3 Nov 2020 · 23m · Sam Ransbotham, Shervin Khodabandeh, Mattias Ulbrich · 23 quoted facts
Porsche CIO and Porsche Digital CEO Mattias Ulbrich argues that AI succeeds when it is built jointly by business experts and technologists, not handed down as an IT project. Using specific, narrow applications in production, design, supply forecasting and engineering, Porsche pairs AI with domain knowledge, wide staff training and a culture of continuous learning.
The central tension is between a technology-driven view of AI and the human work needed to make it useful. The standout story is a lab specialist's coffee-machine listening app, which inspired an acoustic assistant for detecting door noises overnight. The hosts conclude that creativity, expertise and cultural change matter as much as the algorithms.
Ulbrich says AI is developed by the business area, so it is no longer just an IT task. The acoustic anomaly assistant only works because engineers and AI experts spent months teaching it the noises of a car part.
A Berlin lab specialist wrote an app that could identify coffee types by sound. R&D staff, who couldn't listen to doors overnight, saw the same idea could detect door noise anomalies, and it spread to several R&D areas.
About 200 people work on the AI program, linked to others across the company, and all IT staff were trained for a shared view. Some engineers moved into IT to run the AI platform alongside IT colleagues.
Engineering already has people who understand technology, but finance has less technical knowledge and needs more outside hires. Across Ulbrich's organization, about 40% are new people and 60% were already there.
On the production line it supports workers, while in engineering and design it works like a colleague, and in order management it forecasts demand. Forecasting is hard in markets like China, where it is difficult to know what customers will buy in three months.
A specialist at Porsche Digital's Berlin lab built an app that could tell from sound whether a coffee machine made a cappuccino or an espresso. That inspired an AI system that listens for unusual noises in car doors.
No. Ulbrich says AI is developed by the business area, so it needs a collaboration between AI experts, business experts and the machine, which differs from normal IT projects.
Ulbrich says about 200 people work on the AI program. In his IT organization about 40% are new hires and 60% were already at the company, with the mix varying by department.
Designers use AI to reshape digital car models before building the first physical model. In production and sales, AI helps predict orders so the right cars reach dealerships, which is hard in China.
Ulbrich says reactions vary. Technically minded colleagues see how it helps them do their work better. Sam Ransbotham notes that introducing specific, narrow applications helps people understand the tool without fearing abstract AI.
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Mattias Ulbrich is both CIO of Porsche and CEO of Porsche Digital, a Porsche subsidiary focused on new technologies.
“I'm the CIO of Porsche and at the same time the CEO of Porsche Digital, a subsidiary of Porsche, where we really focus on the new stuff”
Porsche Digital has locations in Atlanta, Silicon Valley, China, Tel Aviv and Germany.
“we are in in the United States, like Atlanta and Silicon Valley, but we are as well in China, for example, we are in Tel Aviv, and of course, in Germany.”
Porsche's US headquarters is in Atlanta, which is why Porsche Digital operates there.
“the Porsche headquarter is there, right? That's the reason why we are there, because we have a very close collaboration with the headquarter in in the United States, in Atlanta”
Ulbrich began his career at Hewlett Packard in the 1990s.
“I started my career at uh Hewlett Packard a long time ago in the nineteen's.”
Ulbrich joined Audi at its Neckarsulm plant, where the A8 is built.
“Then I had the opportunity to join Audi at the time in Neckarsulm. This is a plant where the A eight, for example, is built.”
Ulbrich served as CIO of SEAT in Barcelona.
“I had a great time in Barcelona, where I was the CIO of SEAT at that time.”
Ulbrich spent six years at Volkswagen Group headquarters in Wolfsburg, responsible for worldwide IT services.
“After that, I had six years in Wolfsburg, the headquarter of the Volkswagen Group. I was responsible for the IT services worldwide.”
Ulbrich was CIO of Audi for six years.
“And after that, I became the CIO of Audi. I took this position for six years.”
Before joining Porsche, Ulbrich studied digital transformation at MIT and in Saint Gallen, Switzerland.
“So I moved to MIT and Saint Gallen here in in in Switzerland to learn more about the term of, let's say, transformation, digital transformation”
At the end of the 1990s, cars contained many electronic devices that initially communicated poorly with each other, which drew Ulbrich to automotive work.
“it was a time at the end of the nineteen's, where there were so many different electronic devices in the car.”
Porsche uses AI to predict orders so the right cars reach dealerships, which Ulbrich says is difficult in China.
“how we can predict for order to deliver markets, uh, to have the right cars in the dealerships. That is not easy, for example, in China.”
Porsche's design studio in Weissach uses AI to support design ideas.
“I was in the design uh studio here in Weissach uh two weeks ago, and we had a long discussion how they use AI to really support their design ideas.”
Porsche designers can reshape digital models with AI support before building the first physical model.
“they can use different models and and really reshape the digital model before they go into the first physical model.”
Porsche has an acoustic anomaly assistant that works with engineers to understand noise anomalies such as those in car doors.
“we have, for example, an acoustic anomaly assistant that works really with the with the engineer to understand um anomalies, uh, that we have, for example, in the door”
A specialist in Porsche Digital's Berlin lab wrote an application that could identify the type of coffee a machine made by listening to it.
“an IS specialist uh wrote an application that could listen to the coffee machine and know what kind of coffee is done.”
Porsche R&D staff said they could not listen for door noises overnight, which motivated the acoustic AI project.
“we are listening to our door all the day, but we can't listen at the night, for example, because we are not there.”
About 200 people work on Porsche's AI program.
“the AI program is, um, let's say, it's 200 person that really work on that in in the Porsche organization”
Ulbrich trained every IT person in his organization on AI.
“I started to train really every IT person in my organization.”
Some engineers moved into Porsche's IT organization to help run its AI platform.
“The they they moved from engineering to the IT to drive this as a platform. We have an AI platform”
Ulbrich says about 40% of his organization are new people and 60% were already at the company.
“we have maybe 40% new people and 60% that were already in the organization.”
Ulbrich says finance needs more external hires for AI because it has less technology knowledge than engineering.
“In finance, for example, you have totally different approach, and you have to bring more external people to really drive those ideas”
Ulbrich cites the Porsche Taycan as an example of Porsche's effort to put better cars on the road for sustainability.
“Porsche is doing a lot to to getting better cars on the road, like the Taycan, for example”
The next episode features Arti Zegami, head of AI at H&M.
“Join us next time when we talk with Arti Zegami, head of AI at H and M.”
Ulbrich explains his dual role as Porsche CIO and Porsche Digital CEO and the international sites where the subsidiary works on new technology.
“we really focus on the new stuff, for example, for AI, for cloud technology”
His career path through HP, Audi, SEAT and Volkswagen explains why he links cars, IT and production knowledge.
“I started my career at uh Hewlett Packard a long time ago”
Examples show AI predicting demand for dealerships and supporting designers before physical models are built.
“it's sometimes it's uh really a challenge to understand what kind of cars they would buy”
Ulbrich says AI differs from normal IT projects because business experts and AI experts must build it together, as with the door-noise assistant.
“it's not an IT task anymore”
A coffee-listening app led to an acoustic assistant that monitors noises when engineers are not around.
“that came really from drinking coffee”
Porsche trains all IT staff, runs a roughly 200-person AI program and balances new hires with existing employees.
“we have maybe 40% new people and 60% that were already in the organization”
Ulbrich says working with people motivates him more than any gadget and sees AI as key to societal challenges.
“I believe really that AI is the the biggest challenge that we are facing right now”
Sam Ransbotham and Shervin Khodabandeh highlight culture change, narrow use cases and creativity applying a technology in a new field.
“I really feel like Porsche is getting it.”
What does coffee have to do with artificial intelligence?
In this episode, Mattias Ulbrich will talk to us about how Porsche is driving a culture of innovation and digital transformation with AI across all functions in the company.
Welcome to me, myself and AI, a podcast on artificial intelligence and business. Each week we introduce you to someone innovating with AI. I'm Sam Ransbotham, professor of information systems at Boston College, and I'm also the guest editor for the AI and Business Strategy Big Idea program at MIT Sloan Management Review.
And I'm Shervin Khodabandeh, senior partner with BCG, and I co-lead BCG's AI practice in North America. And together, BCG and MITSMR have been researching AI for four years, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and deploy and scale AI capabilities and really transform the way organizations operate. Hi Mattias, welcome to the podcast. We're really excited that you could join us today. How are you?
Uh, it's a pleasure for me to be with you. Thank you for your time.
Can you tell us a little bit about yourself and and your role at Porsche?
My name is Mattias, and I'm the CIO of Porsche and at the same time the CEO of Porsche Digital, a subsidiary of Porsche, where we really focus on the new stuff, for example, for AI, for cloud technology, for for other like blockchain, and where we are, let's say, more international. And uh that means we are in in the United States, like Atlanta and Silicon Valley, but we are as well in China, for example, we are in Tel Aviv, and of course, in Germany.
So this is a company where we really look for for digital projects, uh that we can really support within the Porsche organization. But at the same time, we are looking at new ideas in the market and looking for new business ideas and business models within this organization.
You know that I'm actually from Atlanta?
Oh, I didn't know that. No. But you know that that the Porsche headquarter is there, right? That's the reason why we are there, because we have a very close collaboration with the headquarter in in the United States, in Atlanta, where we're really looking to find the best solution for the US. market.
I like the phrase new stuff. Tell us um a little bit about how you yourself got interested in new stuff. Uh you mentioned a a fair number of new technologies as new stuff.
Yeah, I'm personally very interested in new technology. Well, I I started uh electronics when I started my career. I worked with Hewlett Packard. That was an very innovative at that time. It was very innovative organization. So I really like to to understand what are the the advantages of new technology and what can it bring to the business, to bring really technology and business together. This is what drives me and where I'm interesting in finding the best solutions.
So tell us how that background started, like how you got from from Hewlett Packard or from electronics to Porsche.
I started my career at uh Hewlett Packard a long time ago in the nineteen's. Then I had the opportunity to join Audi at the time in Neckarsulm. This is a plant where the A eight, for example, is built. And there was a lot of new stuff between the car, the IT and the production side. So it was very interesting to see all the electronics and the software in the car and how you can handle that in in the production process.
And so I learned a lot about car business, IT business and production at that time. And after that, I had the opportunity to move to SEAT, that is as well part of the Volkswagen Group as Audi is. And uh I had a great time in Barcelona, where I was the CIO of SEAT at that time. After that, I had six years in Wolfsburg, the headquarter of the Volkswagen Group. I was responsible for the IT services worldwide.
And after that, I became the CIO of Audi. I took this position for six years. And then I decided to to get, let's say, a wrap up of the new technology. So I moved to MIT and Saint Gallen here in in in Switzerland to learn more about the term of, let's say, transformation, digital transformation, about new technology. And after that, I came to my role here at Porsche to drive the transformation of of Porsche and as well getting all this stuff as well from this uh Porsche digital organization into the Porsche organization.
You mentioned your interest in AI, you mentioned your interest in technology and innovation. Why cars?
Yeah, that's a good question, because I thought when I started with Hewlett Packard, it was really interesting. And I saw a lot of different branches and and companies starting from trucks and going to ships, for example, as well for other things like chemical stuff. And then I had those projects with Audi, and it it was so interesting to see what technology really did to the car. So it was a time at the end of the nineteen's, where there were so many different electronic devices in the car.
And and it was a huge challenge to manage that because in the beginning they didn't communicate well with each other. And so the there was a huge technology challenge. And and this was really interesting for me to understanding this challenge and bringing together production knowledge, IT knowledge and and the car technology together. And so that was the moment when I changed to the automotive company.
What kinds of new stuff are you interested in right now, particularly at Porsche?
Yeah, the main focus is really AI to to be honest. Um, of course, we are looking how we can organize as well with, um, let's say, agile working. What can we do as well in our development centers for for the car business? And of course, software in the car is a very important thing as well, because this is uh changing the world of of driving. But what we are focusing as well is to look how we can use AI, for example, to improve our internal processes.
How can we use AI to to get a better contact and a better understanding of of of our customers? That is very important for us as well. And of course, um, yeah, to look, uh, what can we do in in our product and and really increase as well our product portfolio to have digital products that we can deliver for B two C markets? Uh, like we started, for example, with some ideas in in the direction of uh supporting uh sustainability as well, like uh Porsche impact, where we developed uh a solution.
How a customer can compensate, for example, his uh CO two footprint that he drives with the car, and how we can really um increase our product portfolio for our customers and make driving more attractive.
So is there some particular example of how AI has made a big difference in a way that other technologies would not have that you're particularly fired up about?
Yeah, what we really found out for for example in in production, that we can really use AI to predict, of course, a lot of things, uh, for example, in in the order management systems, how we can predict for order to deliver markets, uh, to have the right cars in the dealerships. That is not easy, for example, in China. So it it's sometimes it's uh really a challenge to understand what kind of cars they would buy in three months.
So it's very important to get feedback from from the market and and understand what is the the the most important driver for that. But as well to understand what can we do in the production uh process, for example.
These are some very, very great examples of the use of AI. You mentioned a whole bunch on the marketing side, a whole bunch on the supply side and supply chain and also production. Are there also some examples where roles or functions that are typically more like engineer driven, like design or some of the trade-offs in terms of performance, et cetera, are aided by use of AI? Cause we see that in other industries, which is sort of a new thing, because these are typically things that engineers have a strong sort of discipline and playbook for.
Uh, is that something you can also comment on?
It's very interesting that you mentioned design, for example, so because uh I was in the design uh studio here in Weissach uh two weeks ago, and we had a long discussion how they use AI to really support their design ideas. And they have a program that they are using for new cars, for example, they can use different models and and really reshape the digital model before they go into the first physical model. And and they are using AI to supporting the designer, for example.
But as well, we have, of course, in other areas of engineering as well, uh, support tools that really use AI to improve uh technical development.
When you guys implement these AI tools in these areas where a strong collaboration between man and machine is needed, do you find there is a fair amount of sort of culture change needed or reskilling or change management needed for the human to become friends with AI?
I think the collaboration between the machine, the IT machine or the AI machine and and the person is very important, and you need to understand how you can use that. And it's totally different if you go to production, for example, or to engineering, because sometimes it's a supporting tool and and uh it supports the worker, for example, in the production line. In the other area in engineering, they are, they are like friends, li like colleagues that work on the same issue.
And it's very important uh to understand that AI is further developed by the business area. So it's it's not an IT task anymore. It's it's really a collaboration between the AI expert, uh, but as well the business expert and the machine. And and you have to to manage that. So this is a huge difference to normal IT projects. So we, we have a totally different approach to to go together on on such pilot projects. Uh, we have, for example, an acoustic anomaly assistant that works really with the with the engineer to understand um anomalies, uh, that we have, for example, in the door, and how he can really understand.
Wha- what is the reason for the noise? And and this support, it's only possible if you worked really together for months to understand all the noises, and then you can adapt it to a different part of the car. But it it takes a while, and you learn, you, you have a learning curve like this, and and it's only possible if you have a very good collaboration.
When you implemented this system, what were people's reactions? Did they say, Oh my gosh, this is taking my job? Did they say, Oh my gosh, this is a great way to do this? How do people react the day that you turn the system on?
Yeah, like always, people react totally different. As of course, yeah, we have a lot of very technical oriented guys that see the the opportunity and, uh, see really how this can help them in doing their task better, doing their work better.
You mentioned a little bit about, you know, the fears that some people have about when you say AI in general, and they think a machine will rule versus machine vision to help understand the right label or the acoustic anomaly system on the doors. Like, how did you come up with these sorts of like the acoustic anomaly system, for example? I wouldn't have thought of listening to a door, I would have thought of looking at a door.
How do you, how do you get someone to think about listening to a door? My doors haven't said anything as far as I know, but maybe I'm not, I'm not listening.
Listening to the door is already a task that has been done in Porsche and at Volkswagen, for example, in the whole industry for years. But the idea to use AI to really improve that noise that a door makes, that came really from drinking coffee. Now it was in the lab uh that we have in Berlin, an IS specialist uh wrote an application that could listen to the coffee machine and know what kind of coffee is done.
So he knows this was a cappuccino, this was an espresso, for example. And then there was the idea, what can we do with that with this acoustic system? And and then we had some discussion with our R and D people, and they said, Well, we are listening to our door all the day, but we can't listen at the night, for example, because we are not there. And sometimes there's a noise and uh, we're coming back in the morning.
So together we we started this project then, and, uh, it was the beginning of a success story. Because right now we are doing this in several places in R and D, and it was a good example how it can work.
Yes, for sure, and the step to really educate. I'm sure it really, really helps too, because, as you said, lack of information could create a lot of anxiety. And so how widespread are these programs? Is it tens of people or hundreds of people or thousands of people?
Yeah, how many people in the organization need to know about these technologies? Everyone?
I think, yeah, I I started to train really every IT person in my organization. So that we have a broader view on that and a common view on that as well. But of course, the AI program is, um, let's say, it's 200 person that really work on that in in the Porsche organization, and that are connected, of course, to others. But you need a very small core team that is driving this change and and bring the best use cases in place and really tell about that.
And and so we have. We have a good mixture of business people that drive that, and some people came as well to the IT organization that was new. The they they moved from engineering to the IT to drive this as a platform. We have an AI platform, and and they had really fun to to drive this program together with the IT people. And so we have a good mixture of business and IT people that are driving this uh program.
That's great. Thanks for clarifying that. And then another question I had is you mentioned 200 people. How many of these people are new people or reskilled people? Because, of course, some of this training is to educate, but some of it could also be to really build new skills. Do you find that you have to bring talent that may not exist, a massive amount of that talent from outside, or do you find that it's more about reskilling and training the existing workforce?
It's different in the different business areas. For example, in engineering, you have a lot of good trained people that understand technology and that really understand as well the possibilities of AI. In finance, for example, you have totally different approach, and you have to bring more external people to really drive those ideas because there's no technology knowledge on on on the business side. So if I look at my organization, we have maybe 40% new people and 60% that were already in the organization.
But some are, of course, really focused in in the past as well to new technology, so they are very adaptive and they would like to learn. So for us, it's a very good mixture because you need really to understand the business process to find the best solutions. That is very important as well.
It's all about continuous and ongoing learning at all That's right, yeah, levels, right for people, for the algorithms, for for everybody.
yeah, that's right. And creating a culture that is really open for that, that is very important as well.
So what are you excited about? I mean, you've you've mentioned new stuff, the new stuff from the beginnings of of uh Hewlett Packard days to new stuff now. What's the next new stuff that you're excited about?
Yeah, the the most thing I'm excited about is uh to work together with great people, to be very honest. I I like technology, of course, but to to work with people, it's uh it's the best thing you can do. And, uh, really, to learn together as well. So we are just in the beginning of AI, so there's a long way to go because we have done some really great project, but we are still having a lot of things to do with AI.
And I I believe really that AI is the the biggest challenge that we are facing right now, still, and all the other things are not such a game changer like AI is right now in in the technology field. And and I believe that AI and as well other technologies, but AI in a very special situation, can really drive the change that we have right now in in society, all the challenges that we have in society, for example, sustainability.
And I believe that digitalization can really help, uh, in in those very important fields. And if you look, uh, Porsche is doing a lot to to getting better cars on the road, like the Taycan, for example, to to have better cars. That really helps, uh, as well the environment and doing a lot in in terms of sustainability. And I think those things are driving me to to really creating a better world and and looking what can technology really do for that.
And, of course, supporting people to really focus on the most important things in life, but as well helping the world to getting a little bit better. And I think this is a great motivation for me, and this is very helpful for me.
Mattias, thank you for joining us today. Uh, we appreciate all the time you spend in the interesting conversation.
It was great with you. Thank you for your time.
Thank you very much. Yes,
That was great. Mattias was really interesting.
Mattias hit a lot of very key points that we we see in our in our work, we see in the survey this year, we saw it last year. He made the point around culture of innovation and and creation of new ideas. He made the point around it's not tech, it's about business and technology coming together. He also, you know, underscored the importance of, I mean, just by being who he is and the role he has, the the importance of this being a senior executive role.
Uh, and then the other thing I really liked is the point he made about different roles of AI. I mean, we say that in our report that there are at least five different modes, and then he talked about it. He talked about vision and automation of that. Then he talked about design, where AI gives some ideas to the engineer or to the designer. He talked about AI as generator of insights for supply chain or for the marketer or for the, you know, product designer.
I really feel like Porsche is getting it. It's really, really, uh, impressive.
Yeah, Mattias had a nice combination of excitement about technology, but also a methodological approach towards it, which I thought was a nice combination. He he both exuded excitement about technology, but at the same time, a very realistic view about how to implement it,
Yeah, the other thing I was quite impressed is the extent of culture change that Porsche is undertaking that he underscored with training, with education, with reskilling, with getting a base level education for everybody in his organization. And it's hundreds of people, not a handful of people.
right? And they have to because of the number of applications that they're. I mean, you, you listed off dozens of places that they're using it, and they were just all across the organization. When he was talking about introducing it, he said, Oh, yeah, we don't talk about bringing in AI as a big, scary, general thing that's abstract, and who knows what baggage it's bringing with it. He said, We're bringing in this application, it is very specific and very narrow, and then people can get their mind around what that means versus bringing in a lot of baggage from every sci-fi movie that anyone's ever seen.
yeah. The other thing I really liked is the coffee example. Uh, it was a great question you asked, you know, why sound? And I, I didn't think he's gonna go to coffee. I thought there's a million other places he could go, but that's a great example. And I think we also heard that from some of our other interviewees that a success story or an or an anecdote, or or an example somewhere else translates into a completely a new idea in a completely different field.
Yeah, that's a the very human role of of still, that creativity, of recognizing that this situation well isn't exactly where we saw that technology before, but it's a great place we can use the technology now, and that's still a very decidedly human creative part. But it still requires expertise, and he was very clear about that, that you, you had to have that expertise as a as a baseline. And it sounds like they're working very hard to make sure that most people in the organization have some expertise in there.
You want to ask him about new technologies. He's excited about he didn't bite on rattling off the the latest, greatest new AI thing. He's he's all about getting the people excited about it. That that was not where I thought he would go there.
yes, that's right.
We really enjoyed speaking with Mattias today. Uh, he really combined an enthusiasm for technology with an enthusiasm for business, and that was a fun combination.
Join us next time when we talk with Arti Zegami, head of AI at H and M.
Thanks for listening to me, myself and AI. If you're enjoying the show, take a minute to write us a review. If you send us a screenshot, we'll send you a collection of MITSMR's best articles on artificial intelligence free for a limited time. Send your review screenshot to SMR feedback at MIT dot edu.