Join us for an engaging conversation with Philipp Wehn, Vice President of Innovation & Venture Building at Siemens, as we dive deep into the exciting realm where artificial intelligence and real-time location systems intersect. In this episode, Philipp unveils Siemens' remarkable journey of harnessing AI to achieve unparalleled accuracy in Real-Time Location Systems (RTLS)
Join us for an engaging conversation with Philipp Wehn, Vice President of Innovation & Venture Building at Siemens, as we dive deep into the exciting realm where artificial intelligence and real-time location systems intersect.
In this episode, Philipp unveils Siemens' remarkable journey of harnessing AI to achieve unparalleled accuracy in Real-Time Location Systems (RTLS). He shares inspiring real-life stories, including a deeply personal account of how this cutting-edge technology played a pivotal role in ensuring the safety of his newborn child in a hospital. Discover the myriad applications of AI-powered RTLS in healthcare and manufacturing, and how it's revolutionizing these industries.
We also embark on a journey through the evolution of smart buildings. Philipp compares traditional homes with their smart counterparts, examining the transformative influence of voice models like Siri and Alexa in reshaping our interactions with living spaces. Dive into the world of computer vision, with a special focus on its rapid growth in China, and explore the pressing challenges of safeguarding data privacy in the future of smart buildings.
Philipp sheds light on the integration of AI into corporate strategies, discussing its profound impact on productivity. Together, we contemplate the concept of autonomous enterprises and the potential for AI to liberate humans from mundane tasks, empowering us to tackle more complex and impactful challenges. This riveting conversation is filled with intriguing insights and real-world examples of AI in action that you won't want to miss.
Key Quotes:
Time stamps:
(00:43) RTLS accuracy with AI
(3:45) How hospitals utilize RTLS for increased security
(11:30) Expanding to smart homes and centers
(14:45) Expanding machine learning through the use of cameras
(18:50) Venture building and energy efficiency
(23:08) Protecting data privacy
(28:27) Building Trust and Sustainability of AI
Philipp: [00:00:00] With AI first, I mean, whenever I do anything, I ask myself, can AI help me? And I have different applications for that. This is also part of the Gen AI strategy that we put together here in the company, but for different fields like writing collateral, summarizing meetings, putting together decks, um, preparing for a session, summarizing a competitive report.
I would say generative AI in the past six months, 5X my. Personal productivity in my job.
Vahe: In today's session of Hybrid Minds, I'd like to welcome Philipp Wehn. Uh, he's Vice, um, President of Innovation and Venture Building at Siemens in San Francisco. At Siemens, Philipp is responsible for driving and accelerating the innovation program through in house development and strategic partnerships.
He also leads the development of the generative AI strategy for his business segments. Aiming to drive top line growth and bottom line optimization. [00:01:00] In November 2022, Philipp was named to German Capital Magazine's top 40 on the 40 list. Welcome Philipp and congratulations!
Philipp: Thank you so much, Vahe. Thanks for having me.
And yeah, good to meet you. And I'm excited to talk. So
Vahe: I read that Siemens has achieved close to 100 percent accuracy in real time location systems by leveraging AI. Correct. Can you explain how AI enhances this actually traditional
Philipp: service? Yeah, absolutely. Maybe a quick, quick recap on RTLS itself. It has been around for more than 10 years, right?
We have been hearing about real time location services and manufacturing and in healthcare for a while now. And whenever I talk about it, and people don't, don't know it or don't identify with it, I always take our Apple AirTags or our Apple Wallets and I tell them, You're actually using RTLS every single day.
A couple decades ago we used maps and we tried to find places, but now we [00:02:00] have this technology using GPS in our consumer life, and it makes our life better. So what we did here at Siemens and Lighted, um, we wanted to deliver a consumer like experience to the commercial RTLS world with a focus on healthcare.
And before we revamped and relaunched our products through 2022, we didn't have any machine learning in our location services. And with that revamp, and the shout out goes out to my head of architecture here, Nikolai Gigov, He and his team, um, they pretty much built two levels of, uh, machine learning, um, that run in the cloud.
And with that, we, with our sensor network and our asset tags, we create first room level accuracy. And then in a second level with communication outside of the room, we create even sub room level accuracy. Almost 100 percent of an accuracy level and [00:03:00] very low latency through machine learning. Latency
Vahe: is also a big issue there, right?
Because you want it really to be real time, basically. And how does AI actually really help you there? Is it, is it on cleaning sensor data or is it more of learning locations better or how do you...
Philipp: Yeah, let's, let's be honest. AI right now is like the big word. Um, a year ago, we would have just called it machine learning, um, which is part of the general AI space.
But, um, yeah, what, what we do there is we take the data after we train the model in a setup process. And, um, the data points that communicate with different sensors that are in the ceiling around where the asset actually is. And the model calculates where the points are collected, and then it predicts where the asset actually is.
And before that, we had like a more geometrical, just mathematical approach to it. And that first was way more [00:04:00] cost intense on our side, because it was And then second, it was less accurate than just training the model on how the sensor data looks when you are in a certain place.
Vahe: Very interesting. And, and what are the typical use cases for this and where is it like really
Philipp: used a lot?
So I'm happy to share a personal story and then also, uh, The moneymaker, uh, let's start with the moneymaker. Uh, the core ROI comes from, um, asset efficiency. So especially in healthcare, if you look at a healthcare operation, they lose about 20 percent of an asset in any asset class within one year. It just disappears.
Some, some might get stolen, some might move to another station. Um, so the first thing there is we can tell a hospital the exact utilization rate of an asset or an asset class. And then secondly, we can make sure that the assets that are not used are put away, [00:05:00] and then we can do also asset loss prevention through virtual geofencing.
So we can make sure if an asset leaves a certain area or enters a certain area, like the exit zone, we can send an alert to nurses or to staff. So that 20 percent asset loss can be reduced to approximately 5 percent through real time location services, and that is a huge moneymaker for healthcare. The second use case that I experienced seven weeks ago is, um, we can, with our badges and our wearables, we can also, um, if the client wants to do that, Accelerate safety in patient tracking and patient journeys.
So my baby had an RTOS tag on her right foot during her first six days of life in the hospital, paired with her bassinet that also had a tag. It wasn't from Siemens, but I told them that there are better products out there and they should look at it. Um, but these two were paired and they were also [00:06:00] connected to the station.
Um, the maternity station. So if, if, if I took the baby out of the room and it wasn't in her bassinet, alert to the nurses. If the baby would have left the station with that, with that tag on, the full hospital would have gone into like security mode. So those are just two use cases, uh, specific to healthcare.
And then in manufacturing, any lean management principles that we did for the last Decades, you can actually now see all your assets move in real time, timestamping, efficiency, acceleration. Um, you don't have to do any spaghetti diagram drawings anymore. You get the data in real time. I understand that
Vahe: you have a tag in the sensor, right?
And, um, I would assume that on the Tag, you don't have any or very little energy consumption and the sensor is the part that does more. But how is that, is that efficient enough? Like I would assume you need many of those sensors and then how is the efficiency, energy efficiency of such a thing?
Philipp: Yeah. So [00:07:00] that is the cool thing about our system.
When we look at real time location services, a major bottleneck in the past was the hardware and installation cost. Um, the Siemens Enlighted RTLS system is a lighting based RTLS system. That means the first core value proposition we bring to a client is that with a sensor in every light, We reduce the energy consumption through lighting by 70%.
So this is how you get the ROI out of the system immediately. And then with the RTLS system, you get a compound ROI. And that makes us better than the competition where if you are a hospital or a manufacturing site, you want to do RTLS, well, guess what? You have to buy this hardware. So for us, we bring an ROI through, um, through the energy savings.
We make our customers more sustainable. And then energy efficiency side, you just said it. Our asset tags have an accelerometer. So when they don't move, they only [00:08:00] ping every five minutes. As soon as they move, they ping constantly. And with that, um, firmware, uh, we are able to accelerate our battery life to five years.
Vahe: Putting into light is actually really smart. How do you come up with such idea? Does AI actually help with such a thing? Could AI be used or do you use AI to generate ideas, uh, in, in, in this area or?
Philipp: Yeah, I mean, kudos to the, to the enlightened founders. They had that brilliant idea. The idea that was born here in Silicon Valley, um, was to do, to create a neural network for buildings.
That's, that's what they wanted to do. They wanted to. Sense, motion, temperature, and then also, um, through BOE, asset movement in the building. And they wanted to solve that problem that buildings actually don't know what is happening inside of them. And then they looked at how can we best achieve that goal.
[00:09:00] And they saw that the lights are distributed over the floor in a very perfect way. And they are. in every single building and they all have electricity, right? So that's where they said, okay, let's put that sensor into the light, work with OEM so that you can actually buy light fixtures that have the sensors in them and make sure that the product comes with the light.
And then the software capability can be added. So, in that field, um, we're pretty much, we do have a, um, a sensor on the hardware side that does, um, daylight sensing. So as soon as there's enough daylight, the lights go down, the lights go out. And then we have a couple software applications, um, where we do have, um, I would say regular machine learning and, and, and software, where we do have a lot of AIs in our partnerships.
We extended our portfolio through our data [00:10:00] APIs with two partners in the last month, which is Tagnos in the healthcare space and ZenCompute in the smart cleaning space. And with ZenCompute, we take real time occupancy information from buildings, feed it into their Zenitor system, which is a smart cleaning AI, and then we pretty much Decrease cleaning costs by 30 percent through, um, real time occupancy driven cleaning.
And then that system also learns based on days, weather, calendar, it predicts how many cleaning personnel will be needed in a certain space. So this is again where I would say the human. The human capability gets accelerated through artificial intelligence. Very
Vahe: interesting. And where do you see the future of this going?
Especially with all this generative AI and the advances that we have done with AI, I'm sure also deliver new possibilities, right? Where would you see the future of
Philipp: this technology? [00:11:00] I would say the big step when we look at generative AI is that it makes First, it makes AI more tangible for most people, and it makes it accessible for us with a human language in a very simple way.
And then at the same time with these LLMs, you don't have to create your own machine learning model and train it with a specific set of data anymore. But it's, it's pretty much, it can ingest so insane amounts of data. And give you amazing results. So, we're currently testing how we can use this in our product strategy.
Um, looking at extending the data sources feeding into our software applications, that is a big piece. Right now, our own systems feed into the system, and then we're API open into multiple other products. But now we also look, what else can we ingest and use AI to faster get data and information out of third party systems?
Um, in general, I would say, In the smart building space, [00:12:00] and Siemens is pushing hard on that field with the Siemens Accelerator and the Siemens Building X product, um, whether it's energy management, sustainability management, safety management, or just having a central command center, there are so many data sources in a smart building, and AI will make it possible to just have the right information when you want it, when you need it, use it, and best case, Automatically make the right decisions to run buildings in a more efficient way and yeah, create a more sustainable planet through buildings.
Vahe: Yeah. And that's a really an important point, right? Because, uh, I, I have a smart home at my home. I have to say not Siemens, but you like it after, after this conversation, I have to change it to Siemens. But, uh, but before that, uh, it wasn't, you know what the problem is with it? The problem is. I don't think it's smart because at the end, it just removed the button from the walls to my phone.
It really doesn't do all these things that you guys are doing, which I [00:13:00] saw with like automatically putting on the light and so on. Uh, it really just, just moved the button from the wall to my phone. That's it at the end of the day. Right. Um, and I think there, you've got a really important point, right?
Because what you actually need to do with this information is make decisions based on this, right? Like,
Philipp: I actually have a different experience here in the US, um, with my smart home, to be honest. So yeah, it's, it, the first thing is it's all also coupled with my, um, With my Apple HomePod and my Apple TV, so it's all controllable through voice. I think that is a major step we will see. And I think, um, also through GenAI, we will see an acceleration of these voice models, Siri, Alexa, and others.
And then the second thing, geofencing, as soon as I leave the building, the house, my AC and heating goes to an echo mode. And then I can also opt into like a smart, um, AI driven schedule that is pretty [00:14:00] much based on all other houses in the area. And that adapts my schedule to maximize comfort and efficiency.
And then my watering system is connected to the weather forecast and has sensors actually that measure whether my plants need water. That might be Silicon Valley specific, but, uh, That part I have too,
Vahe: from Gardena. That's awesome. But yeah, that's awesome. That's true. But, uh, the in house, I think I made just the wrong decision to be honest, because I went with KNX, which is a very old system and, uh, I don't know, it's, it's not that smart.
Mine at least isn't. Um, Awesome. That's interesting. So what do you think about, I'm sure you have heard, you know, Tesla, for example, right, is using self driving cars based only on cameras. Do you think that locationing could be done by cameras only, like literally instead of having sensors and tags, use cameras and process that information?
I'm sure we're very far away from this because When you said that [00:15:00] language models are universal, you're very right, but graphical models aren't. So those actually would have need to be trained specifically, and that's a lot of energy and a lot of problems connected with it. But can you see a future where that is, is, is the status quo?
Philipp: I see a lot of potential in computer vision. And I saw when I was working in urban mobility with Siemens before, I saw A couple of really interesting startups and also AI technologies. Um, one of them is called G2K, good to know. They were just, they just exited a Berlin based startup and they did, um, they did occupancy analytics through cameras in train stations, including aggressive behavior, drunk people.
So If we look at countries, um, like for example, China, where there are cameras everywhere and, um, you track all this information, I think there is a lot of potential with that. I think in our Western [00:16:00] society, the one thing that keeps us away from major progress through the results from computer vision is...
Data privacy. We, many people, they really care about their data privacy, and I get that. I am the opposite. I always say I'm, I'm, I'm, I'm a glass person. Take my data as soon as, as long as you make my life better, take it. Um, but I do see a lot of potential in that, to be honest. Um, in, in the field of smart buildings and the business that we Siemens Enlighted, you talked about energy efficiency and then also size of hardware and everything.
Thank you. Um, I would say especially tracking assets and then saying whether it is the asset that ends with 347 in its ID or 349 and the asset looks exactly the same, you have, you don't have a chance to reach that with computer vision, but anonymized people counting, for example, at elevators, um, where you use, [00:17:00] um, yeah, cameras and computer vision to, to track occupancy of, of, of a floor of a building, um, I think there is a lot possible.
As long as... We're in those very strict boundaries of data privacy. Yeah, it makes sense. Of course, you're
Vahe: right. Because if the asset looks exactly the same, but has a different serial number, a camera won't help you there. Yeah.
Philipp: Or if the baby looks the same. How do you
Vahe: know it's my baby? You talked a lot about healthcare. I understand that that's a very big, uh, big sector for you. What other sectors are actually big in, in, in this technology?
Philipp: Yeah. Absolutely. And Leiter did grow in the commercial real estate and big tech vertical. So, uh, we are installed in, in many, many of the big campuses across the San Francisco Bay area, some of the biggest, um, manufacturing sites of tech automotive players.
Um, and then education is another one. So our four verticals is really, um, healthcare, as you [00:18:00] said, commercial real estate, uh, manufacturing and education. And, um, we always look at sectors based on certain criteria. So the first one is, is sustainability important for that sector right now? Is our rising energy prices a problem?
Are companies willing to hit certain, uh, carbon goals? And if that's a yes, that's a goal for us, because as I said before, the ROI of the product comes through a significant saving from energy. And then the second piece that we ask ourselves, where You pretty much have to differentiate even in smart lighting control, there are the Toyotas and there are the Ferraris.
And we are considered more of like a high end product with our applications and our machine learning, um, offering. So we look at the vertical, but then we also look at the client profile. Is the client willing to spend extra to get more value out of the product than just saving energy? [00:19:00] And then this is where we are really strong in all those verticals.
Um, and, uh, Yeah, I would say make our customers happy. By now we have more than 5 million sensors in the field. Um, and, uh, yeah, customers across those verticals across the globe pretty much, uh, rely on and light it to not only control their lights, but really to make their real estate portfolio smarter.
Vahe: The point that you bring up with this energy cost and, and this efficiency, I mean, it's really a big point, right? I mean, we just read last week that Microsoft is considering building its own nuclear plant already. Uh, and, and I just want to, I just want to reiterate that. I mean, before they had about 3%, 4 percent market share in search.
I mean, just imagine Google would go through all these technologies with, uh, with Genium AI. It could seriously be a problem for, you know, for our sustainability goals in general. Absolutely. Already in danger, but after that they are like totally off. Yeah. That makes, uh, that makes, uh, [00:20:00] really sense. And, but you're also doing venture building, right?
At, uh, Edmunds. And, um, so how does, uh, how do you see there the AI development? How does AI help you? And what are AI, um, uh, AI fields that you, you are investing in or looking
Philipp: at? Yeah, so, um, quick definition of how we define venture building at Enlighted. What we do is we, uh, make and we partner. So Siemens has a corporate venture capital fund, X47, um, and that is really all the venture capital that Siemens does is in that, in that fund.
Um, what we do here is, um, in house venturing. So building our own, um, new products, new small companies, and then partnering to build new ventures. And, uh, for that, we operate like a startup, um, we follow the lead startup method. We pretty much do an extensive customer discovery before we build anything. And then, um, after doing the discovery, we put it on, um, like a beta program, [00:21:00] test, test it with the customers before we launch it.
So this is how we do the venturing. And to your question, how AI helps me there, I transformed completely within the first year to be an AI first person. With AI First, I mean, whenever I do anything, I ask myself, can AI help me? And I have different applications for that. This is also part of the Gen AI strategy that we put together here in the company.
Um, but for different fields like writing collateral, summarizing meetings, putting together decks, um, preparing for a session, summarizing a competitive report. Um, I would say Generative AI in the past six months 5x my personal productivity in my job. Just because, um, whenever I do something, I ask myself, which tool can I use?
to accelerate the process to a result. And let's be honest, the result isn't perfect immediately, but I get from 0 to 80 percent within 30 seconds. And that is the time that, that [00:22:00] before generative AI, that drained my head, that made me tired. Getting from 0 to 80. But now I get this 80 percent perfect piece of something, and I can use my human brain to make it awesome.
And I think this is where we should all tap into GenAI. Even more than we already do. Yeah, I agree.
Vahe: I think after the GPT and philanthropic and so on, I don't wanna single out one company. I, I think I haven't written an email from zero to end myself anymore, , except it was a yes or no. Like yes or no. I'm still right.
But the
Philipp: rest is, it's, or you ask, tell me if I should yes or no, . Exactly. Please don't do that. Please don't do that. But yeah, I, I, I, I feel you. And, um, if you look at enterprise AI right now, I think we have to talk quickly about one, one boundary. Um, every single enterprise out there right now has the, let's be honest, a clear guideline not to use the open JetGPT product.
Because when you put content text in there from your work laptop, um, this is considered [00:23:00] confidential information. So what I'm personally looking forward to a lot is the Microsoft co pilot launch in, I think, 1st of November, that is what Microsoft announced. Um, but then also really like specific products in each field, marketing, operations, first level support, software engineering.
Um, I think companies, individuals or enterprise customers, they really have to tap into those products because if they don't, I would say individuals at some point will just go on JetGPT and say, I want to summarize my emails, I want to have this write my emails, and I think that can be really dangerous for enterprises.
Yes, I
Vahe: agree. Yeah, and so I get the point with data privacy. That's actually a big issue. Also, you know, for other industries, not only healthcare, obviously, but also, you know, banking and finance industries, of course, much driven by privacy. What do you think about when we are already talking about, what do you think about publicly available language models that then you can deploy yourself, right?
Like Lama [00:24:00] version 2, for example, and others are out there. And then Falcon was at 180 billion parameters. So they are also very, very good already. And you could deploy them like internally yourself and keep everything private. Like, have you considered that or are you doing something in that direction?
Philipp: We're not doing something in that direction. Um, I would also say my, my perspective is a little bit, a little bit limited here. My opinion is. As long as you understand what the LLM does and how it gets to a result, you can do it. But what I've written, what I've read so far, what I've heard so far, what I've discussed so far is that, um, core assumption is that we as humans don't really understand.
How the result comes together and where it comes from. So I think that could, that could be a limitation, but as I said, my perspective there is a little bit limited. I
Vahe: don't think that it's limited at all because that's exactly the [00:25:00] way it is. Um, at the end, I heard this quote in one of the interviews, which is we're actually not building AI, we're growing aliens because we don't know what we're doing, like we're trying it out.
Right. And it's really just finding a needle in a haystack. And, uh, and, um, the, the reason why it even works, because usually doing such an operation is just takes too long, right? But the reason why it works is the lightning speed you can do this, so it doesn't be tired. You can do it all the time, 24 hours on, you know, thousands of computers, hundreds of computers.
And that's. That's the reason why you actually still make advances without knowing what you're doing. You can bring this example, look at, you know, Leonardo da Vinci, one of the greatest minds wanted to fly, but he wasn't able to do that, right? Because every experiment that he was doing took him a year and then it failed and then he started again.
So if we do advances in AI in that speed, we would never achieve anything. But very interesting.
Philipp: Yeah. Very good point. I agree with that. And, and it also
Vahe: brings, of course, you know, the [00:26:00] limitations, as you said, where you use it and how you use it. Because sometimes you can't use it because you cannot explain why you get this, uh, results, right?
There's this whole thing about explainability and x AI and so on where, where it gets really important, right? Um, especially in healthcare. Just imagine, just imagine you are deciding. Um, transplants based on AI. You could do that. You might actually get a better result, but is that ethically okay? Because you can't explain it.
Right? That's the question. Like, why did AI choose that person and not you? How would you explain? Yeah.
Philipp: I would, I would say any, any company that's building something in that space right now, the holy grail will not be to take the open AI API, put a different UI on it. And, um, make it possible to upload a document to search for, pretty much.
I think that the Holy Grail will be understanding the model. Pre filtering data that flows into it, fine tuning it in the end, and then creating a solution that [00:27:00] is trustworthy and that solves a problem that wasn't possible to be solved before the rise of GenAI. Because what we see right now, also in the VC space, is we see tons of money flowing into products that are a homepage and a UI built around JetTPT, like around the opening API.
And... Even with the launch of Microsoft Copilot, I think many of those enterprise gen AI players, they will struggle. Because then, any enterprise account just goes up, I don't know, I think it's 17. 99 per account. And the person has the open AI technology built into PowerPoint, Excel, Outlook, Teams. And that's what we use most of the time, right?
So I think that's going to be very, very interesting how the, the, the launch of the MS co pilot, but then also the Google AI co pilot that they have in their workspace, how that will change the landscape of enterprise gen AI startups. Yeah,
Vahe: I couldn't agree more. [00:28:00] I call it AI adequacy because at the end of the day, we have proven with, with AI that it is, it can be superior in many things, but can you actually operate?
Operationally use it. Is it something that you, helps you to do your work? Is it really integrated into these, uh, applications? Is it private? And so on and so on. There are many questions. Trustworthy, you pointed out is very important. Uh, and that, that I think still needs to be achieved, right? For many, many use cases.
And
Philipp: I am convinced we'll get there. I personally have the, I have the vision of autonomous enterprises, a couple, couple decades in the future where we, we maybe have a society where we don't have competition between enterprises anymore and the enterprises interact with each other and just generate products and wealth for us so that we can return to do what humans can do best, be creative, invent things, save this planet and get results.
Get all the knowledge workers out of these out of these jobs where they do repetitive things [00:29:00] like they would be in an automotive car factory and do the same thing every single day. These people are smart, these people are creative, and I think AI has the potential to free them up and do what we as humans can do best.
which is solve problems, be creative, invent the future. Yeah, it's, it's
Vahe: refreshing. Um, I couldn't agree more. It's refreshing because very often you also see, of course, the risks, right? And, um, obviously lots of people, um, that are very influential that see also lots of risk and rightfully so. Uh, but at the end, I think it will be what we make out of it.
So if we are positive enough and really work into this direction, we can actually make a very positive
Philipp: impact. Yeah. As always nuclear energy could be great. And that is,
Vahe: uh, here in Germany. I know you, you, you were, I assume, born in Germany, at least raised in Germany. Yes.
Philipp: Born and raised.
Vahe: We have our own, we have our own strategy around nuclear plants.
Um, but, uh, I think that needs to be revisited here [00:30:00] in, in, in Germany. Um, do you think actually that, uh, because Siemens is doing many other things, right? Uh, besides what we talked about, where does AI fit in with the whole corporate strategy? I assume there are many other areas that really look into these, um, uh, possibilities now, right?
Philipp: Yeah. Um, you just said it, Siemens is, I mean, we were founded in 1847, right? Werner von Siemens, uh, his, his, uh, most famous slogan is I will never sacrifice short term profit over long term growth. He was one of the, one of the biggest entrepreneurs we ever had in Germany. And by now, and that's also the reason why I joined Siemens in the first place.
Siemens, no matter what Siemens does, it is always technology with purpose. Whether we do healthcare products, the Siemens Healthineers, or digital industries, or Siemens Mobility providing mass transit, train, rail infrastructure, um, across the globe. [00:31:00] Many people don't see the products of Siemens, but I would assume almost everyone experiences a Siemens product every single day.
And you are absolutely right, um, machine learning and artificial intelligence, they play a key role. And once again, it's on two sides. It's top line driven and it's bottom line driven. On the bottom line, I think, um, Siemens is on a path to pioneer and use these generative AI tools the right way.
Everything we just talked about, um, I think we're on a good path to use this technology the right way. And that will make us more efficient, more productive. It will make our products even, even better, faster to deliver. Um, and it will change the way people do their jobs. And if we can have more creative people in the company, we will become even better.
And then the other piece is, is really the, the top line piece. Core technology that Siemens is currently, um, pushing [00:32:00] as part of the Siemens Accelerator, which is the marketplace for digital products in the industries, is the industrial metaverse. Where Siemens partners with NVIDIA to, to bring not only a digital twin, but a real time metaverse for, um, industrial clients across the globe.
And, um, yeah, if you look at any software application. Whether it is a simple generative AI chatbot built into the product so that you don't have to search for certain problems but you can just, you can just express it in your natural human language, um, that plays a huge role. And then also learning about system failure and predicting system failure, whether it is in, in, um, the rail industry or whether it is in digital industries and, um, Industrial software.
So yeah, it plays a huge role. And, um, I think on the transformation to, to becoming a full tech software company, [00:33:00] um, Siemens is on the right path using what's out there. So
Vahe: right now I would assume, uh, the hardware business is still larger for Siemens than software, right? Or, or...
Philipp: Yeah, I mean, that's across the financial reports, uh, we're growing fast on software ARR.
Um, but, uh, we still deliver safety, fire, HVAC products, trains, like high speed trains, NEIC, uh, in Germany, um, but, uh, yeah, you know, the, the, the right entrepreneurial spirit is to take. Now I'm talking management consultant, uh, business administration, university stuff, but, um, you take the money that comes from your cash cow and you invest it into your question marks and stars, the BCG metrics, and, um, that is exactly what's happening.
And I saw many, many other companies in the past that didn't do that, that just. Kept milking the cash cow, some [00:34:00] automotive players maybe, and now they're facing severe problems and, um, we're not doing that. We're really inventing the future with our customers.
Vahe: Can you explore a little bit more about the metaverse?
Because, uh, it sounds interesting because you said it is a metaverse that you're building for corporations, right? Did I understand that
Philipp: correctly? Yeah. In manufacturing. Manufacturing.
Vahe: And then what is happening there? Uh, how can I, how can I imagine
Philipp: that? So this is, this is part of Siemens Digital Industries.
So this is a little bit outside of my, my smart building expertise and, and my mobility expertise. But, um, the idea we, we, we had, we had the word digital twins for a couple of years now, right? But digital twins are really used a lot in the planning stage. Before you build a plant in manufacturing, um, you build a digital twin and you see how your production could flow.
The step that Siemens is taking now with the customers is to take that into real time [00:35:00] operations. So if you are a plant manager or lean management manager or supply chain manager, um, and you want to see your factory in, in real time, you can access that metaverse. Not only with like VR, AR, we always think about the metaverse with like glasses on, but you can actually access the metaverse with your laptop.
And you can see exactly in real time in a virtual, um, twin of your, of your facility, what is happening. And then you can also simulate, like what would happen if I would add this, this, um, very high importance, uh, order from one of my major customers tomorrow into the field. And you can simulate through.
How the outcomes would be on certain orders, delays, um, and then also take that data, put it into other third systems, um, for other analytics. It makes
Vahe: sense. I could simulate if I change my lighting to Siemens, uh, uh, uh, your lighting, how much energy I would [00:36:00] save and how much smarter it gets at the end of the day, right?
Philipp: Because... 100%, 100%. We're happy. We're happy to help with that. Wonderful.
Vahe: Thank you so much. It was really great talking to you. Very impressive things that you're doing. Thank you. And greetings to Silicon Valley. You're in Silicon Valley now, right?
Philipp: Yeah, Silicon Valley. It's, it's, it's actually cloudy today.
That's the first, first time since I think April or May, but the sun will come out later. And yeah, we are, we are enjoying the, the AI hype here right now. It is a good time to be here. I
Vahe: can imagine. It was great talking to you. Thank you so much for your
Philipp: insights. Thank you very much. Thank you. It was a pleasure.
Announcer: Thanks for listening to this edition of Hybrid Minds. This podcast is brought to you by Cognize, the first of its kind intelligent document processing company, which automates unstructured data with hybrid intelligence and a data centric AI platform. To learn [00:37:00] more, visit cognize. com and be sure to catch the next episode of Hybrid Minds, wherever you get your podcasts.