Too Much Data, Not Enough Action - The Industrial AI Dilemma - with Richard Ella
Show notes
Welcome to the Trend Detection podcast, brought to you by Senseye Predictive Maintenance – which gives you visibility and insights into all your assets, from single machines to full plants to help you reduce downtime, increase knowledge sharing and accelerate digital transformation across your organization.
This episode covers:
Why having too much data could be holding your business back—and how to break free from the “paralysis in analysis” trap.
What makes predictive maintenance efforts stall, even when you have all the right tech—and how to get things moving again.
How AI and machine learning can help you finally turn data into action, not just dashboards.
The real reason so many sites repeat the same mistakes—and how you can shortcut that learning curve across your entire enterprise.
Why the biggest challenge with adopting new tech isn’t the tech at all—it’s knowing who to trust and where to start.
You can find out more about how Senseye Predictive Maintenance can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants, by visiting: www.siemens.com/senseye-predictive-maintenance
Transcript
Uh, anymore around adoption of, um, I, I now think it's more a a reluctance of who to choose to go with. So I think the, the value is there, uh, and, and I think that's well articulated. Everybody now can see, jeez, this is a, a real game changer and it's, it's, uh, I want to change. I want to adopt, I want these benefits. Welcome to the trend detection podcast powered by Siemens, and now your host, Nel Sullivan. Hello and welcome to the latest episode of the trend detection podcast. Um, today I'm once again joined um by my very esteemed colleague Richard Ella, all the way in Australia, um, you know, he's been on the podcast many, many times and he keeps getting invited back cos we talk about such interesting stuff. Um, and today is no different. Um, we're talking about data. Um, and this, uh, let's say, issue where, um, about having too much data and not enough action. So you're sitting on piles and piles and piles of data but not actually getting the insights from it that, that it deserves, really, I guess. But, um, I'm gonna sort of hand over to Richard to get us started. I think we all know who Richard is by now. Maybe one sentence of who you are, Richard, for new newcomers, and then maybe you could just give a bit of background on the topics. I know you brought this to sort of brought this to the table, so over to you. Yeah, thank you Nia. Um, again, just thank you for, for, for doing these, um, these podcasts. I think they are absolutely brilliant. It's really feeding into How we change, um, learning, you know, we, we've, we've become a micro-learning, um, kind of community. It's the only way we can stay relevant and you really are feeding into this, this micro-learning where you can listen to a podcast on the, on the tub on your way to work or you You, you know, you, you, you have this gap where you just, uh, incrementally, uh, improve, um, and, and, um, this really feeds into that. And, and thank you for the, the opportunity. Today's topic is something that I've been very interested in, um, throughout my career, um, and even on a, on a personal note, you know, so, I know, you know, we talk about Sensii and we talk about the value. Um, that, that what, what since I can actually do with your data, but before that, what we are seeing are quite prevalent in, uh, when we talk about digitalization or we talk about a digital adoption. is how do you get started? And this is not a customer, this is not an industry. This is, uh, really comes down to, um, you know, personal beliefs and, and how do you actually change and, um, the, what we're going to talk about today is basically too much data and not enough action. Um, and one of the phrases that I absolutely love, and I picked it up very early on in my career when I, when I, I bought my first, uh, Fitbit, uh, uh, to track my steps was paralysis in analysis. And I was like, that's a bit of a geeky word. What does it actually mean? Um, and, and what it actually meant at the time for me was, Um, because I had this Fitbit and it was technology and I was counting my steps in my sleep and I was studying, I was trying to get fitter, um, and I'm always trying to look at, uh, being healthier and, um, and being a data guy, absolutely loved the data. And, um, but what actually happens is you get into a trap where the more information you get, the more educated you get, the more stuck you get. Um, and Um, it's, it's only until somebody comes along and actually breaks that thinking and breaks that cycle of analysis to say, listen, What are you're actually trying to achieve here? Are you, um, be that predictive maintenance or be that better health. Um, and so there's this, there's this common phenomenon where we call it, um, paralysis in analysis and, you know, coming back to the UK it even goes back as far as, uh, Winston Churchill. I think he had a quote that was, uh, perfection is actually better spelled paralysis. Um, and what that actually means is that, um, we are, to get started, we've got a lot of data, we've got a lot of options, and we've got, we, we, we, we, we get stuck there. And so, Um, and since I in, in the predictive maintenance space is the, the most powerful tool I've seen break that cycle of traditional condition monitoring to, hey, let's, let's, what are you actually trying to achieve here, um, on your return on investment? Are you looking at, um, improving your schedule? Maintenance or your downtime. All of those things that everybody is trying to do with the, with the traditional condition monitoring and it requires a lot of analysis and it requires a lot of effort. Um, and so now Sensai comes along and says, listen, we can achieve the same thing, but in a better way, in a smarter way. And so, Um, so the first part of this para paralysis in analysis is understanding that our customers are stuck in their data. Um, and it's, um, and it's with all the good intentions in the world, um, you know, there, there's a reason why they got the data is because they wanted to use the data, but, um, with traditional condition monitoring and traditional tools. Um, it, it got stuck in the event, uh, rules or the rules-based world. And so, so that's really what we want to, uh, to talk about today is, uh, how do we break that cycle of, of, um, I've got all this information, but how do I get started? Who do I choose? Why do I choose them, um, and, um, getting the tap on the shoulder and say, listen, let's rethink this a little bit and, and lay a roadmap of how to get started. Um, now. Yeah, so I guess first of all, um, I think that's the first time Winston Churchill's been quoted on the podcast. Hopefully not the last, but, um, good, nice, nice quote in there. Um, but I, I guess the first point, um, to make is you, you mentioned that it's not like a recent phenomenon, so I guess people would presume with technology and things like that we're talking about, you know, data analysis now, but obviously data collection's been a thing for much longer, right? This is a, it's been going on for many years. Yeah, certainly. And uh with the, this whole shift of generative AI it's um it's, it's breaking the way that we interact with that data. All right. So if we look at what Sensai has adopted with a generative AI into traditional machine learning AI it's a completely new experience. Um, you don't have your traditional fault code spitting out something that you've got to go read the manual and it's not conversational like and, um, and really hats off to, to Sensai for, for leading the way on this. You know, there's a lot of, um, generative AI talk about, uh, how we, um, wrap this into, into products and, um, and There are a lot of products that wrap generative AI around it, but not many, uh, solutions that embedded in its, uh, offering to, um, to, to not just be a chatbot, not just be something that, um, is just an extension of chat GPT. It's actually a built-in language learning model so that you really derive um. Insights into, into, into your data and you're having that conversation, which look, for me, um, um, it's just a breath of fresh air because uh that's, that's what customers want. That we want to move beyond, um, just spending time Reading manuals and those kind of things. We just want to get as close to the diagnostic as soon as possible. So this data problem is not new. Um, I, I think people, we, we can, you can call it overwhelming. Um, people have it all the time. But, um, in the industrial world, um, this is very prevalent because, um, a lot of sites are coming on board now and connectivity is a prerequisite. So they've got all of this data without any effort, um, because of edge capability, IoT integration. All of these things are coming, uh, kind of really connected. Um, and, um, how, how do you actually convert that is important. Yeah, and I guess, um, that what I was thinking first of all is, you know, how, how do, you know, if we talk about the industrial setting, you know, how do organizations end up with too much data, you know, how, how does that happen, I guess, um, and the, and I guess from my point of view I've spoken, I've been at events and spoken to people and. Uh this was a few years ago now, but when they were talking about, oh yeah, we've gathered all this data, but we, we've no idea what to do with it now. And I guess that's part of what we're talking about today. But do you know how people sort of get, got into that situation to start with, of sitting on mounds of data. Yeah, look, whenever a project goes to work, um, or a project gets deployed, um, there are minimum requirements that get set up for communications. Um, and this is now becoming, um, you know, standard. So if I put in a drive or a flow meter or some kind of device, um, that data is now becoming standard. So, um, and usually during the project onboarding, That information gets, um, gets declared with all the best intention, uh, in the world. Every customer, um, thinks that when they're on board, they'll bring this in, it'll go into this magic box and, you know, it's just going to tell, tell them everything. And, um, and traditionally, uh, it required data scientists, it required subject matter experts, it required coders, um, and all of them to study the application and then derive uh value out of that. And so, Um, the reason why, um, our customers are experiencing this is because that's not scalable. You, you just can't, um, deploy, you know, these people just to go and do individual things. You need to change the way that you look at your data. And, and so, Um, so when we look at mine sites, this is massive. So, uh, we've got lots of data. We, we're commissioning new equipment, all of it, uh, every day. And, um, all of this information is just going into the skater usually, uh, and then usually kind of sits in a historian. Um, and then unless something goes wrong and they do a, you know, a, a lessons learned or Five Wis or some kind of, uh, investigation, the data never gets used. And so, um, and this is, um, becoming more and more prevalent. And so, It's, it's not because customers are doing anything wrong. They, this is not a, a failing of, of, um, of the industry. It, it's, uh, I think it's more of a failure to, um, to leverage that data properly. Um, and Up until um AI, um, it was really hard to be able to scale that. So, you can go to many different sites and you will, uh, you can quickly see if there is a data science team or there's a reliability team. They are usually quite energized. They are usually, they know their data. Um, and you will see wins at that site because they're engaged, they basically, um, but that's again, um, not scalable and is, is very ad hoc. Um, so what you want to do, I take those pockets of um energy of and focus and really leverage that across many sites so that you can actually gain those benefits, uh, rather than it just being something that, um, an engineering team did on one side and they're looking at scaling it, but they need to study it. And so, so, so yeah, this, this. Data issue is massive and it's getting bigger, um, and we need to maybe communicate better to our customers as to why traditional tools, um, get stuck and they're good for traditional condition monitoring. Um, and what the benefits are of, um, machine learning, um, and, and generative AI combined. It's a, it's kind of a real game-changer and can break that cycle for your data. And I guess when we, we, we talk about data as well, so, you know, uh, you know, we Sens as an example, it's not just about one plant or, or one production line, it's across multiple sites as well. And obviously if you add in multiple sites, that's more, more data and stacking up. I mean, what, so if we think about the, the data paralysis issue across the enterprise, how does that, uh, what's your sort of thoughts on that, Richard? How does that, how does that change what we're talking about here? So let's, let's just take for example, where we've got, um, you know, 5 sites and uh we've got exactly the same pump across 5 sites. Those 5 sites don't communicate with each other. They are very siloed in traditional and even across on each site, you might have 3 to 4 shifts. Um, so even on one site, you've got 3 to 4 shifts. Uh, where they don't get the full communication with each other. So, you get a lot of duplication in effort. Um, and what actually happens is, let's say, um, you know, they've got one piece of machine. area that's known to, uh, for the pump to get blocked, um, if they don't clean the filter. Let's just say that is a known issue. All right. That, that in isolation will become a known issue at that one site. All right. And um yes, they can go put service bulletins in place and do something for that, that one site. But let's say a brand new site comes with the same piece of equipment. For that next site to go through that same pain is an absolute waste of time, but happens all the time. All right, it's because each site needs to learn their machinery, they need to learn the uh rosyncrasies that word um of the machine. And so what you actually need to do is you, uh, what Sensi does is it's actually taking the, what's happening within existing site. So when a new site comes on board, it's taking those learnings across. With it. So that if that happens, uh, it straightaway tells you, listen, we've had 10 cases like this exactly and it was because of this filter issue and straightaway, um, the, the, the benefits are gained. But this happens, um, all the time. So, and, and I'm just talking about one application. So, you know, on, on a given mindsite or be it, uh, in any application, you know, we, we can have, you know, up to 300, maybe 400 motors, all with their own unique characters. And so, um, we need to break that, um, traditional way of This is a new site. We will learn, uh, what, what this machine does and adapt and accord. We don't need to do that anymore. We can hit the ground running within existing data. So, um, again, how do you break this paralysis, um, in data is because what, what every site will do on its own, they will study it and then they will go and I get stuck in this paralysis and analysis, and then I'll go, all right, let's go throw some more sensors in it. Let's get some more data. Uh, we still don't know exactly what the problem is. Let's call the OEM and let's, let's maybe get a contractor in to maybe give us a third party. And so you can see how this really snowballs, um, Into, uh, of no value. You, you might eventually get on top of the issue for that site, but where did all that pain get gained? And that's what we really want to, um, lever is uh starting with a with fresher data and a better way of communicating with your machinery. Yeah, so, because I know you sort of shared some, um, sort of bullet points before the court, you know, just to outline, um, what, what we're gonna talk about, and you, you, you have this phrase, the, the overthinking trap, and that's exactly what you're, you're describing there, right, um, this, you know, too much, too much data, you know, trying to find, trying to dig dig into it and understand every last part of it, but I guess that's. The, you know, the key to it all really isn't it, it's, it's, it's spending too much time on the data and not putting it into action. Yeah, that's right. Um, you know, we are all very, very busy, but We are, we, we, we've only got a certain capacity and this is what um Sensii brings to the predictive maintenance phase. It improves your capacity because you're not focusing on all of the different um things that are going on in your, your plant, you, you now have something to do that for you and now you are focusing on the results. So you, um, you don't have to worry about trying to analyze everything and trying to derive things. You, you know, OK, uh, it, it, it takes a while for you to convert your trust over to Sensai. This is a natural, uh, resistance that people will have. I see it every day. I was exactly the same. So, um, so when it comes to this, way, a different way of doing things. Um, this is what actually causes this resistance is, what is a program going to tell me that I don't already know. All right. And so only when you actually interact with uh Sens and you actually see it's, it's pulling things out that One you already knew, but also things that you might not have known. And so once you have this trust in the detection and it becomes an intuitive thing, that's why we call it an attention index is because what Sensii is all about is grabbing your attention, uh, and providing it and focusing it in a certain area. Um, and so what, what it allows you to do, uh, is get out of this paralysis and analysis and focus on, on results and value, um, and just let, um, let Sensai run in the background, um, and basically do a lot of the things that you were trying to do ad hoc and, uh, manually before. And, and of course, um, the, you know, there's, there's a lot to consider on the customer side, but, um, if we sort of flip it around and think about maybe from our point of view, from a Siemens perspective or just vendors in general, how can sort of vendors, um, help customers sort of overcome this sort of overthinking or falling into this trap of like overanalyzing data themselves. Yeah, there, there's, there's a couple of parts to it. So, um, you know, when, when it comes to value add, um, you know, I've been a, a service engineer for a long time, 30 years. Um, and if a, if a customer, uh, questions what value we are providing or uh if there's any question around the value, I will purely bring it down to a simple line that if a customer asks, why should I engage you or Siemens or a vendor? And my simple answer to that is because I want to call you before you call me. I want to bring something to your attention before you bring it to me. If, if that is the relationship, I'm of value to you, um, and it doesn't matter what tools I use. All right. But uh this is one of the reasons why I'm involved with Sensai is because it's one of the strongest tools that we have to be able to do that because our customers, um, they, they, they basically Uh, want to be contacted prior to a failure. And the last thing that they want to do is contact the vendor or contact Siemens, tell them, listen, something's failed, and then go on to explain, uh, everything about it. That's, um, so what Sensii enables a vendor to do to their customer or whatever the relationship is. It, um, it allows, uh, you to strengthen your relationship with the customer to say, listen, let me call you before you call me. And, um, as we move into this, everything as a service, which I absolutely love, you know, um, where everything's becoming subscription-based, um, I can really see this being the future. If, if, honestly, if, um, if I had a new plant, or if I had a plant and I didn't have, a digital transformation team or I didn't have a digital champion, but I wanted to transform. I would go to Siemens and say, listen, I need this transformation. I don't have these capabilities in-house. Can you provide that as a service? And you know, this is what Siemens does. So we offer the products that you can, uh, you know, Empower yourself and, and Sensai will go along the journey with you because it is software as a service and we are constantly engaged with the customer to make sure that they're getting value. Um, but the, the biggest thing Uh, when it comes to this vendor-customer relationship or what, what does Sensai bring to the table is this ability to speak to a customer before they speak to you. Because if you have that relationship, um, it, it's, it's a win-win. They are paying for value, you're providing value and, um, and Sensai is the tool to achieve that. Yeah, slightly biased, but I 100% agree, of course, but, um, but no, absolutely, absolutely. So, um, I, I think, you know, we've got a couple of, got a couple of questions left. …