Cultural Adoption of AI PdM & the Power of Strategic Partnerships - a Panel Discussion - Part One

Trend Detection Podcast ·

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.

In part one of this episode, you will learn about:

How AI is transforming predictive maintenance: Learn how AI-driven solutions like Senseye make predictive maintenance scalable—from thousands of machines at enterprise level down to individual components—without the need for complex modeling or manual data input.

Making PdM accessible for all business sizes: Discover how small to mid-sized enterprises (like those in New Zealand) can start with as few as 50 machines and still achieve strong ROI within 3–6 months.

From data to decisions—without new sensors: Understand how existing automation data (like current, torque, or timing from PLCs) can be used for condition monitoring, often without needing to install new instrumentation.

Preserving institutional knowledge with AI: Hear how the system ingests everything from manuals to repair photos and videos, helping capture and share expert knowledge across sites to support the next generation of maintainers.

A practical, people-first approach to AI: Explore how Senseye’s philosophy prioritizes augmenting human roles, integrating with current systems, and delivering real-world value—not replacing people, but empowering them with better tools.

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

Welcome to the trend detection podcast powered by Siemens. Welcome everybody. Uh, we're here in the, uh, wonderful Brew Lab Innovation Center. Uh, our backdrop is our automated craft brewery, and it's all about innovation here, and, uh, we're very privileged today to have, uh, some visitors, uh, from the Siemens Sensai, uh, business, and, uh, to my right is uh Alex Hill, founder of Sensai and, uh, head of AI for the Siemens business globally. Uh, and to his right, Nick King from NZ Controls. And Nick is a solution partner we work with here for local implementation of the AI solutions. So I guess let's just start with a little bit of a, a broader introduction, uh, for the both of you. Nick, perhaps from your side, right, so we do uh system integration and we're working with the customers in New Zealand and around the world, but here in New Zealand, and we're looking For tools all the time to add value to data and to try to really help them move forward and in New Zealand I think it's well, well known we've got this um productivity problem and I think um there's, there's a lot of tools out there like Sensei that can really make a difference, um, and scale. So quite excited to, to be part of the team with Sens. And Alex, you're normally based in the US, as you can hear by my accent, yeah, there's a deep southern accent there, usually based in the US. I spent a lot of my time in the UK and since I was originally founded in the UK and now I get to spend my time coming and meeting wonderful partners like you, they can help us accelerate what we're trying to do, which is ultimately. Predictive maintenance at enterprise scale, and when we say enterprise, tens of thousands of machines, things that people have never really considered possible before. And just a minor correction, I'm, I'm one of, one of the co-founders, uh, so there's, there's a, there's a, there's a couple of others, and uh, we are, we are leading the charge for predictive maintenance AI within Siemens. So where we are here is all about innovation and thought leadership, and I guess you can't find a more bleeding edge topic than AI at the moment, but it's also a scary topic. So, uh, what are the tools, I guess, in general life that you are seeing out there in AI? That's a broad topic, isn't it? That's, um, I mean, AI is touching literally every single thing. I think it's Unimaginable, it was unimaginable two years ago that AI is going to start affecting the creative industries, for example, and now, any of us can log in and, you know, I want a video of a cat eating a rose and, you know, flying into space. Something that would take a creative team really quite a long time to do, and now you can get something pretty decent immediately. And we can take that kind of concept and that creativity into spaces that have generally not seen a tremendous amount of creativity, um. Which I've got an argument against actually, I, I think engineering should be tremendously creative, but AI is disrupting everything, um, and you guys are, are, are at the forefront of implementing it, so I think the question is actually better turned around on you. Yeah, so for me it's about finding tools that are real and practical that I can deploy, um, and you know that's been a challenge. Like there's a lot of hype out there and a lot of excitement. We use it every day for little bits and pieces, and we're using it more and more and we're seeking new opportunities with it. But, um, you know, the likes of um Sight gives us a really powerful tool, real boots on the ground application. You can see value being derived from it really quickly now, so it's really exciting. So speaking of scale. You can go huge, you can go multiple countries, all of that sort of stuff is pretty clear. But when does it become viable? How, how, um, accessible is this for a smaller business? You know, New Zealand has got a lot of small to medium enterprises with one factory or a small collection of factories. How accessible is it for them? Scale is something we'd like to talk about at the enterprise level of the 10s of thousands of machines, and that's really exciting. But it's also exciting for those smaller organizations, where as a predominantly an AI focused organization, we're learning a tremendous amount from all kinds of different places. We don't share that learning amongst customers, it's very important that we have that data sovereignty, and we protect that customer information. But we are ultimately maintenance and predictive maintenance experts. And so we can help these small organizations become more efficient. Typically we say that uh from an engagement point of view, start with 50 machines or so, and you can start to see a real return on investment within 6 months on that time and hopefully we can achieve a full ROI. So, so for the audience benefit, what's a machine? We've got the brewla behind us. Is that a machine or that's a really good question. So yeah, we have the brewer behind us. And, and one of our um pain points with predictive maintenance has always been that it's typically been very model based, and so that means you need to sit down, you're gonna need to get a, you know, do a full um failure mode analysis and understand that machine, and is that going to be worth it for your $100 valve or drive that, that could fail, but also could also bring down your, your whole piece of equipment or or facility, but traditionally the answer has been no, this is a $100 drive, I'll just replace it instead. With Sensei and in a scale, we can go from that $10,000 sorry, $10,000 machine view down to the single component and make it worthwhile because we are learning individual machine behavior based on the data. So something like this little, uh, I don't even know what this is to be like mess flow meter mass flow meters. So I'm happy you know. I mean you can tell that we're not. So we can now monitor something like that, and we don't have to spend $50,000 making a model and then tweaking it and then passing the cost of maintaining that to someone else. It's automatic. You have the data, bring it in, great. Since I will look at it, and every other single component within this machine. You ask what an asset is, and it's really actually our, our definition is whatever you call it. Yes, so it's part of the design phase or the the journey you go on is to define the assets and what is what is an asset, yeah. And if I've got an existing system I'm using for managing my maintenance or you know, the, the traditional rules-based maintenance program. Can I still work with that? Absolutely, and again a core philosophy of ours is we want to work with what you have, we don't want to throw away any of those existing investments. What we also do accept is that traditionally a maintenance management system, um, the quality of data in there is, Not good. People will often put, yeah, you know, last week, I fixed it. And the failuremo was other, and, and that's the bare minimum that they've needed to get through, and there's a part of the human element to this, and, and relying on partners to come in and lead these workshops is helping people to understand, OK, well you might have done it in the past, but now that we're using AI it's wanting to learn from what you did, and that's a core, uh, part of how we scale. The more you feed it, the more it's going to give you back. Yeah, so the better quality information you put in, the better quality information you'll get out, and there's a lot of work that we do to help improve the contextual information around that. But now we're starting to see, well, if we can educate a reliability engineer to say, OK, I've fixed it in this way, and I, I'm tighter now in how I record this in my maintenance management system. The AI will give me better results, so that I can be better and more efficient at my role, and that's a really nice powerful story that often people don't fully appreciate, but they, they understand once we educate them. So the, the, uh, we talked earlier about being able to upload, for example, instruction manuals uh from the OEM uh and, and SI will automatically look through that and feed up the data, feed up the information to uh to the repairer. Um, you mentioned a feature earlier when we were talking about the ability now to make a video, like a how-to video, or perhaps that engineer that's been there a million years, take a photo he's doing and take the photo. So all that is also ingestible by the by the Sensi tool, and it will feed that up here. Potentially all of the sites around the world and in particular is this institutional knowledge capture. What happens when your highly experienced maintainers leave? Uh, how can we start to, as an organization, help our customers not lose all of that information, and it could be videos, it could be photos, it could be somebody saying, I fixed the thing, but now we have more contextual information that we can wrap around that and use AI to augment that, and that becomes much more powerful. And that helps the new generation of maintainers who can't come in to a facility and put their hand on something, or just here and OK, well I know exactly what I need to do. We need to use AI now to help augment their roles, and we have the solution to do that, it's proven, it's deployed, and we can do it at scale. And that information stays locked down within the organization. It's not shared to other customers. So, and it's a software as a service, uh, solution. One of the barriers to assess model is often security, uh, data sovereignty, those sorts of things. So how do we address that? That's again a very important question. Um, as a startup, and, and there was a significant cost to us for doing that, we had our own ISO 27,0001 certification, as well as 9,0001, and we contracted, and we still do, an independent third party to come in and try to break it, try to do some penetration testing every year. We published those results to all of our customers. We've never had any significant issues. Uh, the, you know, the most we've got is like two medium things which are, Pretty easy to fix, and we fix those in a week and, and that's done. So that data sovereignty is extremely important. We isolate everything, it's encrypted in transit and at rest. There is not, and we spend a lot of time thinking about how we engineer this so that there's no potential crossover of customer information. And you touched on an interesting point when we started Sensei. Part of it was how can we help the little guy compete with the big guy. Maybe we can use that data to, to help, and, and I love the story. But humanity or people or whatever, they're not ready for it because ultimately customers say this is my data, and you must protect it, and I understand that, I get it. You know, there's this dream of what we could achieve, and then what we also need to achieve from a business perspective, and we respect the need of the business to be private and secure. So we're here in our innovation center, which is uh our investment in thought leadership and innovation, and we're always trying to show people what's possible. With Sensi, how do we show people this? What's the best way to demonstrate it? If someone out there today is interested to know more and wants us to get in touch. I guess, I mean, I guess we're talking to a New Zealand audience. I mean, reach out to see to see yourself. um, we're um like set myself up here to to help deliver the projects, but we're here to talk about what it can do, what it can't do, how it might work for you. So, you know, give us a call, you know. It's as simple as that. And that's what comes with scalability, this recognition that look, we can't do this alone. We need partners, we need people who have that deep relationship with the customer, who can do the integration, the connectivity, and ultimately, at Sensai, we, we're focusing on how do we solve this. This um grand problem and philosophy, and then we need the implementers and people like you guys. Yes, there's quite a few steps and there's a model to follow that you've developed and it's well proven, but um there's that's there for us to help take your our combined customers on a journey and help, help guide them right the way through the process, right through to getting, making sure they get value for their investment, you know. So I think that's super important, you know, the, the The amount of time we've spent talking with customers about this solution until now it's quite clear that unless you have a local team that can actually implement it, then you're a long way from Uh, you know, where it all happens, I guess it's all philosophical, isn't it? It's a long flight. Yeah, but there's a, there's a big, there's a whole team there, like, so we can call you, you'll help us, but we can also be like, it's about looking at local to local support because ultimately we're talking about AI, we're talking about automation of these things, but this is again about people. It's about making sure that when we work with the reliability team, we're showcasing them and we're doing that with local support and building on those human relationships. There's a lot of fear about AI replacing people. It's not about that. It's about AI helping people and how we've really build people up to say, you did a great job. And that, that productivity that you've talked about is, is really where it's at for for New Zealand Inc. I guess. That was my phone. Always listening, right? So I guess in practical terms if someone wants to see it in operation, we do have a demo, uh, ability to log into a demo set of data that's looking at real. Uh, a blend of real and simulated data across different industries. Uh, so it is something we'll have here in, in Brew Lab and have access to, uh, to take customers through and show them what it, what it looks like in real practical, uh, terms. Um, come, come to the Brew lab and we can show you running on Brewabb and, and it's a cool story. We have cold beer, freshly brewed beer, um, which are off camera, by the way. Is there anything else, Alex? You put me on the spot now. I ran out of questions. Yeah, I mean. What I really want to emphasize is, is this is about people, and, and we talk about AI, we talk about machines, we talk about condition monitoring, but this is about showcasing how people and teams can improve the efficiency of an organization, and building up that story and making it easy for often engineers are, and I'm, I'm an engineer by training and background, so I don't feel too bad saying this, Engineers are very bad at communicating, how they add value. And what that means to the organization and ultimately what we're doing is a beer they'll probably complain about no one listening to them, but you know, and then we're really about helping them illustrate that story. Yeah, yeah, for me, for me in working in the industry. And why I really, uh, was really excited when I saw Sensi is because, you know, I'm always looking for that, um, application that's very real and very deployable and will get value quickly or have real value to my customers because I don't want to go and sell them, you know, sell them or tell them about something that doesn't add value, you know, it's gotta be, it's gotta be a win-win, you know, um, so, you know, I was really stoked to see that and for me, the New Zealand productivity problem is my problem. I feel like, you know, I want to help solve that problem. So, you know, the likes of Sside there is just like a great tool, you know. So thank you, you know, for bringing it to us. Thank you for, for believing and, and, and, and you know, focusing on, on the, you understand the people element, you understand what we're trying to do here, which is not replace people, it's augment people, help people, stop people focusing on the things that really aren't that interesting. Do you want to spend your, you know, 8 hours a day looking at squiggly lines, trying to figure out what to look at and what that means? Probably not. Humans are much better at understanding. Um, very complex, very strange problems, very, uh, unusual situations. AI is very good at understanding very well defined problems. And we've spent 10 years very expertly defining condition monitoring and how you relate that to predictive maintenance. So let us take care of the mundanity of things, and you, as an expert reliability engineer, can focus on the really complex, really interesting problems that deserve your attention. Well, I guess the core benefit of AI is that it's monitoring everything all the time. So it's not a snapshot in time where a guy's coming in and doing a vibration analysis on an IR, uh, thermography survey or something like that. It's being monitored continuously. And so your focus moves to the important things and actually acting on some of those. Actually, you hit an interesting point there that, um, often when people talk about predictive maintenance or condition monitoring, they think vibration, they think rotational machinery, and they think complex, expensive data capture. It doesn't need to be, because when you're using an AI that has been specifically developed to understand condition monitoring and predictive maintenance, it's very good at taking inherently bad data. So, if a reliability engineer was to come along and take a snapshot of this and say, well, this is the current, this is the torque, this is the timing. In isolation, it doesn't really mean much, but when there's an AI looking at that all of the time, it's able to notice fractions of a percentage of change that will have meaning, and it can add context. And now it's able to say, well, Mr. Reliability or Mrs. Reliability Engineer, you need to come in and pay attention to this because it's changing in these ways. And if you were just to take an instantaneous look at that, it wouldn't be valuable, but the AI can come in and really help to prioritize your workload. So it's an interesting point. Like, uh, would you, if you go into a customer site or a site and they, they only have a certain number of instrumentation, they don't have as much as you might like, but do you just deploy it and find out, do we get good data, or do you ask them to start putting sensors in? That's a really good question. Um, in the past, and the dream was, uh, just give us your, your data that looks like a time series data and we can relate to it. And we, we found some failures there, uh, often people want to experience something immediately, so we come in with solution partners, and we usually then put in some sensors, so that we can get you playing with it and working with it in a really just a quick way and get you some quick wins, and then, working together, we can do the integration with the complex historians, and the maintenance management system, and that's the really powerful thing where, Look, it's, it's not worth your time and effort as an organization to put a vibration sensor on really, well, any of this. But you're already capturing this data. The PFCs controlling it can already tell you the timing and the torque and the current, and previously that wasn't used for condition monitoring. But with AI and with Senseii, we can do that. And so it's, it's the real dream of we can come in, we can instrument things that are really critical immediately, get you some value immediately, and then you can bring in all of this other information that traditionally wasn't thought of as condition monitoring information, but now can be. And some of our most successful customers have never installed an extra sensor. They've just queried the PLCs and they don't necessarily need to be seen as PLCs. Uh, you know, we have all kinds of different manufacturers. As long as you have some form of automation installed in the last 20 years, you have benefit to be gained from Sensai. Hm. 2. It's, right? …

Full transcript (WebVTT)