Stop Waiting for Perfect Data: AI in the Automotive Supply Chain

[Transcript]

[00:00:00] Jan Griffiths: This is the Auto Supply Chain Champions Podcast. We are on a mission to bring you real conversations with the leaders who are transforming supply chains in the automotive sector. These leaders are true champions of manufacturing, and we're here to share their stories. I'm Jan Griffiths, your host and producer, and I'm joined by my co-host, Tom Roberts, Vice President of Strategic Industry Development at QAD.

[00:00:30] Tom Roberts: Great to be here, Jan. What I see every day is simple: manufacturers don't have a data problem, they've got an execution problem. This show is about how artificial intelligence, systems of action, and empowered teams can help close that gap.

[00:00:45] Jan Griffiths: Let's get into it. This podcast is powered by QAD RedZone.

[00:00:53] Hello and welcome to the Auto Supply Chain Champions podcast. This is our Q3 review episode.

[00:01:00] Let's check in with my co-host, Tom Roberts. Tom, what have you been up to?

[00:01:04] Tom Roberts: Hi, Jan. Lots of great things going on. We had Champions of Manufacturing in Chicago. We've got Champions of Manufacturing Munich coming up. Great event in Chicago. it's getting bigger every year. We had Scottie Pippen as a special guest, interviewee with Sanjay, and he's a Hall of Famer, great player.

[00:01:23] You know, Michael Jordan, of course, I think is one of the greatest players ever, but he didn't really start winning championships until he had a great team around him, and Scottie Pippen, I think, was part and parcel to that. So it was a really great event, really great. Even though he's a Chicago Bulls player and I'm a Detroit Pistons fan and that's always a tough thing, but great to see Scottie up on stage and he's a great guest.

[00:01:45] Jan Griffiths: There is a lot that we can learn from high-performance athletes about leadership and teams and coaching. When you start to view leadership Through the lens of being a coach instead of the person that's just telling people what to do, it's a totally different experience. And it's funny you mention basketball, and you know I'm not a sports fan of any sports whatsoever, but I remember when I first came to this country, to Michigan, actually, I came to Indiana first and then to Michigan, and Isiah Thomas was always in the news.

[00:02:14] It was right at the peak of his career. And, I was always intrigued by it. And you know what? I actually get to interview him on stage. ISM, Institute of Supply Management, has an event December 2nd in Detroit, and we'll put a link in the show notes if people want to attend. But I'm gonna be interviewing, Shilpan Amin from GM Purchasing, and a gentleman from Karmanos, and Isiah Thomas all at the same time, and we're gonna be talking about my favorite subject, Tom — leadership.

[00:02:44] Tom Roberts: That's awesome. Awesome opportunity to interview Isaiah. I can remember being a high school student and my friends and I would take those, the $4 Elias Brothers tickets and go to the Silver Dome. They were still playing at the Silver Dome, and I got to see Isaiah a few times.

[00:03:00] Jan Griffiths: Yeah, no, that's interesting. Well, let's talk about Q3, shall we?

[00:03:06] Now, Tom, there are three themes are predominant in Q3, and that is data, I don't think that's ever gonna go away. Data, resilience, and leadership.

[00:03:18] So let's start off, Tom, data. What have we seen in Q3 and what are we looking forward to in Q4?

[00:03:25] Tom Roberts: Well, it's funny that data always comes up because it's affecting each of these particular process areas or compliance areas. We had Josh and Ian back on in July, and whenever we're talking about USMCA or we're talking about tariffs and trade compliance, you have to have great data.

[00:03:45] And, they hit on that. And, when Ambrose came in from Seraph, data, data was incredibly important. And the same thing we had Ted from UHY Consulting, data is important. If you are trying to do cost engineering, if you don't have great data, then you may as well forget about it because you can't attack the concepts without the great foundation of data.

[00:04:06] Jan Griffiths: But what I find interesting from a leadership perspective, as leaders look at data, we are never gonna have perfect data.

[00:04:15] I know you should never say never, but it's... Tom, it hasn't happened in our lifetime.

[00:04:19] Tom Roberts: Right.

[00:04:20] Jan Griffiths: It's gonna get better, but perfect data is not something that's gonna happen anytime soon, and what I see happening is that sometimes people are gonna say, "Well, we'll wait, we'll wait. We'll wait until we get the data cleaned up. And then we'll do the thing, whatever they wanna do. Then, we'll put AI into it. We gotta get the data right first."

[00:04:39] What do you say to those leaders, Tom?

[00:04:42] Tom Roberts: So this is, It's like everything else. There's gonna be a balance, right? So can AI build cross-reference tables very quickly? It can. Is it perfectly reliable to do so? Let's talk about commodity codes, Jan. So you have primary commodity codes. Everybody wants to put something in different, like someone will use the UN code, somebody will use their own code, somebody will use a number, somebody will use a word like steel. Everybody's gonna have their own thing.

[00:05:10] Can AI look at probabilities, look at old data, look at some old pieces of information and make suggestions about what the commodity code cross-reference should look like? It can, but you've gotta create somewhat of a foundation for it.

[00:05:28] The other thing that you can do with AI possibly is can you extract your data, do an ETL function where you extract it, you clean it up, and then you reinsert it into your system? Again, the minute that you do that, if you don't have a great process moving forward, the first person to enter something, it goes off the rails unless you've got some good poke yokes and, other good process steps in there for it to be disciplined.

[00:05:54] You're right. All that come together at companies with thirty thousand employees, seventy thousand employees, twenty plants, fifty plants? I don't know if that'll ever happen until, until it's all truly automated. But I think it's worth the fight and it's worth, at least understanding the state of your data, and I think a lot of companies don't even understand that they have...

[00:06:14] They might call steel ten different things, in their system. You just have to at least have a good fundamental grasp of it, I think, at the very least. And any effort to clean data is a good effort, in my opinion.

[00:06:27] Jan Griffiths: Yeah, I agree, but it's a little uncomfortable from a leadership perspective to start to execute projects with AI, when AI is using data that's not perfect.

[00:06:38] What I'm hearing, Tom, and tell me what you heard when you were at Champions of Manufacturing, but got to start.

[00:06:45] You've got to start somewhere. We can't wait.

[00:06:46] Tom Roberts: Yeah. 'Cause right now, I mean, with the way the tools are changing, and you and I were just talking about this. With the way the tools are advancing, if you don't understand concepts like just simple things like how do you share an AI project or effort between multiple people inside of a enterprise tool or, how do you prompt properly?

[00:07:08] Because you have to have some creativity to prompt properly, to really get what you want, or you have to prompt the system to write the prompt for you, and you've gotta be good at that. So I think that people who are advancing their knowledge to understand these basics as we move along this linear flow. Well, it's probably not very linear, but, as we move along in this, I think the people who are advancing their knowledge, they're gonna have a huge advantage over the people who are just like, "Yeah, I'll just kind of wait until it's all coalesced together and it's all great and we can move on."

[00:07:40] I think the people who are taking steps now are gonna have an advantage.

[00:07:43] Jan Griffiths: I agree, and there's different levels of understanding of AI. And I think that, most of us have played around with ChatGPT or Claude, and I thought I was doing great when I could figure out how to do a Claude project and co-work and all this stuff.

[00:08:01] And then, I joined an AI boot camp, and I'm only on the second session

[00:08:06] Tom Roberts: Oh, nice

[00:08:07] Jan Griffiths: I'm only on the second session, and I've already learned so much, but it's making me realize how important it is for leaders to build their knowledge on AI. You don't have to understand it down to the coding level, but you've at least got to understand its capability and how you can use it in your business. And that's why I'm glad that we had Cheryl Thompson on here to help talk about that, and we probably need to bring her back.

[00:08:34] Tom Roberts: Absolutely.

[00:08:35] Jan Griffiths: This morning actually, I was looking at a LinkedIn post from Sean Gaskin. Now, if you don't know Sean, he is a genius, actually, I think he is. He is the head of purchasing supply chain and logistics IT for General Motors.

[00:08:50] And he just posted that he completed the professional certificate in AI and machine learning through Michigan Engineering, which I think is a very, very intense course that he put himself through. And this is what he said, his biggest takeaway: "AI leadership requires more than understanding the concepts. You need enough depth to know what's possible, where it creates value, and what it takes to scale." And I thought, "You know what? You're absolutely right."

[00:09:24] So I think that's something that we need to provide to our audience as we look at Q4 and rolling into '27, Tom, is what exactly do you need to know and how do you get it? How do you get that knowledge?

[00:09:37] Tom Roberts: Right? And it's one thing to just kind of use it and, it's like going to play basketball or something. You go, you play, there's only certain movements that you're gonna make during a game. But if you don't practice, if you don't take the 100 free throws, and you don't take the 500 three-pointers and all those kinds of things, you don't put in that kind of work, you're not going to really get as good at it as you could.

[00:09:59] I just think you're right. I think that the more training steps that we take, along with the real-world application of it, I think that will be the best scenario possible.

[00:10:07] Jan Griffiths: Yeah, real world application is everything, right? And, I like, what Sanjay talks about, proof over promise.

[00:10:15] Tom Roberts: Yeah, absolutely. 'Cause there's so much, AI noise out there, that AI is doing this and AI is doing that, and then we get some news thing that everybody thinks that AI is gonna end the world.

[00:10:25] Jan Griffiths: Yeah

[00:10:26] Tom Roberts: But, it's the difference between a news article that says, "We think AI is gonna end the world," when a manufacturing company can't even take the most elementary step and automate it, it's just there's such, cognitive dissonance with those things.

[00:10:43] It's like, well, if you think AI is very close to ending the world, can you maybe help me automate my month-end close?

[00:10:49] Jan Griffiths: Yeah, right.

[00:10:51] Tom Roberts: Before we go destroy the world, help me do a little bit better with some of my inbound demand signals from an OEM. And I think that creates some of this dissonance out there.

[00:11:01] And that's what Sanjay's talking about is, we are trying to show people, here is where it's out there, it's working, it's in productive environments, it's in actual companies, it's live.

[00:11:12] And that means more than kind of these atmospheric claims or these high-level things that people are claiming out there that aren't really bringing advancement to their daily lives.

[00:11:22] Jan Griffiths: Yeah, no, I totally agree. And then the one thing that really made me sit up and take notice, I think you too, when we interviewed Mia and Daniel, the two students from Wayne State, who have been in the workforce just a little while, maybe a year or even less, and just enough to get a taste for it. And this idea of perfect data, they're often taught that you have the data and this is what you do with it. But then they get into the real world, and they realize it's on seven different spreadsheets, and they gotta convince seven different people to give them information. Reality hits hard, doesn't it?

[00:11:57] Tom Roberts: Yeah. Or they have to find it, period. Somebody might give them some action to perform and they're like, "I don't even know where to ask for the data." And they're like, "Well, ask Jimmy over in engineering. He might know somebody." There's that whole step of even finding out what the sources are, and then finding out there's 20 of them and nobody ever talked to each other.

[00:12:14] So you're right. The difference between that and, say some kind of use case in college, like you said, they give you the data set and then they say, "Okay, now go boil this down and come out with these particular, outcomes from the data." It's different in the business world. You gotta find it first, and then there's probably 20 things you find, and then you gotta boil it down.

[00:12:33] Jan Griffiths: And then you gotta convince people to give you what you need, and that's a whole skill set in itself too. So I think we got a lot to do bridging the gap between the world of academia and the real world, knowing how fast the real world is changing today. Yeah, no, I think it's really fascinating. Well, that's data.

[00:12:50] Now let's move and talk about resilience. Resilience continues to be an issue. We had Ambrose Conroy on talking about an office of resilience. But Tom, I mean, still every single day we're dealing with supply chain complexity, supply chain moves, black swan events, things coming at us left and right, aren't we?

[00:13:13] Tom Roberts: We are. And Jan, whenever we're talking about resilience in supply chain, I wrote that blog not long ago, the LinkedIn post about the death of JIT, and what predicated that was General Motors took steps to stockpile four and a half billion dollars of specially needed parts with Procura in Asia to hedge against these black swan events.

[00:13:36] In case something happens, they're taking steps to make sure that they've got stocked parts. Now, nobody's gonna walk away from the technology you got from JIT, right? Nobody's gonna walk away from all those connections and the fact that you can send demand signals and have all those back and forth between a supplier and an OEM.

[00:13:55] But it acknowledges, "Hey, you know what? We just gotta do something to make sure that our supply chain is resilient, that we are storing parts if we need to, and that we have that insurance against these black swan events." I mean, you see things happening in the Strait of Hormuz still.

[00:14:12] There's other things that are happening, where you have neodymium magnets, rare earth minerals. Our guest that we interviewed who's gonna be on the show coming up soon. He's a supply chain leader, and one of his areas of focus is, I wanna make sure that I've got backup somehow in case I'm not able to get a supplier to supply what we need from them.

[00:14:35] Is it alternative sourcing? Is it some safety stock? What is the approach that we need to take? Very important.

[00:14:40] Jan Griffiths: Yeah. No, it absolutely is. And from leadership perspective, it's about questioning the norms that we know are the norms in this industry. You talk about questioning Just-in-Time. No longer can we just assume that's the way we need to operate. Will it be the predominant factor the way we operate?

[00:15:01] Maybe. But is it the driving force for everything? Is having inventory in key strategic places now no more important to support a resilience strategy than just blindly following inventory reduction last, to the letter of the law? And the same is true for cost reduction, and I think Ambrose talked about this when he was on. Purchasing has been so driven towards purchase price variance, lowest product cost, are we now going to have to look at, truly look at, total acquisition cost and look at it through the lens of a more resilient supply chain rather than just pure cost? These are big leadership questions that have to be asked and answered.

[00:15:48] Tom Roberts: Absolutely. Supplier quality, supplier development, resilience in the supply chain, alternative supply. If it's possible in automotive, you can't just necessarily spool up another supplier and get a line online very quickly. But, you have got to have an answer and a lot of it is market share.

[00:16:05] There's not a single OEM, especially in the West right now, that wants to give up one micro percentage of market share because this is a bare-knuckle brawl for market share. And if you have a supplier shut you down. If you can't meet a certain, delivery quantity where customers want to buy a particular vehicle, it's an awful situation. And there's gotta be planning to help insulate against that.

[00:16:31] Jan Griffiths: Yeah, absolutely. Let's close out, Tom, and talk about the need for speed. What are your thoughts on that? 'Cause we talk about China speed all day long.

[00:16:43] Tom Roberts: Yeah, I mean, speed is key, and that's speed of movement, speed of decision-making. I had some experience at a Chinese automotive supplier. I think it was an awesome experience. What I noticed is if the decision wasn't made quickly, the follow-up was very quickly and everybody aligned to move forward.

[00:17:03] And that was breathtaking sometimes where you would have people line up behind a decision. That I think is something, we need to decide more quickly as an industry. But it's just getting behind that decision say, "You know what? We've gotta make this decision work. This is a strategic direction."

[00:17:21] If we're always, sitting on the edge and, throwing rocks in at things, there might be some very valuable rocks being thrown, but in the end, people have to line up behind those decisions and move very, very quickly.

[00:17:35] Jan Griffiths: Yeah.

[00:17:36] Tom Roberts: or they'll be left behind.

[00:17:37] Jan Griffiths: Empowerment is important. You have to empower people to make those decisions. And I would like to close out today, Tom, with a call to action to our audience.

[00:17:48] There's a lot going on in the world right now. We know this. Just take one process. One, whether it's the RFQ process, where we interviewed Ted Mabley on from UHY. Whatever it is, it doesn't have to be a big complicated process, but take an action to reduce the time that it takes to make that decision or produce that output. Reduce it by 50%.

[00:18:14] Just start somewhere. It might be the internal RFQ process, the external one. Just take something. Do something. Start somewhere. What's your call to action to our audience from a data perspective, Tom?

[00:18:28] Tom Roberts: Yeah, I would say number one, understand your data. So you may have some anecdotal thoughts that, wow, this is really not a good situation with our data. But actually dig into a little bit of the dimensionality of the data and understand.

[00:18:40] Pick something. Go look at your supplier master. Go look at your item master. Again, I always pick on primary and secondary commodity codes in purchasing because there's almost no bigger driver than understanding that because if you can align your spend, you can go talk to suppliers about scale.

[00:18:57] And when you can buy in scale, you can get a better price and, all those good things. It's just simple concepts. But if you don't understand the data, if you don't understand the nature of what you're doing, then you can't get those advantages. So I would say number one, just understand your data.

[00:19:10] If you can go further than that and start cleansing the data, that's great. But also have that process moving forward. And I don't know, depending on how big your company is, I don't know if you need to have a master data team. I have seen those be very effective.

[00:19:24] If you're gonna add a new, say that you're gonna add a new account to the chart of accounts, well, it goes through a team that asks why. Why do you need this? Why do you need this different dimensionality? Why do you need this, whatever. And ask those questions rather than kinda create them pell-mell.

[00:19:38] And usually, people are, "Okay, well, maybe this isn't something I need right now."

[00:19:43] Jan Griffiths: Yeah

[00:19:44] Tom Roberts: But a lot of the time there's no pushback. And that's when those things proliferate, and you lose data reporting dimensionality, and it just becomes a big mess. But I think that's the biggest thing.

[00:19:53] Jan Griffiths: Great advice. Well, we're gonna keep bringing content to you, our listeners. If there's something specific or somebody that you wanna hear from, please let us know. You can reach out to either Tom or myself directly. We are here to provide content for you to help you on your journey into the future as we transform our beloved automotive industry.

[00:20:16] Thank you, Tom, for joining me today.

[00:20:19] Tom Roberts: Thanks, Jan.

[00:20:20] Jan Griffiths: We wanna hear from you, our listener. Tell us what are your challenges right now? What conversations do you want to hear across the airwaves on this podcast? Drop us a comment on our podcast website.

[00:20:36] The link is in the show notes.