Industrial machinery and AI – episode 4 transcript
Chris Pennington: We have seen a few industries that have actually moved away from selling products and started to sell a service instead. They still provide the machine, but they continue to serve a machine. And I think with industrial machinery, that’s something that has been verge of happening for quite some time. But technology like this really enables that. Doesn’t that help the machine builder actually see what’s happening with their machine real time?
Rahul Garg: Yeah, yeah. No, I think that the whole ecosystem of data sharing Ralf that you identified that actually is also a really fantastic now because many customers are not willing to share that data out right and giving them that control. And the ability right the with the control and the ability to share the relevant data with the relevant ecosystem members, I think that’s really a fantastic innovation from Siemens.
Ralf Wagner: Yeah. Rahul , you mentioned the executable digital twin. This is something really interesting and innovative. Could you put a little bit more color on it?
Rahul Garg: Yeah, you know, this is actually a really, really innovative solution as far as I’m concerned that we have enabled for our customers. We have all heard about how the digital twin makes it easier for you to design and build machines and even use it during a production process with virtual commissioning and whatnot. The executable digital twin is really bringing a lot of our capabilities together for a customer. What it does is bring brings in the simulation capabilities that customers typically use.
Bringing in what we call as a ROM builder, a reduced operating model builder, and AI and package all that into an edge device. What we are doing is now we are saying, hey, the executable digital twin or the digital twin that you had that you were using inside your design process. Inside your engineering process, we are going to take that digital twin and bring it to the shop floor. We are going to take that digital twin and make it usable and valuable while you’re running the equipment inside the edge devices.
The edge devices that we get from inside, we have got that set up where you are now able. To bring the digital twin inside the edge device and run it as an executable solution. That’s why it’s called the executable solution. It’s a runtime solution that’s constantly monitoring the information from the sensor data and then using the digital twin information that was created during the front-end design engineering process and using that information along with the measured points. And using that that that simulation model as a reference point and using that information to change the behavior of the machine in a runtime basis, now we can use the digital tune as a simulation model and use that simulation model as a reference point, take the sensor data and then use that information to make your machines perform even better.
We have started getting customers beginning to use this. Some very interesting key areas, a company called Heller, they make very high-end sophisticated milling machines are using this to even limit the vibrations on the spindle. Typically, when a spindle is moving, there are metal chips. That come in and these metal chips sometimes can be very, very small, micrometer small, micro millimeter small, and these metal chips can have a direct impact in the way the spindle motor is moving.
By monitoring the spindle motion and by using the reference model from the executable digital twin, we are now able to change the way the spindle is moving so that the vibrations can be reduced, or they can be monitored in a much more effective manner so that the smooth and the features that you’re trying to get on your milling machine can be the most optimum. This is really a fantastic capability that’s been released by Siemens. I think it can have a big impact in in the future of how machines get designed and built by using the digital twin capability right on the machine itself.
Chris Pennington: It still amazes me just how fast technology is changing. And you know, I think it’s a really good example of that, that we’re already looking at production deployment with that executable digital twin. What I wanted to do, Ralf, was just ask you a little bit more about futures. I know at Hanover this year we launched a Copilot plug-in that was for instance. It’s only in an early access program at the moment, but what benefits do you think that Copilot will add when it’s fully launched?
Ralf Wagner: There’s two aspects to that, Chris. The Insights Hub Production Copilot was actually launched end of 2024 in in November, which actually I think we were the first to have a copilot for operations which helps the operator.
To come back to what we said at the beginning: at the line to when there is a talk help required, just to start to interact in a kind of chat bot ChatGPT manner with the data, with the events, what’s happening on the shop floor to get a talk help. This is what we launched at the end of 2024. And what we launched in early access at Hanover Fair was the Copilot Studio. This is the bringing this to the next level. And by the way, it will be a GA in September, so only in a few weeks from now the Copilot Studio is allowing customers build their own agents and skills.
Now you don’t have only the chat bot where you can start to interact with data that sits already in Insights Hub, whether time series, events, documents from your supplier, documentation of the machine and equipment, maintenance tickets, everything which is available and accessible via Insights app. Now you can build your own agents giving clear instructions what to do as well as building skills, which some call it tools as well. We call it skills, which is capabilities the agent then can call to get very specific answers, because you give you one example, you might have configured your OEE formula in a special way. In Insights Hub OEE that will be made available by an agent to the Copilot, and he knows exactly how your OEE formulation and calculations being done.
It can be very precise in any answer it gives you back when you have a prompt in the Copilot and Production Copilot. This is what we gave our early access in the hands of our customers in Hanover Fair and I have to say the feedback was overwhelming. Yeah, we had already on Realize Live which is our last customer event which we have in the industrial software in Siemens. We had customers presenting agents which they have been building in stage presentations. This is really opening up the next level of how to make use out of this technology in the production environment.
I’m very excited about Production Copilot and Copilot Studio at this point. It’s the fastest growing capability which we have in Insights Hub over the last few years because it’s really changing from going to dashboards, going to data discovery, going to alerts and triggers and rules and it adds this kind of completely new interaction. How to get value out of the data which is already there. Really exciting.
Chris Pennington: Sounds great. Thanks for that, Ralf. And I imagine many of the manufacturers would like to see machines that have got pre-installed AI capability. Maybe you could talk a little bit about that. Do you imagine that eventually we’ll see machine builders shipping out machines that have already got AI agents built into them?
Ralf Wagner: I think that is exactly what is going to happen because right now, to give you a concrete example. If an SMT line triggers an alert because there is a standstill on the step #3 and there is an error code which is 0X4686, the operator probably has no idea. You can then ask the Copilot because you have uploaded the documentation of that equipment and he gets then an answer with the exact steps. How to deal with that error on how to fix it can create a maintenance ticket because all the context is already there.
But you still need to do this today in order to drop the documentation into the inside integrated data lake to make that information available. In the future, I would expect exactly what you described, Chris, that the machine builders will have building agents already shipping with their machines, which then will interact in agent-to-agent communication with an overall Production Copilot so that you set up these kinds of systems basically in no time. It’s already very convenient today, but it will basically work out-of-the-box because we will use a kind of agent communication standards, which I will assume the machine will adapt as well.
That agentic world which we are seeing in front of us will come very fast as we all saw since the ChatGPT moment 2 1/2 years ago that innovation speed is unseen and agent will not slow it down. On the contrary, I think the agentic world will speed all of this even more up because the value is simply out of this world which you can get there.
Rahul Garg: This is actually really amazing, Ralf . I think in terms of what we are bringing to the world in innovation for machinery companies and manufacturers, I would imagine this is unparalleled.
In, in, in, in the industry, you know, based on your years of experience with Siemens, not only on the software side but even on the hardware side, do you think there’s any other company that can bring this level of innovation at the pace that we are doing this today?
Ralf Wagner: That is that is a very interesting question. I get this often when I have meetings with the industry analysts where they say wow Siemens, you have basically the broadest industrial software and industrial automation portfolio and things. How you pull this together from the automation layer, the PLCS, the SCADA systems, the MAS system, the DCS system from what we call the virtual integration with data sitting on top, but then also the horizontal integration.
Rahul as you mentioned before that what you find out during production either with the equipment.
Itself or with the products you’re producing feeding this back into the engineering and development process of your products or the equipment which manufacture your products, which we would call then the horizontal integration to bring this all together if you. If you don’t have a pre-integrated offering like Siemens is providing and you have this all from different vendors, you’ll probably spend millions in integrating this. that’s the big, big, big advantage if you have this from that one supplier that you have the integrations simply work.
Rahul Garg: I think that’s a great point and actually you know obviously while that’s our biggest value add where we get all these solutions to work a lot more seamlessly in in in an earlier point you mentioned something about another emerging capability with agents and companies shipping machines with AI agents and along that same process you also mentioned something about agent-to-agent communications.
While we are obviously leading in this whole process in the industry, I would imagine there are other companies also doing some similar things for this kind of agent-to-agent communication. Are there some standards that are now evolving? How is that? Because this is kind of a new horizon for the industry, if you will, right?
Ralf Wagner: Absolutely, Rahul. And there’s right now two main standards in the AI or LLM or agent space emerging. One is called MCP, which is basically being launched by Anthropic last year and it’s very quickly and widely adapted in this. And the other one is Agent to Agent, which was defined by Google just a few months ago. And these seem to be the two standards which are the front runners. They will probably come more, but these are the ones that are mostly talked about, and which is Siemens.
I’m working already closely with the other aspect I want to want to add to there why I think Siemens is in a very favorable position there. What we also announced at Hannover fair is that we are developing an industrial foundational model. What is this? Yeah, so we all know the LLMs, the large language models from the open AIS, from the anthropic, from the Googles of the world. They’re all very, very well trained on the general know how and expertise of what is stored on the Internet as well as on audio and video and images, this is basically the multimodal capability that these models have and they’re very, very good.
But these models are not aware about 3D geometry. They’re not aware about the concept of time series. They’re not aware about the concept of engineering and talking the language of engineering or the language of production, not even thinking about PLC code. They’re very good in all kinds of languages generating code, but PLC code not so much. Siemens is building that industrial foundation model which adds exactly these kinds of capabilities which the general LLM’s lack, the kind of industrial flavor and industrial modality we call that. And then having this going forward underpinning our AI capabilities as well will probably leapfrog us in the next dimension.
Rahul Garg: This is fantastic. What Chat GPTs and the Googles of the world are doing for the consumer, for the common man with AI, we are bringing that to the industrial world with the industrial foundation model. This is, this is really fantastic. Ralf , thank you for sharing some of those insights.
Chris Pennington: Yeah, it’s been a great conversation, absolutely information packed, and I thank you both for that. I would like before I wrap up, a last-minute thought, you know what? What was the one thing that you want our listeners to take away with them from today’s conversation? Ralf , maybe you go first.
Ralf Wagner: I would say dear manufacturing company, the time to get started to use this technology is now if you’re not getting your sleeves rolled up and get to data-driven manufacturing taking advantage of these technologies. You probably will run into competitive disadvantages going forward because others will do and the value and the benefits you can get out of that. I gave you some examples earlier are significant and can make a big difference going forward when it comes to your competitiveness on the manufacturing floor.
Rahul Garg: Yeah, Chris, from my end, I think Ralf sent something pretty critical, right? The whole IoT based solutions have been around for the last 4-5 years or maybe more I think, but they have no one’s talking about IoT anymore. Siemens talking about the value right from the from these solutions that can be achieved.
But the biggest turning point that’s come around in the last couple of years is the speed and the urgency with the with the with how the AI revolution is moving. And to me I think the IoT based solutions that we have to offer and some of the other capabilities that we are delivering as a company are so valuable and the AI push that’s coming around today puts this whole process into a hyper accelerated mode.
And to Rahul’s point, if companies do not figure out how to move on this quickly, they’re going to be left behind. It’s going to be a question of not just, hey, is this good to use? It’s going to become a question of survival to use, right? Because your competitors are moving in at a very, very fast pace. To me, I think this is, if there’s one call to action, I would say there’s some fantastic capabilities, some new innovations that are coming in. Companies should be looking to leverage that in the fastest possible way.
Chris Pennington: Great summaries guys. Thanks very much for that. Been a real pleasure talking to you both today and thanks for joining me and just sharing all that expertise and knowledge. I’d also like to thank the listeners for tuning in and we hope you stay tuned for future episodes where we’ll carry on looking at developments in a I and what the future might bring.
Thank you very much.
Siemens Digital Industries Software helps organizations of all sizes digitally transform using software, hardware and services from the Siemens Xcelerator business platform. Siemens’ software and the comprehensive digital twin enable companies to optimize their design, engineering and manufacturing processes to turn today’s ideas into the sustainable products of the future. From chips to entire systems, from product to process, across all industries. Siemens Digital Industries Software – Accelerating transformation.
