all know that healthcare has been one of the areas where everybody thought IoT would really take off, and it’s really been very tough to get healthcare going. Although it wasn’t a technical issue — you can put sensors on everything. You can put sensors on doctors and nurses to find out where they are. You can put sensors on patients. But it still didn’t take off. What this company did was they used AI to create a series of agents that represented horizontal processes for hospitals. So they came from AI back into IoT. I would encourage people to listen to this — it got great feedback — the healthcare pod. They started off with a business procedure. Let’s take a procedure like coming in for an x-ray. Patients, when they come in, they get a wristband. It was something similar to what you get at a Coldplay concert, by the way. They use the same manufacturer, I think. But then every piece of equipment has a little sensor on. And then every person working at the hospital has a tag on. So they’ve got staff, they’ve got things, and they’ve got customers. That would normally just be an IoT project, and it might fizzle out because nobody sees any benefit. But what they do is they don’t present it as an IoT project. They present it as: what is the cycle time for the coming-in-for-an-x-ray process, and where are the blockages? For instance, they did it in A&E — accident and emergency — and they said one of the reasons people are spending eight, ten, eleven hours in A&E is that they get triaged quickly, and then they wait and wait and wait. Even after they’ve seen the doctor, they go back and sit back on the plastic chair with everybody else, and they wait for the medication to come to them. So by highlighting the process inefficiencies with the AI agents, which are all talking to each other based on being fed by the data on things, they actually create a dashboard for the manager of the healthcare institution — the hospital or the clinic or whatever — to say: if you improve these parts of your business processes, then your hospital becomes more efficient and you get to see more patients. That is something that is very successful in the Middle East and in Europe, and then just heading into the US. So I think that’s what you’re saying. Should we even think of it the other way up? Is the way to solve all of these things to take an AI-first approach, where IoT feeds it, and think about agents and think about use cases, as opposed to thinking about technology? Afzal Mangal: This is a great example, and thanks for giving me a new perspective. So, this started with AI, then they added IoT later, right? Yes. But here comes pessimistic Afzal again. Is the community, or the movement of people who are making, developing, implementing AI solutions today, are they aware about the possibilities around…? Nick Earle: Well, in this case, no, they’re not. In this case, by the way, they have aggregation devices around the hospitals to collect the data. This company, whose name escapes me, forgive me if they’re listening, but it is available on our podcast list. This company actually goes in and installs the sensors. They’re just little sticky labels that go on things in the hospital. Patients get the wristband when they check in, and staff will wear a clip-on thing on their uniform. So it’s not a big technical problem. But you’re right in terms of your question. I’m in the UK, you’re in Holland. I don’t know any hospital in the UK that does that. You may say, I’ve never heard of that in Holland. It’s interesting that their customers were… they went to the Middle East, and they’re now getting some customers in the US. So there’s still the same adoption challenges, but they are approaching it a different way. For me, one of the big A-HAs in this episode of the podcast is: can AI solve the IoT awareness and adoption challenge? And I want to ask you, because of the limited time we’ve got left, doesn’t AI potentially solve a third challenge? Let’s go back to assembling the car on your path from all the different pieces. It would be a nightmare. You have to buy them, source them, which is what goes on today. But you also then have to integrate them. You talked about the portal, which is one level of abstraction, but given the power of AI and agents to link it into business processes, do you believe AI could actually solve the integration of all of the data, which we’re trying to do at the lower level? Connect this to connect that, and then an API to this, and if this goes here, send it to that, and it goes into that application — and why isn’t it in that application? Could some of the logic of the integration — because it’ll be multiple sources of data, billions of things — couldn’t the intelligence and the integration of it all, the big data warehouse, be another benefit of AI that would abstract it into a higher layer that doesn’t have to be physically programmed? Because it can use artificial intelligence to say, “I want you to assemble the data this way.” Do you see AI playing a role in creating better insight based on integration of all the hundreds and thousands, potentially millions of data points? Afzal Mangal: I believe in this. I think it’s possible, but I just think that there is still a human being that needs to create a connection. Like in a hospital that a doctor is using a certain AI dashboard, maybe even talking to an AI agent about patients, and the AI agent in the platform is getting data from medication that the patient has used in the past couple of months, doctor’s appointments. We would still need someone who says, okay, but now go out there and take data from more points than only the patient data and the medication data. One funny thing is — and I know that AI is much more than an LLM — but let’s take the LLMs as an example because the knowledge that the LLMs have reflects the general knowledge of human beings of the physical world, right? Recently, a month or two ago, I asked Gemini and Claude and ChatGPT a question: I’m a healthcare worker and this is my role. Can you give me examples how I can work more efficiently if I use digitization? I did the same exercise for construction and for farming and city development. All three LLMs came back with a lot of AI automation use cases, but there were no IoT use cases at all. Nick Earle: Exactly, and that’s my point. By the way, I was going to ask you: did they all come back with the same answers? Because that’s another issue as well. Sometimes if you ask the same question twice, the answers are different. The two points you’ve made there: the human still needs to be involved, that’s why it has to be augmented. And the data has to be… it’s only as good as the training, it’s only as good as the data. That brings us back to where we started, which is: is the next thing for IoT to be sentient IoT in an AI world? Does AI get more relevant touching the issues that we’ve covered in the last 45 minutes or so? Is it easier for people to be aware of it, not just the product department? Will it get wider adoption? Will it solve some of those integration issues, the plumbing, the connecting, the integration issues which slow down projects, and the technical issues which slow down projects? But if you can connect everything back to Things — if information about a Thing is more valuable than the Thing itself — if you can connect everything and feed it in as a new form of training to the AI model, yes, there’s still going to be a lot of issues, but you are suddenly able to create new insights, which then get interpreted by humans. A lot of people are talking about that, and its early days. There are just a few case studies of people doing it. But it comes right back to your point about the device, because unless you can get the data out of the device, you can’t get the data into the abstraction layer. Therefore, you can’t get the insight. Therefore, you can’t get the business value. So, I don’t think this is going to make the device issues go away. If anything, I think it’s going to make issues to do with device and basic connectivity even more important. Afzal Mangal: Yeah. Something that I realize now because of this conversation, this great chat, is that we have a lot of IoT solution providers, IoT integrators, with devices, with connectivity, talking directly to the operation teams of hospitals. Maybe we should change that. Maybe they should go after the people like the AI consultants who are talking to the hospitals. That’s something that we should do. Nick Earle: Well, you know what, that’s probably a prompt for the next podcast. But we probably got to the end of this one, because the moment the light goes on, as you just said, the moment you think about it this way, you suddenly start thinking. Like I said, we turned the model upside down. Should we start off with AI and back into IoT? Listen, the purpose of going from IoT Leaders to IoT and AI Leaders is to exactly put some of these questions on the table and have a debate with thought leaders. I think that’s what we’ve achieved in the pod. So, what I wanted to do is wrap it up. First of all, I want to thank you. If people are interested, I’ve mentioned the book. I believe your website is AfzalMangal.com. I’m going to spell it; make sure I’ve got it right: A-F-Z-A-L M-A-N-G-A-L dot com. And your book, which you’ve sold out, and you need to go into another print run, is IoT: The Hype No One Knows About. Thank you for being on the pod, for pushing back, for asking questions. Let’s see where this thing goes. Maybe there’ll be a second episode, maybe the next book. We’ve started to get some material for you to think about the next book and the AI element. Afzal Mangal: Thanks for having me. Nick Earle: You’re welcome. Thank you for being the guest on the podcast in the new podcast studio, if you are watching this, and thanks for listening to IoT and AI Leaders. Outro: You’ve been listening to IoT and AI Leaders. We hope today’s insights help you drive smarter, faster business innovation with IoT and AI at the centre. Thanks for listening. Until next time.