The Future of Automotive AI: How Technology Is Transforming the Parts and Service Industry

What happens when artificial intelligence meets automotive expertise? Tilak Kasturi, founder of Predii, takes us on a fascinating journey through the rapidly evolving landscape of automotive AI and its transformative impact on the parts and service industry.

Drawing from his diverse background, spanning from NASA's space shuttle launch systems to radiation oncology software, Kasturi explains how Predii developed specialized AI solutions for the automotive aftermarket. Unlike generic AI applications, these tools understand the unique language and technical requirements of vehicle maintenance and repair.

The conversation explores how AI connects previously siloed data across the automotive ecosystem. Parts suppliers can now track customer experiences through the entire service journey, predicting demand patterns with unprecedented accuracy. Service advisors use AI assistants that understand when customers say their "car is making a funny noise," translating vague descriptions into specific technical issues and part requirements.

Most striking is how quickly this technology is being adopted. While the internet revolution took 25 years to fully transform business practices, Kasturi predicts AI will achieve similar penetration in just five years, driven by affordable cloud computing and open-source technologies. Despite this acceleration, he emphasizes the critical importance of implementing AI carefully in an industry where safety and accuracy are paramount.

Looking ahead, Kasturi envisions increasingly personalized vehicle owner experiences powered by "human-centric AI" that builds trust through transparent reasoning. Throughout the conversation, one message remains consistent: successful automotive AI must blend deep industry knowledge with technological capability, making human experts more efficient rather than replacing them.

Whether you're a shop owner considering AI adoption, a parts supplier looking to improve forecasting, or simply curious about how technology is reshaping automotive care, this episode offers valuable insights into the future of our industry. Listen now to understand how the right implementation of AI can transform your automotive business.

To learn more about the Auto Care Association visit autocare.org.

To learn more about our show and suggest future topics and guests, visit autocare.org/podcast

Chapters

• 0:08 Welcome to Indicators
• 0:42 Meet Tilak Kasturi
• 3:43 Introduction to Predii
• 5:30 AI Throughout the Supply Chain
• 11:27 Generative AI Applications
• 20:08 Data Sources and AI Integration
• 24:34 AI Innovation Cycles
• 34:19 Future of Automotive AI
• 38:56 Challenges and Final Thoughts

Transcript

Mike Chung: 0:08

Welcome to Auto Care ON AIR, a candid podcast for a curious industry. I'm Mike Chung, senior Director of Market Intelligence at the Auto Care Association, and this is Indicators, where we identify and explore data that will help you monitor and forecast industry performance. This includes global economic data, industry indicators and new data that will help you monitor and forecast industry performance. This includes global economic data, industry indicators and new data sources. Hello and welcome to another edition of Indicators. I'm really pleased to be with Tilak Kasturi of Predii. Tilak, would you like to say hello and introduce yourself to our crowd? Hi?

Tilak Kasturi: 0:44

Hi everyone, it's a pleasure to be here. Thanks, Mike.

Mike Chung: 0:47

Absolutely. And Tilak, tell us a little bit about yourself and how you landed at Predii.

Tilak Kasturi: 0:52

I'm a technologist by background. I spent time in the industry on building enterprise-grade applications for different verticals, and part of my journey included working at startups, at venture capital firms, and that led me to explore new challenges, new problems, and automotive industry is one of the best industries to work and look at. You know how an experience for a car owner can be improved, and so that just kind of one thing led to the other and to founding.

Mike Chung: 1:42

Predi. Oh, thanks for sharing that. And just to dig in a little bit more you said a technologist and you said other industry verticals. Can you tell me a little bit more about those?

Tilak Kasturi: 1:51

So I spent a long time ago working at it's not entirely a vertical, but I was one of the lead architects building the first internet browser for TVs. That was way back in Orlando, florida, working for Time Warner Cable. Those days it was a very popular thing bringing in cool technologies to the to do video on demand and things. So I was one of the architects there and also worked at NASA and the space shuttle launch control systems and it was primarily, you know, part of a large initiative at Lockheed Martin, but working with telemetry data and it's a different scale but it's the same telemetry data that you see in automotive vehicles. So anyway, connecting the dots, past experience always you try to tap into it and most recently, before this startup, I worked in radiation oncology building another software for cancer treatment. So I try to look at different challenges and problems and that kind of helps me stay focused and energized. So automotive is an exciting challenge for us.

Mike Chung: 3:29

Thank you for that. That's so rich in terms of the different perspectives, the different contexts that you're looking at, data, the different stakeholders that may be involved. So, as a problem solver, engineer, scientist myself, I can appreciate the different lenses that you've gotten to look through throughout your career. So thank you, and tell us a little bit about Predi. How long has it been around? What kind of work does Predi do?

Tilak Kasturi: 3:53

So we are an 11-year-old company based in Palo Alto and we have offices in Pune, india as well India as well, and so, in a simple few words, predi provides automotive AI, automotive-specific AI, for parts and service industry, and our automotive parts and service industry is extremely human-centric industry. We work, you know we work with, based on the experience of service advisors, service writers, technicians, you know, doing problem analysis and then trying to come up with the best resolution. Part suppliers you know. Warehouse distributors you know we have all across the board. It's a very human-centric industry and we bring the AI layer into these workflows, and so that's what Pretty does.

Mike Chung: 4:55

So when you're talking about, say, the warehousers, the parts suppliers, is that kind of a? Tell us a little bit more about about that, because I can think about the service advisor. If I'm a technician and I'm diagnosing and repairing a vehicle, I can imagine going to some sort of resource to identify how and what to do. But could you tell us a little bit about the use case for the other audiences?

Tilak Kasturi: 5:22

So in the parts side, I'll start with the parts suppliers. Parts suppliers are always looking for what is the experience of the car owner and which led to a part failure or part-related issues that can allow them to be staying ahead of the customer complaints. That would lead to a better design of the product or also be able to estimate better demand for the product. And that usually happens. You know it's not a statistical analysis. You know we need to. From a part supplier, they would like to know if they're opening the box of that brake pad and it's actually going on to a ERMEC model engine, and that typically happens at the shop. Last mile knowledge can be brought back to the part suppliers in a much more elegant manner that allows them to know yeah, the part failure rates are the demand for parts in a specific region. So traditionally these are business intelligence driven products.

Tilak Kasturi: 6:39

Now, with the AI universe, we're trying to connect a lot more dots together, much easier than it has ever been before, whether an experience that a car owner had in the shop being able to bring at a scale back to the power suppliers to understand what was the cause, what was the complaint, what was the cause, what was the correction that led to that particular part being replaced on the car. So there's so much. You know this industry is well connected all the way and previously we were thinking in silos. Ok, the parts supplier will get you know demand from the distributors and they would actually address it, or data coming from the parts catalog. But with the help of AI, we're expanding that Instead of being silos, you know you could go all the way up to the car owner and try to bring that type of insights back to different value points in the supply chain, to different value points in the supply chain.

Mike Chung: 7:44

I appreciate that explanation and just to kind of play a little bit of that back to make sure I'm understanding it correctly, perhaps traditionally you had business systems which resided within one company that took in data and is perhaps using historical data within a company to forecast demand and then to be used in operations, whereas with the AI solutions it's a wider spread of data sources so that you can perhaps forecast using algorithmic models, using artificial intelligence, to identify the parts that are going to be needed, perhaps more accurately. Am I getting that correctly More?

Tilak Kasturi: 8:28

accurately more predictably.

Tilak Kasturi: 8:29

So, for example, I have the experience of the mileage, let's say I would know, based on the mileage of the vehicles in operation in a region. I would also generally have an idea whether it's coming from IHS market or other providers the VIO population, but also the age of the population in a particular region, and so, based on all those additional data points, I could actually predict the demand much elegantly compared to you know doing just using statistical models or based on the demand data that was presented to the supplier. You're kind of stretching it much further throughout the ecosystem and the more information you have, it makes you a lot more accurate and be able to manage your business decisions. Yeah, very helpful.

Mike Chung: 9:33

So I'll come back to the data sources in a little bit, but I'm kind of backing out a little bit. What kind of things is Predii working on? You've touched on some of those things already, but tell us a little bit about some of the solutions that you're providing across the automotive space.

Tilak Kasturi: 9:49

So we traditionally have been the AI layer for data around repairs. That AI layer allows you to extract insights from the data coming from different sources, whether it's a shop or a parts catalog, transactions or all of the above, and whenever I touch data, I have to kind of remind the audience that there's extreme care taken in making sure the PII data or the privacy is protected. There are information about protecting the, making sure the compliance items are taken into account. The copyrighted nature of the content is taken into account. This is a very well thought out process that needs to be applied and this is a guidance to anyone trying to use AI or some of your audience, building data science products. That is the core. You know it is really important. I'll touch upon this multiple times during this discussion, this discussion, but the things that, going back to your question, the things that we're talking, we're working on, is AI is probably the fastest changing technology and there are new models coming in, new technologies coming in. It's extremely. It's a fast-paced technology space and we've been doing this for 10 plus years and more recently, we are bringing in generative AI products into the market to enhance the experience of the workforce that is, in the automotive industry, of the workforce that is in the automotive industry. So what I mean by that is, a service advisor is now empowered with a service advisor assist, and that is something that we bring to the market, enabling the service advisor to say, yeah, I understand the customer complaint, and the AI is assisting the service advisor to connect the dots on these complex problems and then say, yeah, it's, a vehicle is running rough, and then there will be automatic suggestions provided by the AI and then be able to provide a very accurate estimate, and in a rapid manner. And these are newer vehicles, newer things, and based on these assist applications, you can also help a parts counter person. You know, if I'm asking for a plenum gasket, you know, we understand, but it's a plenum gasket, it's an air intake manifold gasket. Understand, but it's a plenum gasket, it's an air intake manifold gasket. The parts catalogs are never built for a street language or the cryptic way of saying an O2 sensor or A2F sensor these are the models that we built understands that automotive language. And there's one more reason why you don't want to use anything that's generic, right? So if I'm asking for a refrigerant oil for a Nissan Leaf, the model should understand that you're asking for a hybrid car, you have to give a non-conductive oil versus a conductive oil, right. So there are different viscosities. So these are technical things. That happens in each silos, right. A part supplier will understand, a technician will understand, but, as a service advisor, will not be able to just put it onto the repair order right away. And the models being able to understand the nuances is going to be very important. So we're bringing in generative AI products, which is the fastest growing segment in the industry because it brings in the state-of-the-art models.

Tilak Kasturi: 14:02

At the end of the day, the industry is an experience-based, human-centric industry. The more experience you have, the better you can take care of your car owner, and that experience is kind of embedded in each one of our customers', companies, companies, right. So this is not like somebody just gives you a resolution guide and you can just say I can fix the car because I know this resolution guide. And each car is different, each experience is different. So this is the best time to bring in AI layers into these products to make them A more trustworthy, b more efficient at understanding all of these complex vehicles and bring that experience back to a less experienced person who it's not about experience or the age of the person working at it, because they might not have seen all these scenarios of these different variations of the parts and vehicles and issues. So it tries to bring in all of that into a box, right.

Mike Chung: 15:19

It's really helpful and you touched on something very interesting in terms of sort of that residence expertise, that industry expertise where a service technician, a counterperson, will be able to say, oh, it's this make of car, I know that, as you said, with a Nissan it's this type. There are certain implications and ramifications from which to diagnose and get the right part, think about the context in which to solve that person's problem. And can you tell us a little bit about the data sources that you're gleaning in order to give those really specific solutions that are accurate and that I guess a seasoned, veteran, very experienced service technician will kind of know off the top of his or her head, because you know you don't necessarily want somebody going into an AI tool and you know having to type in a lot more questions than you or I could just kind of break the industry into users of insights and providers of insights.

Tilak Kasturi: 16:29

So users of insights are okay, the insights are available and I will tap into it and I will. You know, these are companies that are purpose built to provide that repair insights, to provide that repair insights, and there's lots of companies. Our target is to help these providers of these insights, right? So there's several industry leaders that provide this type of assistance to technicians or service advisors or anything that happens in the service lane. So the touching on data is the data sets originate. Either can originate from the cars, right, which is onboard diagnostic data in different shapes and forms, and that typically results in you know what are the diagnostic patterns when you have a check engine light, you know, is it a P0171, 174? There is a pattern of events happen and based on those patterns, you can make some decisions. So there's that is one source of data, and the second source of data is what happens within the shop. There are several software systems within the shop and they provide different outputs from that systems, and it could be a shop management system, or it could be other transactional systems that the shops use, or dealerships, in the case of OEM dealerships, and so every touchpoint starts to become a data and they're in different shapes and forms and you need AI to.

Tilak Kasturi: 18:18

There's no one format. You know it's not like oh, microsoft business intelligence format is this, so you take it. You know there are hundreds and hundreds of formats and so AI, again with the help of natural language processing, it can. It doesn't really necessarily depend on oh, you have to give me the format, you have to say the part. You know a disk breakpad has to be spelled out this way, but there are hundreds of suppliers and they could be calling it in any number of ways. Some use PCDB, some might not use the PCDB format, and so you have to be able to adjust to all those, as long as AI is really good at interpreting the intent of what you're saying. So if a human can look at three, four things and say, yeah, actually they are the same, ai can easily do that, and it can do that in microseconds, right in a large scale, and so the pattern recognition is at the core of what AI can do very efficiently. Yeah.

Mike Chung: 20:05

And you mentioned all the disparate data sources. I'm also thinking about the format, as you highlighted, that you have so many unstructured data that are coming in, and I think about if I'm at the shop and I describe my problem they may write it down, they may take it in and just being able to synthesize all that data. And I can't help but to think of this commercial from a few years ago where there are people going into the shop and saying my car is making a so yeah, so's my favorite show, yeah, so I'm glad you know it's commercial I'm talking about. So yeah, fascinating. And you highlighted earlier some of the predictive aspects that call on data from, say, industry providers like IHS, market, s&p, global. So I think that's another aspect of a customized solution that, say, predi would provide, versus something that's more shall we say, a general AI solution.

Tilak Kasturi: 21:09

Right, the predicting is on two dimensions, you know. Dimension one is the age of the vehicle, and it could be measured in the miles, on the mileage, you know. And the other prediction dimension is based on early indicators of problems, right? So if you have a diagnostic code pattern that you do a part of the diagnostics, you could actually predict what type of parts you need, not only now but also before it happens.

Tilak Kasturi: 21:59

The simple example is obviously the engine misfire. You know it's a P0300 series and engine misfire happens and you know you need to replace O2 sensor. Or, you know, in some cases, you know you need to replace more parts, you know. And so, as AI technologies are implemented, you could make decisions based on some of these early indicators and in the independent repair shops, they could suggest some other related repairs. And you know. The simple example is yeah, you have a timing belt replacement. This is common thing, right? And then while you're doing timing belt replacement, at 100,000 miles, you probably might as well change the water pump, right? So, as the labor costs for doing certain things are far exceed and you to become more important for the shop owner experience the car owners as well as the shop owners.

Mike Chung: 23:28

Gotcha, and when we were talking before we recorded this session, you were talking a little bit about the innovation cycle of AI solutions and how they're implemented into workflow. I think you highlighted how e-commerce as a new channel was a little bit different from an implementation standpoint. Can you highlight some of the things you were referring to there, Tilak?

Tilak Kasturi: 23:53

So the way the industry we measure innovation cycles in waves, right. The industry we measure innovation cycles in waves, right. So the last innovation cycle, the biggest one, was the internet and the internet innovation cycles lasted 25 years of that impact where, early on, when it was introduced in the Congress, with Al Gore, with the Information Superhighway Act and other things that opened up the Internet and then that led to the e-commerce wave, we never predicted that, oh, suddenly all the commerce would shift to, or significant commerce would shift to, e-commerce or it becomes an important channel. But it took 25 years. Now, with the AI, if you think about it, it's 2023, March around that time frame is when ChatGPT is the first version came in and it's even an alpha version or a beta version right, and it's even an alpha version or a beta version. But the 18 months that we saw with the amount of impact or the awareness let's just say awareness of AI in the talking about, whether it's in Washington DC or across the world, it's amazing.

Tilak Kasturi: 25:25

So what happened is the innovation cycle.

Tilak Kasturi: 25:26

It won't take 25 years for adoption across the board at a scale we're seeing close to in five years.

Tilak Kasturi: 25:33

You know it's going to have a shortest window ever known in the history of any innovation cycle that people are racing to put in an AI layer into their applications. So and we have to caution that, especially in an expert industry like automotive, you have to do that with extreme caution. It's very easy, or tempting, to throw in a chat GPT into your application, but it's not appropriate for a medical field or an automotive field. So the innovation cycles are rapidly changing and then you would see, you know, in the last 12 months alone, pretty ourselves we have experienced that we deployed generative AI applications in production for some of the mission critical applications and I was really amazed with which the industry is following. And so doing it in a secure, trustworthy manner and transparent manner will be the requirement. In the past, we've been doing AI for 10 years, 10 plus years, so we could say that with a lot more confidence that you know, use it cautiously, use it carefully, it will produce tremendous results.

Mike Chung: 27:12

I appreciate that perspective and you mentioned this shortening window for implementation in the innovation cycle. What are some of the drivers? What has made that possible?

Tilak Kasturi: 27:23

Oh, that's my favorite topic. One of the reasons is the availability of, I would say, the hardware, or affordability of the hardware, and I would also attribute that to the open source community and the cost of a provider. Like us, we depend heavily on open source and to build an OpenAI scale model. It costs billions of dollars Just on one day to run those computers. On a daily basis it's over a million plus dollars. No enterprise, no company, no single company in the Fortune 1000 companies will be throwing in and saying, oh yeah, take a million dollars a day and have my data centers run it. No, that's not how the enterprise universe works. Open source technologies available. It also makes the cloud in a secure manner. Whether it's Microsoft Azure or Google Cloud or AWS the three big providers they allow you to do it in a secure on-demand manner. So I don't need to have the compute to run an application all the time. I use it when I want and I bring it down. So a few hundred dollars to a few thousand dollars a day, I could have an equally capable application doing generative AI. That's the cost factor and allows the innovation cycles to be much more broad in applications, but also shrinks the time taken to bring it to the industry. So that's number one. Bring it to the industry, so that's number one.

Tilak Kasturi: 29:42

The second part is listen, we've been doing AI. We're doing AI in production for 10 plus years. But AI as a technology or on the research topic, it's been there for decades. But the awareness amongst the industry leaders are the you know, the boards of these companies. We don't have to, you know, educate anyone that's saying, yeah, you need to do AI as part of your roadmap, and that awareness allowed, you know, if it happens at the board level, and I think it makes it every part of the organization easier to implement, you know. So there are a few things that happen Awareness that it is going to, if applied, it can produce tremendous results and makes your offering more competitive. And makes your offering more competitive. That awareness, along with the affordability of being able to have an AI layer into your products. It's a really amazing time. Actually, this is the best time where we're having a lot of fun.

Mike Chung: 30:52

Yeah, you mentioned the decades long experience and I remember being in college in the early 90s and hearing about AI developments. Then I think it was fuzzy logic was a big thing in the early 90s, and I sort of remember even a rice cooker with fuzzy logic capabilities. So I guess it had something to do with. It's not a yes, it's not necessarily a no, but it's somewhere in between, and it will be able to use fuzzy logic to cook your rice perfectly. Does that sound at all familiar? Oh, yeah, yeah.

Tilak Kasturi: 31:23

You know the AI languages early on Ada and you know it's all in research, community and implementation Used to have specialized AI languages and you study all of that. And it's amazing, mike, even 2017, it appears very recently, it's like 2017-18 timeframe the hardware availability of the GPUs and all that it wasn't even close to what we have today, right? Wow. And it's so shrinking of the cycles, right, it's just moving so fast and the results proving it right. So no one is doing it for one of the things that we hear a lot from our customers.

Tilak Kasturi: 32:13

You know I don't want any R&D. You know this is not a magical experiment. No one wants, especially in an industry like automotive. You want to see results. This is, you know, no one is going to take a gamble that, oh, the car owner might get fixed or might not, or they'll get experience or might not get experience. No one will deploy an experiment, and so it has to be at accuracy levels. It has to have a reliability level at which the industry leaders will say, yeah, I'm going to put this in my application suite. If I don't do it, I'm not being competitive, right? So? And the quality aspects and the quality controls is in the automotive industry and healthcare. I can relate to both industries as I worked in both.

Mike Chung: 33:05

You mentioned radiation oncology. That's a big consequences and ramifications there.

Tilak Kasturi: 33:11

The safety concerns are extremely critical in these industries, and so you have to have a lot of confidence in what you're putting forward.

Mike Chung: 33:22

I appreciate that and thinking about. So we've talked a little bit about the past and the current. How about the next three to five years, five to 10 years? If do you, if you're looking in the, in your crystal ball, what kind of things are you seeing a little bit further down the road, either in AI or the types of products that you and your team are going to be providing?

Tilak Kasturi: 33:44

I would say this the AI today is more about providing, you know, decisions based on the data you provide you know it's basically it's not going to magically make an expert level decision for you. You have to have a historical experience, that based on which it will try to give you a reliable answer. Now that will continue to evolve and we see more and more personalized experiences on the horizon. So I would say, personalized experiences. You have to know more about why I'm coming, my car, my life, my experience, and be able to connect the dots on that. It is still possible to have that today, just the way we're, you know, marching forward with personalized medicine and in the healthcare. You know a personalized experience of taking care of the car owner will become a norm rather than an exception. You know, and the industry is going to you know, in a few years.

Tilak Kasturi: 35:12

As I said, the innovation cycle is shrinking and there will be a next generation of AI which is already in the progress. We call it human-centric AI or we call it the world models, an AI that is coming up in the next five years. Beyond the generative AI, there's already a next version of AI that's already being worked on right now and we see more and more regulations. Some are good, some are not so great, but especially in an industry that has high safety concerns and you would see more AI regulations there are some that already started, and if I'm talking to an AI, an application that is using AI, there has to be a trust from the user. That, hey, what gives some transparency and you know to understand how you are making those decisions. We call it reasoning, and that gives you the user more trust.

Tilak Kasturi: 36:32

So there will be more adoption of the reasoning-based models, and so I touched on these more on the technology side. But on an experience level, just the way we are seeing, whether it's Google Maps or Apple Maps, we automatically put that on wherever we go. It becomes such an easy application. You shouldn't even feeling that there is an AI behind it, and that's how the best experiences are. No one talks about it. They only talk about the experience you're having, right, right. It only talks about the experience you're having, right, right, and you'll see more elegant experiences in all the interactions.

Mike Chung: 37:21

Yeah, that's fascinating and really exciting to see what the future holds, and you know just a couple of last questions here as we wrap up Any challenges, any kind of mountains that you and your team have had to overcome that perhaps you didn't see coming before Throughout the 10, 11 years Predii has been operating, what comes to?

Tilak Kasturi: 37:48

mind in terms of significant hurdles. Well, the biggest hurdle is always data and data privacy or the availability of the data, and so this is an industry where every company has a proprietary data set and their licenses to their data set. So it takes time. It's not something. If it is easy, everybody will be turning on. You know the layers, so invest your time in you know collecting data, or do it in the right manner, right by taking into data concerns and protecting the consumer at all times. Data concerns and protecting the consumer at all times, and so that's so. The data side wasn't a surprise, but it will take time, and so that is really one area that we worked with one customer at a time.

Tilak Kasturi: 39:02

But the AI, pure AI companies providing automotive AI they do not exist before. Pretty, I confidently say that, right, but it takes time. And automotive AI, it takes time to build it. And there's this temptation that, oh, I'll turn on an off-the-shelf model and you know I'll be able to build applications the shelf model and you know I'll be able to build applications, and but I, you know that's one area where you know it took us a long time to get three patents, and you know, in a normal industry. It's. You know, getting a patent is fairly straightforward, but in an automotive AI or in the AI industry it takes time, and so we. The amount of time it took for the patterns also surprised us, and it's well worth it. You know, all our offering is kind of backed by the patterns and I'm of a person I would like to experience as an entrepreneur and it's experience the industry. So even if it takes time to experience, you know it's very fruitful journey for us.

Mike Chung: 40:29

And so really helpful. And two of the things I heard you talk I gleaned from that response are one, privacy so if you're getting data from somebody who's taking their car in, making sure my name and PII is not associated with it, and also, if I heard you correctly, copyrighted information content that you're bringing in from elsewhere as a data point in that model has to be used correctly and appropriately. Am I hearing that correctly? Tilak yeah.

Tilak Kasturi: 41:01

And they're not meant so copyrighted content. You know the legal teams will always tell you you cannot just put it into an AI model. It's so tempting, but you should not and you cannot, and that's the ownership of whoever provided that content. And there are special techniques to leverage that content so that it can be made available, but you cannot memorize that content. That's what I mean by use it. You're not able. You know we have to respect those laws and so, yes, it's very critical and, again, it's out of the box. For a technologist, it looks like everything is easy to use.

Mike Chung: 41:50

But taking care of these compliance aspects, and copyrighted aspects is where some of the, you know, critical methods of how you'd like to highlight to our audience regarding AI and how it is impacting the automotive space a lot with the AI adoption, that's for sure, and all the data scientists out there implementing these solutions.

Tilak Kasturi: 42:43

You know I urge you to kind of have the automotive intuitiveness first, then bring the technology, and you know so it looks like everything, looks like a solvable problem with the technologist mindset, because I played both roles, sure, and every model appears appropriate, every variable looks appropriate, but take time in bringing that automotive intuitiveness into it.

Tilak Kasturi: 43:16

These industries, with AI or without AI, are very human-centric and the AI will never replace a human and it will only make him or her very efficient. And so the one guidance there is yes, automotive intuitiveness is so critical and you know it's actually shifting more, and it's used to be. Oh yeah, the technology is cool, and then I'll bring in some automotive infusion, but more and more, as the technology starts to become more and more accessible, you would say, yeah, it's more, 75% automotive intuitiveness, 25% application, you know so whoever does it in the right manner, and so the need for automotive knowledge will continue to increase, not decrease. Yeah, that's what my guidance is. You know, if it is a nice way to bring in more of the automotive AI experience into automotive industry, to transform it. That's kind of my passion, not only for Predi but beyond Predi right.

Mike Chung: 44:32

Well, Tilak, thank you so much for taking the time to share all these insights with us in your, in your company's journey. And we'll close with this kind of a fun question do you eat ice cream? I love ice cream, so give me some of your favorite flavors, whether it's your go-to or I. Have a special celebration I want to have and I'm going to get something.

Tilak Kasturi: 44:53

Even you know more over the top oh, I like the one with caramel and nuts if there is a flavor available.

Mike Chung: 45:02

But pistachio and caramel flavor is my in a bowl or on a cone cone is waffle cone, regular cone, sugar cone with the chocolate on the cone, waffle waffle cone all right, well, I hope to share a waffle cone of caramel, ice cream and nuts with you next time. I see you to all of our listeners. Thank you so much for joining us. We hope you found this informative, interesting and enjoyable. Please like, subscribe and tell your friends about it, and until the next time, have a great day everybody. Thank you, mike. Auto Care ON AIR is proud to be a production of the Auto Care Association, dedicated to advancing the auto care industry and supporting professionals like you. To learn more about the association and its initiatives, visit AutoCare. org.