The Data Buffet: How TI Automotive is Fueling Enterprise AI

Chad Watt: Welcome to the Infosys Knowledge Institute podcast, where business leaders share what they've learned on their technology journey. I'm Chad Watt, Infosys Knowledge Institute researcher and writer. Today I'm speaking with Apurva Wadodkar, Senior Director and Head of Data and AI at TI Automotive, and Rakesh Gollapalli, Senior Vice President of Automotive at Infosys. Apurva is in charge of strategic data and AI programs at TI Automotive, a tier one supplier of fluid management systems and related components to auto manufacturers worldwide. Rakesh oversees Infosys' work with automotive businesses in North America. Welcome to you both.

Apurva Wadodkar: Thank you Chad.

Chad Watt: Apurva, I have read about your signature idea of a data buffet and yes, it made me hungry. Tell us a bit about it.

Apurva Wadodkar: Yeah, absolutely. Now me being a foodie, I had to get some sort of a food reference in my work. And really the idea here is that we are cutting down drastically on the time to get your hands on data. The north star for my data buffet is just that.

And the way we are organizing our data buffets is by cuisine. So there's gonna be a sales cuisine. There's going to be a marketing cuisine. There's going to be a finance cuisine and so on. And to support these cuisines, now we go back into the kitchen. We are creating parts. So I do have a finance part. I have a supply chain part and so on. Now we have also brought in a little bit of innovation in the way we have structured our parts as well. So there will be people from my team, the engineers, the architects, right? And then we are bringing in a business representative to support this part. And that business representative is gonna be the curator of the menu of that cuisine. And the team here that comes from my space is gonna be cooking that menu, right? So that's how the parts are put together.

Now, this does a couple of things. Because we brought in business into the kitchen, right? They're not just consuming this data, they're into the kitchen. And so now they are more likely to use this data. So adoption now becomes easy, right? Because hey, it's for you, by you, right? And so adoption is, we don't have to try as hard for adoption.

Now, coming back to the buffet itself, you know, if you've seen a real buffet, it's all, standardized, isn't it? You see the same sort of bowls in which things are served. And similarly in data buffet also, we need to see that standardization, regardless of the buffet, cuisine, or regardless of the data product we are serving, right? So standard naming conventions, making sure there are descriptions associated with every data product and so on. All these data products are hosted in a catalog, so it's easily findable, stuff like that. And we see little things, but really enhances the user experience, right? So that's my data buffet.

Chad Watt: Great. Thanks so much for that. Rakesh, what is your experience with tech, data, and AI in the automotive sector?

Rakesh Gollapalli: Thanks for that question. And Apurva, you really nicely put it on how the whole data arrangement and buffet, I really like the analogy that you gave. Coming back to the automotive industry, this industry is a very, very competitive industry. And just by the nature of the product they make, the customers they serve, and the industry in general, they have been early adopters of any technology. And it has been no different even in the AI way.

However, this industry has gone through a bit of reality cycle check. Essentially, they jumped aggressively into the AI across autonomous driving, generative AI, smart factories, engineering, everything. But what many of them discovered very quickly is enterprise is as good as the data underneath it.

So today we are seeing a clear shift. While companies are not slowing down on the AI investment, but they're rebalancing towards a data foundation because that's become the biggest bottleneck to scale any meaningful ROI and any meaningful benefits out of it. Across OEMs, you know, tier ones, tier twos, their suppliers, the challenges are consistent. Data is highly fragmented. It's across engineering suppliers, the supply chain, manufacturing, the vehicle itself, the dealer systems, so on and so forth. No common data model exists across the complete value chain. AI pilots work, but they don't scale. So the industry is moving to a data first AI at scale model.

Essentially what we are seeing is we are seeing more investments in enterprise data platforms and end-to-end data integration, data governance, and building the digital thread from design to service so that they are able to track it completely. So, in summary, I think AI is the ambition, but data is the foundation. And that's kind of where the industry is right now.

Chad Watt: That's really great. We've talked about the data buffet. We talked about data and AI and Apurva, you have written on this recently on CIO.com. You argue that just investing in that data foundation isn't enough for companies to see a good ROI. Why is that?

Apurva Wadodkar: Absolutely, I love what you said Rakesh, AI is a step two and it is essential, we build the foundation right before we step onto AI. And we see this all the time in companies over enthusiasm about AI but hey data investment is not there.

But to take it even further down into this foundational space which is the data space. I think there needs to be a little bit of a mind shift there as well. I have seen a lot of companies heads down, building their data foundation, and really data is not a competitive edge. It is the decisions that that data is supporting is the competitive edge, right? So again, coming back to north star, you know, eyes on prize, which is decisions.

So when I ask people, hey, what are the decisions you are supporting with the data products. And you know, we are so engrossed in creating data products, we lose sight of that. And so I have created the decision intelligence framework, which really puts those decisions in the limelight.

You know, theoretically it's easy to get to it to understand, but it's the practical, when you come to practical, it takes a little bit of a work. So the way it works is you get a full vantage point of your company from different departments and see what are those decisions that are driving these departments and hence driving the company towards its goals, right?

So I would meet with department leaders and they don't have to be top, top leaders. They are the people who are in the zone. They're doing all the work to make these decisions and ask them what are your top three decisions that are guiding finance, supply chain, right? Manufacturing, what is that? What are the top three? And then start creating that inventory of decisions that are happening.

Once you have these inventory, you also now deep dive into, okay, what are the outcomes you are expecting, right? And now with all that information, how many times are you making that decision? What does it take to make that decision? So really it's an inventory of decisions that are building your company or making it successful, right?

And now what we do is we plug in data products first to make these decisions better, right, to support these decisions.

And then in the end to measure if the outcome of these decisions is desirable or not, right? And this way creates a feedback loop which is a circular feedback loop. This way, you're improving decision every time, you're measuring outcomes, and it's all balanced. Now the data products you're building to support these, or AI products, or applications, it could be anything, is what's now driving this engine, right?

Chad Watt: That's terrific.

Apurva Wadodkar: Exactly.

Chad Watt: Great. So, Apurva, you just say data, decisions, outcomes, and circling back around to that data. Give me, that's a great high-level abstraction there. How are you putting artificial intelligence to work in TI Automotive?

Apurva Wadodkar: Yeah, so I'll tell you AI has been so much democratized in recent years and we want to really ride that capability. And so far we have seen only a limited, the AI COE doing AI stuff. I want TI Automotive to come out of that and we wanna bring this capability to masses.

And so earlier this year we opened up enterprise ChatGPT, Microsoft Copilot, for our employees. And then one other thing you want to realize is just bringing in a tool is never sufficient. Because you're in IT, it comes naturally to you, but this, I call it a technology bump. It's a real psychological bump in people's head that, oh my God, this adoption, how do I even start doing this? And so... you have to really hand hold them, take them through this bump and now run as fast as you can, right?

And that means you have to invest in training people. Education is a strategic objective of an AI center of excellence, really, I would say. You want to empower people to come out of their mundane tasks and just automate it or do it faster. So they now are open for much more creative work, right? And it's a shift in culture. It's a shift in, you know, how your company sees things.

And so we did a lot of trainings for ChatGPT and Copilot and simple things, prompt engineering. Like, what do I even ask ChatGPT to do? You'll be surprised, but that's what people are stuck upon is, what do I do? And so you give them all these ideas and they run with it.

Now, another thing we brought in was the Codex feature or any sort of Copilot you might have for development folks.

Nobody wants to spend time in debugging and code refactoring. I put this in the mundane category. What is more interesting is building that new logic and making things happen. That's where we want to focus our people towards instead of this piece, right? And what better than AI to do it?

Chad Watt: Right. Yeah, it's one thing to give people a tool, but you have to deliver that with some training and context. Data, decisions, outcomes, trainings. Rakesh, let me turn to you. Where in the automotive sector do you see AI delivering value right now?

Rakesh Gollapalli: Just queuing on what Apurva mentioned, we are seeing a proliferation of AI across all automotives, OEMs and tier 1, tier 2s. AI is already delivering value today across many areas. But in my mind, it's not comprehensive and we can do a lot more.

So we are seeing the biggest impact is where data is strong, where problems are repeatable, ROI is measurable. And I'll give you a few examples across the value chain. For example, in manufacturing, we are seeing big wins in predictive maintenance, AI-driven quality inspection, product optimization. Similarly, if I go through all the other areas, after sales service, it's about predictive service, AI diagnostics, dealer support systems. On the engineering side, it's helping accelerate innovation by generative design, simulations, and you know, the whole automotive world is moving towards software-defined vehicles.

And what that means is the content of the software is more than the content of the hardware, or rather the value of software is more. So in this case, it's about acceleration of software deployment or called feature delivery onto the vehicle. We are seeing a lot of use cases around there. Connected vehicles. You know, AI is driving hyper-personalization and we are seeing some advanced use of AI, in-car AI assistance, delivering and usage-based services.

Again, on the autonomous driving. A lot of investments going on there, especially as we are starting to move away from sensor-based technologies to camera-based, we see adoption of AI across the board. So if I connect both the questions you asked me, the first one and the one here, AI is delivering value across manufacturing services, supply chain, in-car experience, autonomous.

But I think the challenge has been scaling that value across the complete enterprise. And this truly depends on data.

And that reality is a shift we are seeing now. People are moving away from AI experimentation to data-led AI transformation. So in my mind, I think the winners in automotive won't be the ones with the most pilots or the most siloed use cases. They will be the ones who can scale AI across the complete value chain, that too powered by a strong data foundation.

Frankly, AI is no longer about efficiency, but it is becoming a source of differentiation in the automotive industry. And this is where things get really interesting for someone like Infosys, because we have the ability to help our customers, like we are doing at TI Automotive, in helping build a strong data foundation and then accelerate the AI deployment at scale, leveraging the strong data foundation.

Chad Watt: How can a company use AI to differentiate itself?

Rakesh Gollapalli: The car becomes software defined and a learning system and just not a static product. Now, if I look at on the customer sales process side, on the experience side, AI is driven sales journeys, AI-driven sales journeys, hyperpersonalized advertising and the offers that we put out, predictive maintenance. So a lot of these initiatives could be big differentiators for companies on the AI side.

Again, we talked about on the engineering side, faster design cycles, AI led simulations. If you look at Formula One, for example, they make everything virtually before they make it physically. And that's done leveraging a lot of AI use cases or AI technologies. Continuous feedback loop from vehicle information. And that's going to help drive innovation internally on the engineering side.

In essence, companies that use the vehicle to learn and train proprietary models will out-compete or will out-innovate competition. You know, new revenue models, companies are differentiating by offering a feature as a service, usage-based models, ride as a service, so on and so forth. So in my mind, AI will enable recurring software like revenue streams instead of a one-time sale.

Chad Watt: Anything in the context of AI that has really just wowed you recently in the automotive world?

Rakesh Gollapalli: I think for me, what is fascinating is what is happening on the autonomous driving side, especially with availability of strong AI models and advances that are being made. The fact that we are now moving away from sensor-based navigation to more camera based or visual based navigation. It just opens up a very, very big field. And I heard from several of my clients that some of the biggest constraints was availability of next generation AI so that they could make autonomous driving even more autonomous, even better, even more reliable.

Chad Watt: Apurva is anything you'd like to add on what's wowing you or anything else that we've missed.

Apurva Wadodkar: I think Just the general awareness amongst engineering folks and our manufacturing teams is they are installing sensors everywhere, just in anticipation that, oh, we might need this data for some sort of an AI and so really that intuitiveness amazes me. This was not the case a couple of years ago, right? People were not as savvy as to think, this might help me someday. So I'm seeing quite a bit of that in our plans.

Chad Watt: Thank you all for joing us today.

Apurva Wadodkar: Thank you for having us Chad.

Chad Watt: This podcast is of our collaboration with MIT TechReview in partnership with Infosys Cobalt. Visit our content hub on technologyreview.com to learn more. Be sure to follow us wherever you get your podcasts. You can find more details in our show notes and transcripts at Infosys.com/IKI in our podcast section. Thanks to our producers, Christine Calhoun and Yulia De Bari. Dode Bigley is our broadcast engineer. And I'm Chad Watt with the Infosys Knowledge Institute, signing off. Until next time, keep learning and keep sharing.