Taking the car-buying experience to the max with AI
Episode Details
Transcript
When somebody is shopping for a car, they need real-world reviews and practical information about each make and model. But for retailers with tens of thousands of cars for sale at any given time, offering the right content for shoppers to make an informed decision is a continual challenge.
Today, CarMax is tackling this fundamental business problem by using AI to empower its content creation teams. This episode of Pivotal features CarMax's Chief Information and Technology Officer, Shamim Mohammad, who discusses his company’s interesting approach to building new AI solutions on a foundation of solid data governance and culture of experimentation, which has yield extraordinary results for customer and employee experiences.
According to Shamim, by starting small, allowing teams the space to learn and get better, and then empowering them to put innovations into production, they can create incredible solutions and scale them quickly. And the beauty is every organization can benefit from this approach.
00:02 Introduction and Welcome
00:19 The Role of AI in Business Transformation
01:19 Introducing Guest: Shamim Mohammad from CarMax
01:53 Shamim's Journey and Role at CarMax
03:25 The Advent of AI and its Impact
04:42 AI Implementation at CarMax
07:51 The Importance of AI Governance
10:09 Creating a Culture of Experimentation
12:19 Scaling AI Projects and Governance
13:15 The Role of Data in AI Implementation
27:08 Lessons Learned from AI Implementation
29:17 The Future of AI
30:53 Conclusion and Farewell
Link to full episode transcript.
Produced by Larj Media.
Hayete
Welcome to Pivotal. I'm Hayete Gallot, corporate vice president for commercial solution areas at Microsoft. I work customers around the globe to transform their business through technology. At the center of every transformation are people who give technology its purpose. And that doesn't change with the advent of AI. It's actually being accelerated. People spark visionary ideas for leveraging technology. The release of AI technology, like chat GPT, this year is exciting, but it has led to big question as we grapple with the best way to harness those tools enhance and support the people behind the work. We like to talk about technology. I love to talk about it, but we often forget that technology is most effective when it supports people with purpose. This season will demystify AI by talking to the innovators using new AI technology to uplift their industries and augment their people from education, to journalism, to surfing. And it just illustrates what AI is about. Everybody thinks it's about tech. No. Everybody's using AI. That's what we're gonna show you on this season. Today in this series, We have one of our customers joining us, Carmax. They've had a very extensive experience with data and AI. So I thought it would be great to talk to Shamim Mohammed, who is the executive vice president and chief information and technology officer at CarMax. He's driven a fantastic transformation in this company, and I think we will all benefit from his lessons and learnings, as we have a conversation today. Shamim, welcome. Shamim
Hello, Hayete. Nice to meet you. Hayete
So, Shamim, you've been at CarMax for eleven years. And before that, you are at BJ's Wholesale. You're not a stranger to digital transformation and using technology to drive business outcome. So I'd love to hear more about your journey. How did you end up at CarMax? Shamim
Yeah. Absolutely. Hi. So I've been in the technology innovation, space for almost thirty, thirty five years. And I always found myself at the intersection of technology and business. It goes back all the way to even my college days. You know, my education was computer science and accounting. And, over the years, I always was fascinated by what technology can do. More importantly, what business problem, technology can solve. So that's what I've been doing. I worked for both large fortune five hundred companies, and also couple of startups. And before coming to CarMax, I have my most of my, experience have been in retail, driving customer experience, driving, you know, a lot of data analytics kinda experience, And at Carmex today, I run all of technology. So I role I play the role of the CIO, which is running the technology space, and I'm also play the role of the CTO, which is driving product innovation, customer experience, and the business transformation. Hayete
So how did your interest in AI start? Shamim
The AI interest has been here for a long time. I remember going back to even the college days, you know, thirty some years ago, we were discussing what artificial intelligence could look like. What's really fascinating is after, you know, thirty years, it is like becoming reality. So so it it just has become this critical mass and the technology has evolved to a point where we're actually seeing what we had envisioned or thought about or dreamed of thirty some years ago. So I've been fascinated in this space for a long time, but my passion for AI really is built on data. So if you look at some of my career, even before CarMax or even before that, I spent a lot of time with data and the power of data, power of analytics, power of helping companies drive business transformation, driving great customer experience, driving great associate experience using data. So so that's what I've been focusing on, but now we have this technology available just like open AI at JPD. There's just so much happening in the last couple of years. I think we're just seeing a tremendous evolution and benefit of AI. So I've been fascinated by AI pretty much most of my life. Hayete
So I know you've been on this AI journey for a while, and you're actually already In action at Kermax, can you tell us more about what you're doing at Kermax and how you're leveraging AI? Shamim
Yeah. Hyatt at Carmax, you know, one thing we have, and we believe one of our biggest asset besides having a great culture, great people, is our data. So we always had great data. So what we've done is over the past few years, we put a strong data governance model and team so that we can clean up our data, organize our data and and and have right technology to leverage this data. Mhmm. So we have built the talent base around data science and machine learning. And by, you know, those the team really have been focusing on how do we optimize and leverage this data to create great experience, continue to drive customer you know, a competitive differentiation, and also make our lives of our associates better. So if I look at the uses of AI and machine learning and data within our environment. There's, like, three, examples I can share with you. One is the digital merchandising. So a customer looking for a buy buying a car or shopping for a car, we're using machine learning algorithms to help that process go better and provide a great quality actual picture of the car that he or she might be buying. So that's a space we have tremendous progress. We also have some patented technology that our data scientists and machine learning experts built to optimize that process. So that was a huge advantage there for for CarMax. The other area is customer shopping. You know, we build a recommendation engine where we can understand what the customers are trying to do and then help them with the shopping process. At any given moment, on our website, we could have tens of thousands of cars. And if somebody's looking for a car, that buying process could be overwhelming. So we're using AI to optimize that experience and make that process much easier for our customers. And the last example is really around what what we're using GNI, the power of GNI partnering with, you know, OpenAI and JetGPD. To organize massive, massive amount of data in a way that's easier for our customers to evaluate and look at So he's doing two things for us. One is helping our customers look at the card information in much more friendly and easy to digestible way. The other thing it's doing is helping us from a search engine ranking. So because our data is organized in a great way and it's published in the right way, it's helping when look customer looking for and searching for cars in the internet, we get listed in a preference preferential way. So that's been great, you know, benefit for the company. So those are three kind of use cases I shared with you, but our teams are using AI and machine learning across the company for various things just to make sure we continue to drive innovation and continue to remain, and and and as stay as a market leader in our space. Hayete
So when you you think about GenAI, a lot of the customers right now are asking, how do I get started? How do I prioritize scenarios? So how did you go about it with your team? Shamim
It it starts with having a good culture. And what I mean by culture is really create an environment where the employees or associates are empowered to experiment and they're giving the latitude to try different things and learn from it. So that's where it starts. And then I think it's supported then by a strong data foundation because if a company is not good with data, they cannot be good with AI. So those two things I think teams need to really, company need to pay attention to. Fortunately, for Carvex, we've been focusing on those two aspects. And and then supporting both of those is really having a strong governance model for AI governance What we did, Hyatt, is at Carmex, we had, a good data governance that we put together a few years ago and we can evolve that into more of a AI governance. So between the AI governance and the data foundation and in part culture, our teams are experimenting different ways to continue to drive business outcome and drive great customer experience. And and these teams are cross functional. So that's the beauty because it's not just technology. It's not just marketing. It's not just operation. The teams are organized cross functionally around a common mission or common business outcome. So when they have they're clear on their outcome, they are able to use different technologies to figure out how to go about achieving that outcome. So in this particular use case that we talked about open AI, at JPD, the team had a mission to provide better information to our customers on the website, when they're shopping for the car. They also had a mission to improve the search engine ranking. So they found that JetGPD is a great way to take this massive amount of data that we have and solve the best card data we have and use the chat activity to organize and, structure the data in a way that makes it easier for our customers. Hayete
So two things struck me, experimentation and governance. I wanna go I won't start back to experimentation. So how do you create this environment for experimentation? How do you give technology to to your teams to go about it and tinker with the technology and look at building MVPs for CarMax. How do you enable that at scale through the company. Shamim
Yeah. It started out kinda small. Right? So several years ago, when I first became the CIO of CarMax. I was fortunate enough to partner with a couple of my colleagues the COO and CMO. So three of us are gonna join at the hip, and we said, okay. We gotta transform our business. From being a market leading brick and mortar company to more of a digital company. And we deal for the transformation had to get our teams organized in a different way, and they had to really work towards a common mission, common goal. And they had to really move really fast using, you know, the lab adjusting different type of technologies. So we started small with handful of teams. And today, majority of the work we do at CarMax, are done through those teams. And those team number have grown significantly. We have upwards of fifty, sixty teams now working cross functionally on all these different types of experimentation and driving innovation for the company. So that's really where it started, but we didn't have to go all in day one. We started small, shown the results, and continue to build on that over the last few years. And now the beauty is we know this model works, and we are able to operate that way. Although, we are a big large company. And I think what really helps me personally is that I have worked, in a startup environment many years ago. So by this this the way we work today as teams, the the startup culture is really what we have. So these teams are somewhat empowered. They're small. They're cost functional, but they're mission driven. So what's allows them to do is really try. They they take shared accountability to drive results. And I think that's the beauty of this. So I think every company can do this. It's not an easy transition, but if you start a small allow the team to really learn how to do this and get better at it, then you can skip from there. Hayete
And so as people are experimenting, you must have a lot of project coming through. So I I wanna come back to the governance. How do you govern? How do you prioritize? How do you say we're gonna go for this one or not this one? Because I think everybody right now is trying to understand how they go about this. Shamim
Yeah. I think it starts with being very clear in terms of overall mission the organization, right, the highest level, what are your trends to achieve? What are the goals you're going after? And the way we organize this is our, we call it product teams. Right? These cross functional teams are product teams. And they have, obviously, annual objectives. But they really look at the quarterly OKRs objectives and key results. Mhmm. And then they break it down to, like, biweekly sprints. So we get to see what these teams are working on, what experimentation they have ran, what they learned from this experimentation, and how they're gonna be adapting the next level, experimentation. So if you look at this, these teams, they present in product showcase every couple of weeks. So from a governance perspective, the senior levels of the organization, we're able to actually go and what these teams present in a very short, like, ten minutes presentation, of what they've been working on. So which really helps us as a helps me as a leader of the organization of the company to worry about what teams are focusing on also allows me to adjust anything we need to adjust it. But because these teams are so empowered, they have a lot of latitude, and they know the guardrails, right, what they can do, it cannot do. This is where I think the beauty is he only empowered a cross functional team with a great mission. They will surprise you with amazing results, and we're seeing that every day. Hayete
So through this governance and how's they come, if you see very interesting projects, and now you wanna move to large scale production, that's where you make those decisions. Is that how it works? Shamim
I think the team's really because they'll experiment and try different things. Yeah. And it's naturally the scale when the a good idea comes And we believe we see the results and it's driving the desired outcome. They can actually scale it almost, like, on their own without getting a lot of bureaucracy around, you know, on a lot of hierarchy. They're very empowered to push those things, you know, globally. Like, in this example, in the open AI, the chatty, the customer reviews, and they structured data. I mean, the team did a lot of experiments. They showed results. They learned from it. They optimized it. And then they went, you know, basically entire company. And that was just natural progression. And that's how the teams work today at Carvex, which is really amazing. I feel like I have this massively large company, but it's still working like a startup. And that's the beauty. Hayete
That is actually, a very interesting approach you guys. Have adopted. It's very conducive to, creativity and innovation. It's pretty exciting. So if you think about this open AI scenario, you mean, What has been the business outcome? Like, what results have you seen in terms of your customer experience? At a high level. I'm not trying to get into your business details, but more on the what were you looking for in terms of experience you were trying to create, and what did you observe? Shamim
That's a that's a great question, Aya. So we have, let's say, fifty, sixty thousand cars on our website at any given moment. And when somebody's looking for a car, they're looking for a lot of good information. Okay. The what is the customer review? Is this a good car for we can, you know, adventure or is it good for my family? Is it good for my dog? So they're looking for those kind of information in a human way. So we have those data. We always have those data. We have, you know, we have all the car features and functions and all these things for every model to make. We have information from a customer. They're telling us what the car is, how they like it or don't like it. So we had all this data. We just didn't know how to present the right way. So that is usable and much more friendly for our customers. We have content writers. They're going through all the customer reviews. They're looking for all looking through all the data. And then creating some articles and, messaging for the customers so they can navigate through those. Problem is with tens of thousands of cars, all this make and model, and all the thousands of customer reviews. It's very hard to look at every situation, every scenario, and make that available on the website. So this is where Chad GPT was incredible. So we took our data our actual customer review, we took our cost card data that we have in our various systems, And we used through prompt engineering, we were able to tell chat gpt. Hey, hey, this is what we were looking for. This is the information I'm giving you. Return it back to me based on these parameters through prompt engineering. So once you got the data back, for pretty much most of the cars that we're able to publish it back on our website, and there's a human involved. It's not just we're getting data back. We're just publishing it. We actually have a human in the loop kind of mindset because, you know, we we'll probably talk a little bit about responsible AI because we wanna make sure the information that we're sharing meets our internal standard and also accurate. So we have a human involved So those content writers now reviewing this content before it gets published. So they don't have to create the content, they get the content created, then they review, and we publish it. And it has a two big benefit. One, the content would have taken us for, like, ten, twelve years to develop by many, many content writers. Now we're able to do this in days, maybe hours sometime. So that's a huge benefit. The second thing is allowing our customers to find the information about this large purchase. You know, these cards could be twenty, thirty, forty thousand dollar purchase. Now they they have this meaningful, usable, human centric information available that they can use to make an informed decision. So that's the second thing. And the third thing is I think just from a search engine ranking. Now we have better data about cars compared to most other websites. So when the search engines looking for good data on a particular subject matter. They're looking at Carmex saying, okay. Karmaex actually has better information. So, hey, CarMax should be ranked a little bit higher. So tremendous benefit for the company, the customers, and and in general, it's been a huge win for our team. Hayete
So what does it mean for your content writers? It must be super satisfying for them to not not have to try to catch up with all this information and being able in a position to review content versus having to try to bring all this together. And how does it change their day today? Because they must have more time now. Shamim
Yeah. They well, they actually, have same time, but what they're doing is they're using the time on more productive and more value added task. Right? So because before they were writing all this, now actually they're reviewing it, and they can get to all this cars that they couldn't go before. So it's really the tremendous engagement booster because they feel like they're adding a lot more value in the process and they're able to get to the work that they couldn't get to before. And still, the quality is still on par, if not better than before. So it's a win win for our customers and our for our associates. Hayete
That's awesome to hear. That's what we wanna hear, right, with GenAI. That it's benefiting the humans as well. So when you rolled out this GNI solution, did you face any skepticism or pushback from any group in the company? Shamim
I would not say pushback, but it was like the unknown. Right? So, again, what happens in in our company right now? Because we have all these teams. They are working within a certain guy or guardrails and guidelines and we have empowered them. So they're experimenting a lot of a lot of cool things, cool technologies such as Open AI. So so we're very happy that the the team really figured out how to use this. And it happened actually before OpenAI became this big popular thing where my team was already doing this. And we're like, hey, this is how we just do stuff anyway. So it didn't we didn't think it's a big deal, but now it's like, oh, you do the big deal. It's a little bit ahead of time compared to the rest of the world. But I think, you know, the biggest thing was just education, you know, because this is new, relatively new to broader set of, associates. So I spent a lot of time educating people. Like, hey, gen AI has a huge potential. And so we want to embrace it. But do it in a way that is responsible. That is, you know, that keeps us make sure within the regulatory guidelines and everything else, So we're aware of what some of the downfalls might be. And be by being aware of those, we're able to be using it in a productive safe way. That's the so we spend a lot of time in education. I think that's the biggest thing. And and, you know, and, and, and, and, also, it will evolve the AI governance model over the last year or so because we feel like, again, this is moving so fast. We really need to make sure a cross functional leadership team is starting with me and our general counsel and the executive team all the way down to the, organization that we have a cross functional alignment around how to approach this, what are we learning, how do we evolve this? So it's been a lot of excitement about GNII within the company, but it was it started out with, hey. Well, there's something happening there in a couple of product teams. That's what her company wide, visibility, and people are excited about it. Hayete
So, I'm with you on, there is excitement on the possibilities, and there is anxiety, and the need to be responsible. And you said you evolve your governor model on AI. Can you share more on what it looks like and what you've made as in terms of changes to ensure you were responsible? Shamim
Yeah. I think the big thing is we wanted to make sure that the way we leverage AI and and how we use any technology, is very much consistent with our core company values and culture. Which is transparency and integrity. So that was the thing. Right? And I also wanna make sure that we're following all the regulatory guidelines, we're following all these standards and make sure that we stick current to the regulations that may be coming because there are a lot of as you know, there's a lot of changes happening right now, different government agencies and things like that. So they're so we wanna make sure we sit through to whatever the guidelines are. But more importantly, you know, the one of the big thing we have is a people centric culture. Right? So whatever we do in some way, shape, when I'm, you know, help our associates, help our customers, and the communities that we serve. So I wanna make sure that the AI stays true to those our core value values and principles. So that's one thing we we put in place. The other thing we did is we we, also through education, we're helping our associates, not just technology, not just product but across the board, how to understand what they can and should do or should not do. We didn't lock it down. A lot of companies I I've seen that they shut down, like, hey, cannot go access it. Right? We actually allow our associates to go experiment and try, but we also at the same time to tell them they should be doing and not should be doing. We also created some clear line of accountability in terms of, you know, AI, central excellence. And what the team members there at different levels, not just at the executive level or the engineering level. So having that level of company wide energy and focus and cross functional nature of this. I think it's really what has happened and how we have evolved over the course of the last year. Hayete
That's actually very interesting this AI center of excellence. Can you share more about this and different levels being in the center of excellence? How does that work? The center of excellence. Shamim
Yeah. So at the highest level, it's part of the AI governance, and the AI governance is where I'm involved and the c level executives. And the next level, we have some of the VPs involved, including our CSO. We we have our, you know, chief digital data officer there. We have our head of, you know, data engineering is there. So we have that group next level below us. And then next level, we have actually one of the VPs as a responsible single point of contact for all things with AI, because he's also, like, you know, sort of the program owner. And then below that, we have cross functional team from different business unit. So legal, operations, finance, marketing. So it is I believe this has such a profound and cross functional impact and possibility, we need to have a we need to approach it that way. And that's really how it works. Hayete
That's exciting. I learned something. It's it's actually quite similar to some of the approach we have at Microsoft with our cross functional approach and how we've abandoned that in our entire development cycles from the beginning on the idea to developing a product. Very cool to see. Now I wanna come back to something you said earlier around One of the key thing that really played well for you guys is how you got your data already. Your data in order very early What does that look like to get your data in order? Cause we all say, yeah, it's as good as your data. So I'd love to to hear more from you. Shamim
Yeah. The data is the foundation that fits into AI models. Right? So if we don't have good data, you're not gonna have good AI. So that's just a very fundamental. For us, you know, we knew we had tremendous data assets from the beginning. We have absolutely the best use card data probably on the planet. So we're gonna make sure that those data are organized. We have we trust the quality. We trust the make it available. It's scalable. Is updated. It's refreshed. So one of the things we did early on probably four, five years ago, maybe longer than that. We put that this data governance team in place. So we have the daily storage shift. Again, we we take the way we do things at CarMax has, you know, usually mostly cross functional. So even the daily governance was a very much of a cross functional effort. And by doing that, we really, you know, organize our data for customers, our, you know, car cars, And any any relevant data subject matter, we organized them, and then we made it available through some of the also Microsoft technologies in the cloud. And and and that gave us the scale. Also, it allowed us to monitor the quality of the data. And again, it's tying back into our standards, you know, privacy and, you know, security aspect of it. So all those things we took into account, to make sure the data is usable and available and organized. So that's really what happened. And so by doing that, we're very well positioned for taking advantage of what's happening in AI. Hayete
You've had a lot of experience with data and AI. What are the lessons learned may that you've had, and you would say, you know, looking back, here are the things that we do differently or some lessons on you had that you wanna share with this. Shamim
Well, I think worked well is where we have a cross functional alignment. You know, this is where a CIO or CTO or CMO or CEO, nobody can do this alone. Right? So it makes sure there is a strong alignment and an engagement from across functional team. So that's the number one thing. And and that's something, you know, we did very well early on. The other thing I would say is the this whole school experimentation, empower your team to experiment. Give them guardrails, give them the tools, give them the access. Don't lock it down because if you lock it down, you're not gonna get the benefit. So it's really create that culture, empower your team, make them feel safe to try different things as long as they stay within the boundaries or guidelines you put out there. And the third thing I would say is a talent, and that's something I think we had to evolve. This is changing so fast. You know, data scientists the AI, you know, engineers. Those are and that's a new skill set that's evolving and a lot of demand. So you know, two things you can do. One, obviously, you can go higher. Some of them externally, but the other option is just grow, find your best, data scientists, data engineers, train them, invest in them so they can become good AI engineer. Because it prompt engineering and how to leverage gene AI stay current. I mean, those that's a little bit different than traditional AI, traditional vegetation modeling. So that's an area I think with learn. And we are actually done taking some steps to build that talent internally. Those are three areas I would, you know, I would look forward and by a lessons learned. Hayete
That's that's actually a fantastic summary. I love the concept of thinking about the talent and the evolution of talent and reskilling folks so that they can take advantage of AI. Right? Because generative AI is a bit different than what we thought about machine learning. So now when you look ahead, Shamin, You've been in AI for a long time. Right? What gets you excited as you look ahead? What's coming? Then you're like, yesterday, there's some big announcement as an example. What gets you excited? Shamim
Ala Hyatt, I think I'm, like, pumped in terms of the possibilities of AI. It's supposed like a dream is coming true. This is something I had dreamed of thirty years ago. And now I'm like, okay. It's happening. I think what's I'm I'm most excited is really that it is going to unlock the human potential in a way that we have never seen before. We are gonna be able to do things that we didn't think was gonna be even possible. Even simplest thing, the example we have for OpenAI that I share with you, that's a simple, easy example, but that's just a tip of the iceberg. Every role in every company at every job in some way or shape will be impacted by AI GNI somehow. So I think it is gonna make us smarter because it's not like AI is gonna replace us, but it's gonna make us smarter. Is gonna free up our time by doing things that we did not do, don't need to do, and the time that's freed up, which is the most valuable commodity we'd have is the incredibly valuable. And with the time, I'm just excited what else can we do as a human being? And how much innovation can it drive because of that extra time and focus that we're gonna have? So I'm just overly excited for all the things that's coming. And the pro the good thing is I don't have to wait another ten years for this. I think I'm gonna see changes literally every year or if not every month. Hayete
Yeah. Well, shameem, I I wanna thank you. I have to say this conversation, not only has been inspiring. But I think everybody's gonna listen. It's gonna learn so much from your approach, from the learnings you had. Cause I think this is full of possibility. It's exciting, but I think everybody is thinking about how do I approach this responsibly, but also at scale and get results for my company. And I think you gave us fantastic examples today of how to go about this. So thank you so much for your time. Shamim
Thank you, Hyatt. It's been a pleasure talking to you. And, let's, you know, both of us, and all of us look forward to what's coming ahead. It's just it's an exciting future for all of us. Hayete
Yeah. I agree. Well, Hart Italy. Thank you, Shamim. Shamim
Thank you. Hayete
Thank you for listening to pivotal. I'd love to hear your story in your pivotal moments, so don't hesitate to follow me and share on LinkedIn. Audence information is also available in the show notes. Our show is produced by large media, That's L ARJ Media. Special thanks to Lynn Yang and our partners at We Communications. 
