Podcast #29: Charting the Rise of Industrial AI - by Ganesh Bell

Good day everyone, this is Ed Maguire, Insights Partner at Momenta Partners, with another episode of our Edge Podcast and today we have a special guest, Ganesh Bell who is the president of Uptake, and he’s actually somebody I had the pleasure of meeting several years ago in one of his prior roles, and we had a conversation that started on technology, and boy, Ganesh is one of the most interesting people I remember speaking to. Interests that cover a whole range of areas in technology, life, and the economy, it was such a thrill to see that Ganesh is now in his relatively new role at Uptake, focused on connected industries. Ganesh, it’s great to have you here. 

Ed, thank you for having me on the show and it should be fun. 

First, I’d like to get a sense of what’s shaped your view of the industrial Internet of Things, and what are the experiences that have brought you to your current role? 

Good question. I could go back almost a decade ago when cloud was becoming mainstream, and SAP was mainstream, we were starting to use the term big data, and I was at one of the biggest enterprise companies in the world, SAP. I used to talk to my colleagues about the fact that all these technologies being democratized just means that there will be more IP, more XP in the world, they’ll be codified in software. In fact, I would talk to people about the next big software company is in a software business, and people go, ‘What the heck does that mean? You work in one of the biggest software companies in the world’. I would say exactly that, which is ideas and IT and deep domain expertise will be codifying software, so with Saas, and cloud, and big data, there would be others that will be doing it, and now we talk about it as digital transformation of industries. 

What led me here is, the more I talked about it, I had a moment where I actually believed my own bullshit, when I got a call from GE that they want to do something with software, and I sold them on digital and became the first Chief Digital Officer in the company, we built one of the first digital businesses, and realized that this is a much bigger market opportunity than I originally even thought. And now I’m doing it at Uptake with great customers who are managing the connected industries that they are in. 

That’s amazing and what a ride as well. I think when we were first talking you had been working on analytic strategy, looking at harnessing the value of data for business impact. It would be interesting to get a sense of what are some events, or people, that have had the most impact on your thinking? 

Going back to when I was talking about the next big software company, is in the sort of business where I said the companies that use software and build businesses, but fundamentally they won’t be in the business of selling software; I didn’t have great words to say it, so I think probably one of the most defining pieces of our time is Marc Andreesen’s, ‘Software is eating the world’, that was a simplification of the idea and great explanation of what was happening. Now we have examples in the consumer world, new software company, big start software company, Amazon software company, but none of them sell software.  

But when you take that over to business, I think probably going into the industrial world where I didn’t have a lot of appreciation for how grunging industry was about five or six years ago, going into GE and meeting with amazing customers. I would say they probably shaped my view of the transformation and the challenge that they have to go through, and it’s become much more than technology. Yes, it is platforms and applications, intuitive industry data and content, but it’s also about culture, change management, transformation, and new ways to work. I would say the journey that I’ve been part of with Uptake and my past life with customers always shaped it more the belief in the market. But I would say I would go back to the article by Marc Andreesen, I think that is the future of business at least for the next decade or so, where pretty much every industry, every business will be reimagined with software. 

Yes, it’s amazing, the ramifications are just so broad-reaching and profound. In a recent Forbes piece, you were claiming that enterprise software as we’d known in the past, is over, and business process automation is no longer good enough. Could you expand on that and share a little bit more about your point of view there? 

Sure, I’ve worked in enterprise software all my career. I’ve been an entrepreneur, I’ve cleared my own CRM product which launched a company, worked in some of the biggest enterprise software companies like JDR Wood, PeopleSoft, SAPs, and now I’m a lead entrepreneur at Uptake. When you look at the history of software and you’ve seen many people talk about it, even acts like a replacement market, every time a new technology comes people say it’s a replacement market. What does that mean? They’re basically redoing the same thing in a new technology.  

We went from mainframe to distribute computing, to client server, to Web, to ASP, to SAP, to cloud, and across the last several decades of enterprise software we’ve literally redone the same business profit in a new technology. Yes, it’s gotten more useful, the user-experience is better, there’s better and better analytics, there is democratization of actors, the software is more configurable, consumable, extensible, all that as it happened, but fundamentally the design of ERP or CRM, or any of these business systems are still the same. 

I also think about all of enterprise software throughout the history has been about humans entering data. We’ve built technology for just humans to input data, ERP is just that, CRM is just that, that’s what SAP, Oracle, SalesForce do really, really well. But we’re now at a stage, especially in the industrial world where machines generate more in actual data than humans, and the new kinds of sensors like drones, the aerial sector, or robotic work, or humans with augmented machines generating more data, so you almost need a new architecture on how you look at the data. At the same time, we now have the new technology, not just cloud, but in the last four or five years the amount of progress was made in machine learning and AI, forces us to reimagine the software, not just redo the same thing. It was okay to redo CRM from clients over to web, through web to SAP, to different companies, it’s no longer good enough. Customers want much more than just different sorts of automation, they want outcomes. 

So, I think there’s desire from customers for outcomes, the evolution for AI and machine-learning is fundamentally reimagined and augment human intelligence, and all the cloud native technologies forces to think about enterprise software in a very different way. That is the belief that we have at Uptake which is, customers need outcomes and how do you take all the data that you have; if you ask most CXO’s, ‘Do you trust that data has strategic assets’ and they’ll say yes now, because they’re reading something, they have a belief in it. But if you ask them, ‘What do you do about it?’ they don’t have great answers. And how you take all from the assets of data to outcomes is an idea that we’re working on at Uptake. 

So, the evolution of AI in machine learning, as we’ve seen, an enormous amount of progress, break-throughs and acceleration, but how do you see that driving the evolution of what we call industrial IoT? 

In all the industries that we’re talking about, industries like power, energy, oil and gas, manufacturing, construction, mining, transportation – whether its rail, locomotives, aviation, or trucks, all these industries have operated with some level of software support, and some level of analytics. If you go to a powerplant or oil field they all have some analytics, but they’re all based on first principle statistic analytics, which are important, but not good enough, because they’re operating on a slice of operation data. In almost all these scenarios we see about less than 1 percent of data is utilized, and through really processing all that data you need something that takes into all the domain knowledge that orchestrates a new level of line. That’s really what we’re seeing now is, machine learning is we’re able to do that now. We’re now able to build engines that detects failure, or predicts anomalies, or manage optimization of fuel and so on, and you can do that with machine learning, and you can use all your operation data. 

At Uptake for example, we build some of these engines that utilize more than 1.2 billion hours of operating data, so, they get better at prediction. So, we’re now seeing the ability to do this across many classes of assets, and across systems and systems of assets. 

I’ve seen Uptake described as being an industrial AI software company, how does that tie into the work that you guys are doing with customers? Could you expand a bit on that definition? 

Sure, if you go back to your previous question on IoT, and if you think about it in the context of industrial systems; there was a first wave of IoT and if you look at it, it was just people making connected things, ‘Can you make a product connected?’, and that happened in the consumer world, it happened in… if you made a thermostat, how do you make a connected thermostat. If you make a car, how do you make a connected car? A lot of those companies got acquired by the likes of Cisco, Intel, and everybody else, and that’s good in those industries. Whereas in an industrial world these assets that we’re talking about, like wind turbine, or gas turbine, jet-engine, or a robot in the manufacturing assembly line have always been connected, except there are silos of information and data. 

What we’re doing at Uptake is now applying not just the notion of connected IoT applications, on top of these assets, but also the analytics and domain in terms of how these assets are supposed to go work. For example, we work with Berkshire Hathaway Energy, all their wind turbines are connected to the Uptake cloud and a series of machine learning engines. Because of that we can now predict many different cellular modes, and improve unplanned downtime, and then they were improving annual energy production which is more energy from the same wind. We’re just doing that in locomotives, in transportation, where about 6,000 locomotives of Progress Rail are connected to our cloud, where we can help them to detect more than 90 percent of the failure modes.  

That’s the same engines, the platform, and the application are also useful in the context of another customer we just signed which is the US Army, where we’re now able to deploy across all the five Bradley fighting vehicles, whereby detecting failures and anomalies we can improve uptime, or as the army calls it, readiness of their fighting vehicles. In their words it calculates to billions of dollars when its applied across all their fleet. So, we’ve just started on that journey, and have seen many such examples in other industries like trucking, whether it’s installing a new robot in the manufacturing line. All these ideas can be optimized further, because we can augment all the human domain expertise with machine-learning domain expertise and help in decision-making over time. 

It sounds like Uptake is very much focused on a broad range of industries, and applying analytics to solve very real pain points and business problems. But what about the areas of product focus, what are some of the key either technologies or offerings that are central in your focus right now? 

If you look at the industrial landscape, we see a whole new stack that is emerging. In enterprise software we see this moving forward. In the nineties there was just database servers and applications, that’s it? Then you saw the rise of things like application servers, integration servers, business profit management servers, master-data management servers, portal servers, and so-on and so-forth. We’re seeing the same thing happen in the industrial world where in the past it was just machines, and controlled systems. Then we saw the rise of, hey, we need to connect things and do simple asset management applications, or monitoring and diagnostic applications, but now you see the rise of the industrial data layer. You see the rise of edge computing. You see the rise of data conceptualization, how do you switch all the thousands and thousands of tags or sensors of data into a model that is understandable, you call that data contextualization, and its ingesting all this high-volume data. 

Then there are application development layers, and analytical layers, and application layers and so-on. We think of all these things in a simple construct of platforms, applications and curated content. At Uptake we provide is tagged on that we say you cannot build a simple application development platform, and then IoT platform, and then add machine learning and analytics to it. You’ve got to have machine learning and AI as a first-class core citizen, and that is really what our platform is; we built a series of data science and things on top of which we built an IoT programming model, so that platform allows our customers to build great applications. 

On top of that same platform we built our killer application, our performance management, or optimizing service and parts for managing dealer networks in construction and mining. Then on top of that we have creative industry data, in that creative industry data our focus is building a repository database, we call this our asset strategy library where there’s more than 800 critical asset types, with about 55,000 failure modes, and about 178,000 reportable working conditions, and 10 million components, over 32,000 human working years of experience, all curated. So, what does that allow us to do? It allows to go all the way from connectivity of machines, to data integrity, to having the best platform for machine learning and AI that informs our applications, along with all the physics data analytics, and domain knowledge to predict outcomes in those industries. 

So, we think of it as platforms, applications, and curated content, and we’re doing this in every one of the industries that we’re in. 

So, the curated content dimensions of your offerings are something that really sets Uptake apart. Could you talk about the role of domain expertise industry knowledge, driving the success of your offerings, and how the process of developing curated content as it were, how that creates real differentiation that translates to value for your customers? 

A lot of people would agree with you that domain matters in an industrial world, and we’ve seen the debate, we’ve seen new upstarts, or even horizontal technology, even established enterprise software vendors sometimes do lip-service to verticalization. They’ll say, ‘Domain really doesn’t matter, we can extend our horizontal applications towards a particular industry’, and at the other end of the spectrum you see OEMs, OEMs are big machines in the industrial space, argue a lot about domain not matters. We agree domain absolutely matters, but at Uptake we believe the domain of operating the machines and the assets in the business matters a lot, more than just the notion of having manufactured the machine, or the material side of the machine. They matter, but operation data matters a lot more. 

As I use examples sometimes, I tell people, ‘If you’re building an autonomous car, yes, the domain of driving matters, the domain of building an automobile matters, but what matters a lot is experience and knowledge of having driven a car on public roads for millions of miles. So, if you apply the same thing in the industrial world, we think of operational data, all the physics data analytics matters, all the domain knowledge of building a machine and assets matter, and operating the environment matters, and building software matters, but what matters a lot is also all of the operating data which is what our customers have. 

So, being able to go into it and curating it is an important aspect of delivering outcomes, which is why we acquired asset performance technologies, which is the database that I talked about, and we’ve added to it, we’ve added all the elements of… or approaching almost thousands of critical assets across all these industries. But in addition to that, we’re also looking at other external data factors, like ambient data, satellite data, weather data; and people go, ‘Why would Uptake care about weather data, doesn’t everybody else do weather? And it’s like, ‘Absolutely, the weather for humans and other industries in general. But when we do weather, we take into account six different weather sources, and we predict the freezing point of a particular point on a railway track, so we can understand slippage, and correlate that to failure detection on the locomotive. Or, we understand weather as in wind-speed at the ground level, or the mast level, or the rotor level for a wind turbine performance curve, and performance prediction.  

So, you can look at lots of different data sources in industry similarly, and start building on top of profit data streams a unique insight that would inform you about how to operate, or not operate, a particular environment or set of assets. They all feed into things like improving over time, going from just reliable to uptime, to optimizing apps and services to cars, to profitability of the customer. 

Can you talk a bit about some of the technology aspects for industrial AI adopters? As you look around at the customers you’ve been working with, and from your experience, are there some hurdles or enabling technologies that you think aren’t fully appreciated, either as creating additional challenges, or offering the leverage to potential adopters? 

It’s a great question, again, we’re early in the days of what we think is one of the biggest, arguably the largest enterprise software market ever. The biggest hurdle that we see in value creation for our customers is complacency. Let’s talk about the size of the value here, the World Economic Forum estimates over the next decade, in the consumer world it’s about $10 trillion of value to be created, in the industrial world just applying digital technologies, it’s about $18 trillion of value to be created. The biggest challenges that we see is complacency from CXOs and CEOs because they can’t chart the path fully over the next five years, it takes someone with the courage to have a leap of faith saying, ‘I know the first two steps, let me take it. And let me make sure I have a team that can consistently figure out the next two to three steps behind it’. That takes a different mindset than people who have grown up purely in the industrial world, it takes visionaries, CEOs and CXOs to make that leap. That’s the first hurdle. 

From a technology perspective the next hurdle that we see is, sometimes people thinking, ‘Okay, I’m going to build this myself. Because this is core and competitive to me, I’ve got to build everything’, and sometimes we see people saying, ‘Okay, I just want to build it on top of a public cloud, like an Amazon or a Microsoft Azure’, and we tell them those are great decisions of a great company, they make great infrastructures of those platforms. They’re a fantastic platinum service to building generic applications, but you need a lot more on top of it to get to your outcomes that you care in the industrial world. 

So, what we’ve done at Uptake is, we genuinely believe that any such technology in the modern day should be cloud data technologies, meaning technologies that were born in the cloud. The design principle of that also means that it shouldn’t be native to any one particular cloud, because when you build something related to one particular cloud, and a customer has different preferences whether they want to be on Amazon, or Azure, or Google, or only cloud, or a hybrid scenario, and you’ve got to be able to support it, that means you’ve got to build technologies that are native to cloud, but not native to any one particular cloud. A good container technology is like Kubernetes, that’s what we build on.  

At the same time, you’ve got to be building true machine learning technologies, meaning technologies that are engines, that are first-class citizens in a programming model, versus… and we’ve seen this moving in enterprise software, where you build simple transactional applications, and then you try to shove analytics into it, it doesn’t work. A decade into enterprise software we still don’t have ERPs with first-class native analytics, or CRM with first-class native analytics. That cannot happen in the industrial world because inherently these are analytical applications, so the other set of design principles that we have are about how do you build machine learning engines natively into the technology stack.

Are there challenges in managing the data? You’ve got a long background in working with enterprise data, I guess the human created data, and business analytics, but the challenges, the volume, the velocity of machine data and also the data structures are different; could you talk about some of the challenges involved, or the considerations when your pulling data from so many different sources, and what sort of approaches are helpful to deliver business value from these different systems, and data-types? 

The data challenge in the industrial world is way more complex than in the traditional enterprise apps world. If you look at a modern powerplant it could have 5,000 to 10,000 sensors. If you go to any environment, even like a wind-turbine or a wind farm, you have hundreds of sensors and thousands of tags that are sending you data. A lot of these happen to be isolated data silos today, whether it’s in old site or some kind of an historian database, and sensor database, they are all isolated and these are all data streams that have not been correlated. One of the things that you see over time with people trying to do analytics to the industrial scenario, it just sticks together all the data to make sense of something as simple as an asset model. This is what the direct industry has done, lip-service, like calling a digital model of the asset, and we would say, ‘That’s just a data model, that has nothing to do with the digital model, or the asset’. But even that is hand-stitched and hand-assembled in most scenarios.  

What we believe can happen is, the rapid modelling capability, which is why at Uptake we built a machine-learning engine that does this automatically. We call them the Labour Correction Engine which understands that this is the same sensor that in thousands of other assets it’s just been labelled in different ways, and understand the correlation of the different sensors, to quickly create what we call the Industrial Asset Graph. So, we can quickly understand a set of assets very rapidly, and prepare that data for analytics, as well as training, so we can quickly build models on top of the asset, model the particular asset, train that asset model, the machine learning model on top of the existing asset graph of data, and quickly start to get into predictions. 

So, anybody trying to do this in a manual way is missing the point. You’ve got to apply high productivity machine learning tools to the problem of doing… it’s just like we’d talk about software to build software, it’s kind of like machine learning to build more machine learning models, is how we approach it. 

It sounds like the approach you’re taking again, is you’re approaching the goal of delivering replicable solutions, but along a very vertical-by-vertical, or case-by-case type of basis. I’d love to get your sense of the trajectory of expectations and adoption in the IoT market. I always find this really interesting, if you go back about four or five years ago when the first big forecasts of the number of devices, and economic value that was going to be created by connecting physical assets came out; I think there were a lot of expectations that we would see this massive inflection, an enormous spike up into the right.  

I’m referring to spending, and in contrast what’s happened is I think initially expectations were aggressive, but adoption has been much steadier, maybe slower than some of the more optimistic forecasts. I’d love to get your thoughts on what are the challenges to adoption of comprehensive IoT, and industrial AI solution, and where you think the greatest value is being created? 

It’s a great question when you look at the trajectory of where we are as an industry. I would say, adoption is happening slow and fast, it’s happening slow by people who are unsure about how to begin this journey, and we see many that start doing pilots and they end up just in pilot-land. At the same time, we also see big leaders act more decisively and make a bet and go really fast.  

The best transitions that we’ve led seen are a CEO, CXO, believe led-transitions, where it got outside returns because of their belief in investment, versus a ‘Let’s go do a pilot’, approach, now, that’s hard. At the same time, we’ve also seen slower adoptions by… and I think the visionary CIOs, CTOs have a different approach, but most of them that end up in a, ‘Let me triall it, let me go and prove this in one scenario’, cannot quickly lead do a full-on business case and value, across a full enterprise. 

So, the biggest advice we tell people is, have a look at your entire enterprise, get started, you’ve got to have a basic  tribute and make a leap of faith. At the same time, I would say the adoption is robust, where we see a challenge is data, where most companies don’t have a good picture of the value of data that they have, the integrity of all that data and preparing that data to be able to apply. Then the next step is, ‘I’ll just go try this’, and, ‘I’m going to assemble a best of breed and go build something’, and we sometimes challenge some customers to go. The best visionary customers can be successful by assembling so many different pieces of technology, but they’re going to end up building something that they have imagined today, three years from now, and it’s going to set like concrete, and it will not evolve. 

Then the biggest challenge is, people creating the journey map or road map where they can make a few big bets, and involve architecture, they can continue and run fast, so they can chase ideas from the digital exhausts of their operations. I think there we’re seeing some of the leaders and the visionaries, like the… I still do the example of the US Army, the US Army last week announced that they’re going to be investing $2 billion into AI. Yes, it’s going to go into a lot of different areas, but it’s also going to go into simple mundane tasks like maintenance, producing operations and maintenance costs of their fighting vehicles, or in the Marines, or in the Army, or in the Air Force, and those are all great examples of ideas where you are very big with the belief, and you will actually find out the used case as you deploy and evolve with architecture. 

We see the demand for this going up, and we see this is where companies like Berkshire Hathaway Energy, company’s, ideas like the US Army, these are ideas where you can go run really fast. 

That’s a key point, this idea that with an extensible or an agile platform you’re able to really discover new use cases and value that you don’t necessarily anticipate, going into the beginning of a project. Would love to get your sense on some of the industries where you’ve seen some notable successes, i.e. I think the US Army example is a great one. But from your perspective, are there industries that stand out as being particularly visionary and successful, and any others that have taken unconventional approaches to adopting these connected industry technologies? 

I don’t know if I would say any industry is visionary or further ahead, obviously a lot of these industries are in some ways visually backwards, they’re not telecommunications, this is not banking, this is not finance, this is not autonomous driving, although I believe there’s going to be an autonomous moment in every one of these industries. But, to see the activity in the last two-years, it’s very different than say all the last 25-years or several decades put together, that’s everything from energy, to oil and gas, to construction and mining. So, I’ll use an example as simple as construction and mining, and to go back to your previous questions of getting started and discovering new values. If you look at Uptake, we’ve got more than 100 customers, and there’s a fair bit of those customers that are also managing assets across multiple different kinds of industries. A particular example would be about dealers, dealers of construction and mining equipment, they actually manage over 400,000 assets on the Uptake cloud, and they got started with a simple idea of simple calamitic from the assets, to doing CRM functionality, and over time adding insight; because they had all this data, now they could deploy some asset insights. Once they deployed their asset insight, they are now able to go and implement a full-service workflow and transform the entire service operations. 

Now, the entire dealer, they’ve increased their productivity, augmenting the technicians, they’re improving the scheduling and reducing the number of hours spent, and administration’s down. There’s one such dealer, they’ve managed about 200 assets, they start off with the simple idea of, ‘We just want all the assets connected, and we want them connected to our servers, software, our CRM software from Uptake’, but then they start adding the active insights, they’re now generating millions of dollars in savings.  

So, you see, even an old industry like that can get huge benefits from the moment they start getting connected, and make that first step, are now are able to monetize all that operational data. We do the same things in examples that we’ve done in wind where we can now predict a wind turbine, we predict an anomaly 300 days ahead of the lead time. That can fundamentally impact how you change your business process. Or, in oil and gas, that’s another industry we see leading fast, and growing even faster over the two-years, because of that value. So, we’re seeing our value applications being deployed. There’s a large American oil and gas company has implemented that, and they’re now using our solution across over almost 650 different pieces of machinery, and less than 15 percent in less time, they’re using 90 percent less resources than previous methods, whilst improving their operation safety, and asset reliability. 

So, we see everybody starting to go faster over the last two-years, because we as an industry now understand the value of digital transformation, there are more success stories that are coming out, than the failures of pilot hell, people that big are starting to win big. 

That’s super-encouraging to hear. Are there any common characteristics or stand-out qualities of businesses or projects that have been able to be successful in navigating their business model, or culture, to one that’s really enabled by IoT technologies? 

If you’re asking do I care for companies that are doing this type of role well, is that…? 

Yes, in your view who is doing it successfully, and are there some lessons that other companies can learn from their successes? 

This is something that I get a chance to consult with lots of CEOs and CXOs, and board members across the globe, in terms of how to go run digital transformation, and I came into 124-year old company and got to build the team that literally wrote the play book on digital transformation. From my experience I’ve learned two things that I see as common-purpose, from other leaders who are doing it well. One of the first things they talk about is, you’ve got to build a belief system, and what I mean by that is it doesn’t matter what industry you are in, you have to have a belief that at the critical control point of application, data, or algorithms, that you’ve got to create, or you’ve got to orchestrate. It means that you have to have a belief that software, analytics, or machine learning, or AI will fundamentally reimagine your business. 

You have to have a belief, if you don’t have a belief, I tell people ‘Don’t do it, you can save yourself a lot of headaches, a lot of money, don’t even embark on this’. Don’t do it because somebody else is forcing you to do it, because your board is asking you to do it, or somebody thought it was a good idea, but you don’t believe it. You’ve got to build a belief system, that’s one. The leaders that do that, they’ll be honest with you, they’ll tell you, ‘I don’t know the exact 10 steps I’ve got to take, but I believe this is the right thing. I believe that I’ll figure out the next steps like I said, consistently, and I’ll assemble a team to go with that’. So, build the belief system is the number one thing that I would say.  

The second thing I would tell companies that do it fairly well, acknowledge the fact that you need new metrics, new measures. You can’t measure these new ideas or investment in digital technology, the same ways you measured big investment in traditional CAPEX projects, or construction projects, or if you’re a building power plant, or factory, you’ve got to look at them differently. So, you’ve got to calibrate your expectations in terms of new metrics, new measures. That means you’ve got to do things like, what percentage of the data is being analyzed? How accurate are your predictions? What are the adoption rates? What’s the engagement rate of the applications you’re creating. So, new measures and new metrics, people that are willing to embrace that go further, because they look at signs that better indicate the future upside from their digital investment. 

Third I would say is new ways to work, new ways to work means adopting practices from agile practices, or ideas from the lead start-up, and changing the culture or design thinking into problem-solving in their work environment. So, I would say those three things are some of the things I see leaders do well.  

The fourth thing, I would say the next factor is people that acknowledge the diversity of time, and diversity in general to innovation, where they want to bring the people from other industries into their company, whether you offer a large trucking fleet, or you do manufacturing across multiple lines, or you are an energy provider, or a construction equipment provider, bringing people from other industries. Apple is diverse in terms of gender, and thought, and different exponential can bring and solve problems and attract talent. 

I think those four elements are the things that I see from leaders that do this fairly well. 

It is so critical to be able to think across different industries, to be able to tie together disparate experiences, that’s been a key differentiator, and I think what’s so fascinating about how companies like Uptake and others that are working in the industrial IoT area, which we’ll call very broadly are really helping industry to transform themselves. It becomes pretty profound the ramifications here. 

So, one final question, just looking forward, how you see the markets that you play in, evolving over the next decade. What are you optimistic about, and what are some concerns that keep you up at night? 

I’m really optimistic about the change that’s happening in terms of leaders who want to own this personally. CEOs who want their own personal digital transformation, and leadership schemes across many of the industries thinking big and investing big, I’m highly encouraged by that. And I think we would see more large-scale systems like the ones that we have within Uptake, as an example, the industry so you’ll have more successful examples of people can look into, versus get out of this pilot hell. We also see that we’re an inflection point where people can point to and say, ‘By deploying IoT and industrial AI solutions, here are the values I’m seeing’, and the confidence factors in these systems are going up. It’s radical transference is needed in digital transformation, and we provide that to our customers, for example, we actually deploy thousands of machine learning models in production, and would say to our customers, ‘Here’s what the prediction is, here is how the prediction are getting better, there’s a confidence factor in the predictions, and here is how its improving over time’. We call it the Model Performance Management report. Our customers can track back, and our predictions are getting better all the time, and their confidence in the acceptance of those predictions, which leads to what I call hot-augmenting their domain expertise, and their intelligence with our software. I think we will see a faster rate of that, and as you know every prediction you make in the field of how machine learning, or AI impacts certain industry has only gone faster than people originally thought. 

The best minds in 2004 would have said at one point, ‘Frankly economists wrote that computer cannot handle the task of driving an automobile’, it took exactly five years from that statement for a Google Waymo cover 1,000 miles. I think this is going to happen faster, we’ve already seen machine learning engines and AI engines in many industries, do predictions way better than just purely physics-based models, we’ve already crossed that threshold, and we can already do that much faster than almost all the industries that we’re in. So, that speed’s highly encouraging, the technology changes we’re seeing, especially if you built on a new modern stack in the last few years, the benefits of the cloud technologies and machine learning technology, that’s highly encouraging. 

What I’m nervous about, I’m nervous about customers that have made big investments and bets in digital transformation with certain vendors, and when those vendors change course, they’re going to be left with a path that was no longer viable, and that may make them nervous about investing further. Other potential customers, they’re nervous about investing big in digital transformation, and we saw that, in fact we’ve seen that and we launched a campaign for lots of customers of OEMs like Siemens and GE and so on, who made big investments in digital; they made the right decision, it was the right decision and embarked on the digital transformation, but I’m nervous about the fact that some of those failures may deter others that we offer a program called Digital Safe Practice so they can continue their journey with a vendor who is fully committed to this market, and being a sufficient Switzerland across all of these OEMs. 

But I think the upside, and the value statement there is one of the biggest markets we’ve seen, and we see great missionary leaders betting big, so I’m overall more optimistic about the value that we can generate for our customers, than how fast is this happening. 

That’s great. I think we very much do share the same optimism. One final question that I save for all my guests, a recommendation of a book or resource that you would recommend, that sometimes you call it either the most gifted, or most recommended from your point of view? 

I don’t have a recommendation, but I’m just a few pages into a new book as of yesterday. This is a biased view, it’s from my ex-colleague Beth Comstock, it’s her book, ‘Imagine It Forward’, and this is a book that appears very personal in her journey that Beth has taken, a great leader that I admire and work with. I am just into a few pages of the book and its exciting, so I would highly recommend it, and like I said to buy it as well because it’s a good ex-college, a friend, and a leader that I learned a lot from. 

Well she’s just a tremendous thinker, and I think that’s going to be a must-read on a lot of people’s bookshelves. I’ve already heard people talking about it, so that’s a great recommendation Ganesh. 

It’s been a great conversation, I’ve learned an enormous amount. I think you are really up to a lot of super-exciting work at Uptake, and really looking forward to seeing the story unfold with you at the wheel there.  

This has been Ed Maguire Insights Partner at Moment Partners, with Ganesh Bell, President of Uptake, as our guest. Ganesh, I want to thank you once again for taking the time. 

Thank you Ed, thanks for having me on the show. 

Thanks everybody for listening.