Ganesh Bell on Digital Transformation and Industry 4.0

Why We Need to Fundamentally Reimagine Some of the Oldest Industries in the World.

Over the last decade, labor productivity growth has been in a free fall. However, many people believe we already have the technology we need to reverse this trend. Uptake’s president, Ganesh Bell, shared his thoughts on this topic at the MIT Technology Review EmTech Conference, which explored how industrial businesses are using artificial intelligence (AI) and machine learning to boost productivity.

When you look at productivity trends throughout history, technology and digital transformation have caused key inflection points. Bell says, “If you go back and look at the analysis for the past four decades, every time there was a spike in industrial productivity, you can directly map it to the birth of the PC, automation, ERP systems and the internet. And now, we think analytics and AI.” The Fourth Industrial Revolution, or Industry 4.0, is being driven by software that effectively uses AI and machine learning to drive financial outcomes for industrial companies.

Elizabeth Woyke, senior editor of MIT Technology Review, sat down with Ganesh after his speech to ask some hard-hitting questions about how Uptake helps customers on their digital transformation journey. Check out what they discussed: Check out what they discussed:

Elizabeth Woyke (EW): We’ve written about plug-and-play AI that companies can buy off the shelf and apply to their own data and business goals. Why would the industries that Uptake is targeting not choose that type of AI? Why do they need a specialized platform?

Ganesh Bell (GB):
Great question. When we talk about outcomes like reliability, unplanned downtime or efficiency, every one of these outcomes can be broken into a subset of problems and people want to know that we can predict those problems pretty quickly. I explain it like this: It would be like if every one of us got a smartphone and we had to teach it vocabulary, we had to teach it our language, we had to teach it our dialect. Yes, your smartphone can start to speak, but what if it came already knowing how to speak, and all it has to do is adapt to you. That’s really what customers are looking for because when they call us they ask, “Have you done this kind of an asset? If you haven’t, how quickly can you?”

EW: So it’s that level of familiarity that they need – that’s interesting. Now, I’d like to ask you a little more detail about your data science. To what extent is Uptake using data analytics versus what one might call “true AI?”

GB: Most of our algorithms are a series of engines that we built using a combination of supervised learning, unsupervised learning and deep learning techniques. We say you can “meet an asset” very quickly and model that asset very quickly in our system.

EW: Related to that, which of the companies or industries you work with are able to fully utilize the power of your AI and machine learning engines today?

GB: I think in most of the industries that we’re in, it’s pretty low-hanging fruit. To your previous question, a lot of industries have actually operated on some kind of analytics, but the analytics they’ve operated on are what we call first-principle, physics-based analytics – not learning-based systems that look at all of the operational data. For example, a jet engine flying here in North America is very different from a jet engine flying over in Dubai in harsh conditions – so looking at all that operational data is another important part. I think in every one of those industries, they’re ready to apply machine learning techniques and they actually have the data, but most of the time we see complacency as well as a lack of belief that this can actually supplant a traditional rules-based system.

I think in every one of those industries, they’re ready to apply machine learning techniques and they actually have the data, but most of the time we see complacency as well as a lack of belief that this can actually supplant a traditional rules-based system.

EW: Right, so that’s why you mentioned in your presentation the closed loop. There’s a certain amount of education and confidence building you need to do.

GB: Absolutely, yeah. That’s why we go through a process where we bring domain experts in. We actually build a lot of this with domain experts because domain expertise matters. But knowing AI and machine learning matters as much, if not more, than just knowing domain expertise.

EW: That’s a great segue to another question I wanted to ask you. There are companies that compete with you, such as GE, that make both the physical infrastructure as well as the software that connects and monitors that infrastructure. At Uptake, you have your platform and apps, but you’re not building power plants, for example. What are the pros and cons of those two different approaches?

GB: I think in some ways this is like arguing Windows versus Mac. Which is a better model? Many would agree that Window is probably a better business model, and that Mac required a Steve Jobs to go make it happen. I don’t think there’s any Steve Jobs in the industrial world, so I think that companies that build software that works with multiple hardware will win.

EW: [Laughing] Are you the Steve Jobs of the industrial world?

[Laughing] I don’t profess to be anything! But getting back to your question - does the coupling of hardware and software matter? Domain knowledge absolutely matters, but the knowledge of operations matters a lot and the operating data matters a lot. While OEMs make the machine, the usage data sits with customers. It’s that data that’s more important. So that, married with the right algorithms and learning engines, is actually going to win.

EW: In your presentation, you outlined a number of examples where Uptake helped customers take preventive action, but what about outcomes? Going from preventive to prescriptive analytics and going beyond that to where you’re helping them unlock new revenue streams?

GB: Great question. In fact, most of the time our conversation starts with outcomes. In most cases, in these industries, people are paid on these outcomes. They’re paid on outcomes like eliminating unplanned downtime. They’re also paid on things like fuel optimization – because in some industries, like in energy, fuel is one of the biggest capital costs, so they’ve got to reduce that. We distill the outcome into a set of metrics or KPIs that we’ve got to be able to hit and then work backward from that. In terms of new economic models, we see this happening. If you take a power company, for example, historically they made money by generating and transmitting electrons and those electrons getting consumed. But now, in a world where consumers of energy are becoming producers of energy, companies have to think about new business models. How do you create new ideas and build new applications that understand end-consumer patterns? All of that requires software skills. They need to partner with someone like us to build it. In some industries, we’re in the early days of this, but eventually – in every one of these industries – companies will look at new, alternative revenue streams from digital assets.

In some industries, we’re in the early days but eventually, in every one of these industries, companies will look at new alternative revenue streams from digital assets.

EW: I also have a future of work question for you. In your presentation, you talked about institutional knowledge retiring and digital natives, but I was also curious about this: To what extent are your customers reskilling or upskilling their employees in order to take advantage of the new types of data insights your software is able to give them?

This is an interesting one. In many industries, they really have a huge problem. I think in North America, 50 percent of the workforce in power will retire in the next five years. That’s a huge amount of workforce and experience that’s walking out and we need to capture that knowledge. But the interesting thing that’s happening is, by deploying these technologies, they’re seeing some of the people who would take voluntary time off actually stay on because they’re excited about the journey of bringing on new talent and training them. And also, you’re seeing people starting to build data operation centers and intelligence operation centers. And a lot of these people are getting new jobs and getting reskilled. I think it’s an interesting dynamic and we’re also seeing that when companies have a digital bend – where they’re trying to transform how they work and put new systems in – they’re also able to track the next generation of talent coming in.

See how Industry 4.0 delivers new revenue streams for industrial businesses.

Read more about Industry 4.0