Ep 40: Supercharging Enterprise Operations with AI with Lexmark Global CTO & CIO Vishal Gupta

‍Evan: Hi there, and welcome to Enterprise Software Innovators, a show where top tech executives share how they innovate at scale. In each episode, enterprise CIOs share how they've applied exciting new technologies, and what they've learned along the way. I'm Evan Reiser, the CEO and founder of Abnormal Security.

‍Saam: I'm Saam Motamedi, a general partner at Greylock Partners.

 Evan: Today on the show, we’re bringing you a conversation with Vishal Gupta, Chief Information Officer and Chief Technology Officer at Lexmark

With over 9,000 employees operating in 170 countries, Lexmark is a global technology company specializing in printing solutions and cloud-enabled IoT imaging technology.

In this conversation, Vishal shares insights into how Lexmark is using AI for predictive maintenance and an optimized customer experience, the revolutionary potential of enterprise AI adoption, and how IT leaders navigate the opportunities and challenges of emerging technologies.

Well Vishal, first of all, thank you so much for taking time to join us today. Before we kind of dive in, uh, do you mind sharing with our audience a little bit about kind of your background and the role you play at Lexmark today? 

Vishal: Sure. So my background has been about 30 years in technology. I actually started my career in management consulting. Way back and then, you know, kind of went into different functions, but I've always been in technology, but in different functions. And then, last three of my gigs, you know, I was fortunate to be a CIO or CTO, at a sort of little bit larger companies and my responsibilities at Lexmark cover not just the CIO bit where we think about, you know, how do we create great experiences for our 8, 000 employees, but also as a CTO, I'm responsible for building the software that enables our customers that we sell to our customers.

Obviously that includes cybersecurity as well. I've also had the responsibility in the past for what we call is Lexmark Ventures, which is, you know, around incubating completely new things. And so it's a fun role. I thoroughly enjoyed it. 

Evan: I'm sure almost everyone in our audience is somewhat familiar with the Lexmark brand but I imagine that most people don't really fully understand kind of the full scope of, you know, what you guys do. Do you mind kind of sharing a little more about kind of what the, what the company does? 

Vishal: So, you know, we are, our focus is to be both a leader in imaging, right? Our vision is to enable our customers leave a lasting impression in their world. And so we work with probably, you know, a thousand of the sort of fortune 2000 or fortune 3000 companies in, in really enabling them to do so. We do that not only through the world of printers and imaging, but we're also fundamentally a technology company at heart. And so we do a lot of different kinds of software, for example, we'll do software that will do all kinds of document processing, not just scanning, but image translation.

So you could put a document in a printer and it could print out in any of the hundred languages you want, right, you know, without even going to a computer. So imagine the implication of that if you're in a school or you're in a bank or you're trying to provide a service. 

We've also transitioned a lot based upon the technologies that we have developed into being an IOT and AI company. And so we launched an offering called Optra about two, two and a half years back. And there, we're really doing a lot of work with AI to say, okay, you know, the same thing that we did with the printers, our printers have these IOT sensors, about a hundred IOT sensors for printer. And the last, 50 percent longer than the rest of the industry because we're able to stream the data and we're able to, you know, use the AI to extend their life. Can we do that with other manufacturers for other manufacturers too? Or can we also apply edge AI, for example, to get a bank or a retail store or a transportation company have a much better experience. 

And so the way to think about Lexmark is as a technology company. Obviously, we are very significant in the world of imaging, but also trying to be equally significant in the world of IOT and AI as well.

Saam: Vishal, It's hard to have a conversation here in March 2024 about technology and not, you know, talk about AI. So there's a couple of things we want to double click on around AI generally and also the role of AI at Lexmark, but maybe let's start very high level. Like you've been through a lot of technology cycles. Where are we in the hype cycle around AI and how are you approaching it at Lexmarks? 

Vishal: So I think the best way to answer your question is to maybe reflect back. What does it take for technology to become mainstream? Right? Let's think about cars. How long did it take to get 100 million cars sold? You know, it took 84 years.

So the pace used to be, you know, you have a piece of technology, it's called a car. It'll take 84 years to get 100 million. You know, even, you know, we all think so highly of cell phones. We all have these devices. Even cell phones took 14 years to reach 100 million adoption target. 

If you think of ChatGPT, the hundred million took two months. And the reason I, I share that is we do live a little bit in unprecedented times in terms of how fast the adoption from kind of the tick box perspective, some can happen. Having said that, right, is ChatGPT as transformational right now in people's life as a car or a cell phone was clearly not right we're at the start of the inning.

And so I think of this as, you know, We have a technology that can get rapid adoption that can create a lot of interest, but clearly like anything else, it will take time for it to mainstream from, you know, an interesting thing that top 10 technologies companies use to something that, you know, every company can start leveraging across every process, every product that is out there.

So, so we are probably in the front end of that hype cycle, if you may, right? Like what Gartner likes to call it, the, the, the delusional, the value of the delusion or something to that effect. Having said that, I think there's going to be lots of interesting use cases that will come out of it as we continue to do the work, you know, to make it much more mainstream in a meaningful way, not just in a hype way, but in a meaningful way where just like now we don't think about web, right? It's just part of what everybody does. 

That's what I think will be the right analogy from an AI that, you know, if you're a company, you would just be doing AI, just like you do web today. And you don't think about the fact that you have a website or you have a domain name or, Or you can transact, you know, all your applications are running on the web or the same thing with the cloud. So I think it will become as omnipresent and as mainstream as that, but we're at the front end of it.

Now at Lexmark, we've invested a lot in making us to be a data driven company over the last several years, we, you know, invested a lot in creating things like data lakehouse technology. So that way, you know, from 100 different systems, all the data can be gathered, deduplicated, refined, tagged, all of that, right? And that has fed a lot of our AI. And we believe that will give us a unique advantage in the world of generative AI as well. 

I'll give you one simple example, right? When this technology came out in March, we thought, hey, the best place to adopt this technology is where we have, you know, good data, which is telling us what good looks like. And we thought, okay, you know, one of the places where we need to answer a lot of questions is, guess what? RFPs. We get all these RFPs from customers, and we also have filled lots of them, and we know which ones we want. So what if we train generative AI using the ones we want? And that way we could generate good answers to the new RFPs that come, whether those RFPs are for our products and imaging, or for IOT, or anything else.

And, and that has been a great use case because we have 600 good examples of RFPs that are won, and so we can use them to train it. And so the data investment that we did. It has been fantastic to now take this new technology in generative AI, you know, build on top of open, open AI on GPT, for core and be able to do that.

We wouldn't have been able to do it had we, did we not have those RFP responses completed? Did we not have the product manuals that could understand, you know, what people are asking for and, and, and know what good looks like. And, and also with, you know, if we couldn't have done it, if we didn't have people who are open to change and willing to experiment. And so hopefully that gives you a good sense, in terms of where we are both in the hype cycle and as well as what opportunities are in front of us to drive that differentiation. 

Evan: You kind of talk a little bit about kind of, you know, AI, your, your, your thoughts at the high level. Do you mind kind of going a little, you know, just maybe there's other kind of anecdotes you can share at the lower level. Um, I guess specifically, are there, are there kind of ways in which you guys are using AI that just, you know, people from the outside might find surprising?

Vishal: When I joined Lexmark, we had about five data scientists, right? And so, the thing that struck me as we were embarking on our AI journey was the fact we have to do three things. One, we'll have to scale our talent on AI, right? And so we said, okay, data scientists cost a lot of money, a quarter million dollars, and they typically like to jump companies. So instead of hiring a hundred of them, can we train our existing talent to become data scientists? And so we tied up with North Carolina State and, you know, We got into this program, which is a daily program. It runs from evening from 5 to 9 p. m. And, you know, we assign mentors.

We started with one cohort. Right now, we're on the sixth cohort and we have almost trained 100 people who've gone through year long daily program and they got four certificates at the end of it. And what's amazing is none of those people ended up leaving us. None of those people dropped out of the program. They got a great set, you know, set of certificates out of it. And many of them are not even in the technology team. So we tried to create a center of influence, kind of a concept, because we wanted to embed these people across the business in finance, in product delivery, in supply chain, in commercial, right? Because that way they could bring the most interesting problems to us, right? They could kind of on that cultural side, really understand what data can do for you. 

The second thing we did was to really invest in the data architecture that I was referencing, and that has been very helpful in terms of, you know, creating that data lake house and creating the ability to, you know, have things like master data management, so you can deduplicate and call the data in a unique way, be able to do the join, you know, be able to build a full idea of what a customer is across all the hundred different systems and, and what makes them do things and not do them. Same thing for the employee. Same thing with all the different stakeholders. You're there.

So what that has helped us do is to ultimately, by putting the right culture, by putting the right talent together, and the technology stack together, we have now gone and we have now probably have about Maybe over a hundred machine learning models in production today. So that's like a, you know, a 10x leap of where we used to be. And I'll give you some simple examples of what these models can do from the world of imaging. So you can understand. 

A typical printer lasts about four years, right? For most companies. We said the imaging business is such that really the, the way you're making money is by all the printing that happens. So the longer the machine will be there, the better it is for the customer and the better it is for us. And so what if we can make these things last instead of four years, say seven years, right? And so we used all the streaming data that are coming from these devices. Think about millions of devices, hundreds of sensors on each of the device.

So think about billions of data points coming in. And we're using to analyze this data in a streaming data context. and marrying that with all the customer care systems to understand why machines go down. And so we're able to extend the life of these so we can systematically see what causes things to go down. How can we design them better? How can we use better suppliers, better processes? And now our printers last seven years. 

And so this has created an amazing advantage for us, and our customers because it's better for the environment. Our customers have to upgrade and buy new printers less. And for us, we can sell things as a service. And so this is a great example of AI creating a win win, right? For the company, for the customer, and for the environment. 

Same thing we did with predictive maintenance, right? So that way, Our uptime is very high, and that has taken our retention rates from like low 90s to 99%. Because guess what? People don't want to have to replace the machine, they don't want them to go down. And they want that, whenever the toner is done, automatically to get filled, which also is being driven by machine learning supplies, depletion algorithms, so we're not wasting the toner and so on. 

And so we have found, we created a lot of these use cases, leveraging AI that extended the life for the deprecative maintenance. That automatically, the things like shipping. The machines became customers, right? So the machine would automatically say, okay, I need toner. And without even a human intervention, the toner would get shipped to them so that they would get fulfilled at the right time. 

And we said, ultimately, if we could do all of these machine learning models for ourselves, what if we could abstract the type of device and do it also for others, right? That's why we launched this whole business called Optra. We are right now doing it. For example, dental machines. You know, when you go to a dentist, many times these machines are down for the dentist, it's a loss of 10, 000 a day. We have now worked with one of the largest manufacturers in that space to make sure that they'll now go down instead of, for this particular customer, it was going down almost 12 days a year, it's gonna come down to one day a year. And so that's, that's got a massive impact on them in terms of their retention rates, in terms of their ability to please their customers. 

And we're also now getting much more into the Edge AI, Evan, and I'll give you a simple example. Right? We go for oil changes. We're working with one of the biggest oil changes in the planet, company in the planet, where we can, with our Edge AI technology, detect if a car that has come in for oil change, has it been greeted?

It turns out if you don't greet the car, the people run away, because they don't know how long they'll be waiting. And so the AI is able to just see, okay, has the car been greeted within the first three minutes? If not it paints it as a red dot and then people essentially act on it and where we have deployed this technology, they're already seeing that they're able to save one or two drive offs a day which pays for the technology. The one day savings pays for the entire month or for the technology cost. So so so lots of great examples of how AI is being used today and what we had to do to put that in place 

Saam: Yeah, those are great examples, Vishal. I like the toner example in particular, because I think one of the interesting things about AI and, you know, People see these like really magic like examples of AI. But I think as important, if not more important, are all the ways it's being incorporated into how businesses are run. Right. And some of those ways may sound small, but when you think about like the second order of consequences, they're actually quite profound. And so maybe just kind of continuing on that thread, right? Are there a couple of wins that you've already accomplished at Lexmark, either in your products or in how you run the business internally that you can share? 

Vishal: Yeah. Let me share maybe three or four. Right. We talked about the RFP one, which they are the big advantage that we see is, you know, we think that will create almost 5 million in incremental benefit a year because we used to turn down a lot of RFPs if we didn't have two weeks to respond because that's the time the team needed to do a quality job. 

Only the second one that we've seen from a market sales perspective where we're leveraging somebody else's Gen AI that was built into another product is, you know, you think about prospects. When you are going after prospects, typically there's a 1 percent response rate. What we did was, we leveraged this tool which could create a more tailored  email to the prospect based upon their social feed, based upon which product we think would resonate better with them, and what words would resonate better, so imagine a tailored email going, and now we're seeing a response rate of 2-3%. So almost a 2 or 3x improvement in the response itself of people responding to, as new customers, to things that we want to do. 

The third very interesting one, which is not in production, but we're working on it, which is going to have a massive impact on us, is customer care. I think customer care could actually be one of the biggest impacts of generative AI. You know, we employ several hundred customer service agents because we want when our customers call them to be able to, you know, have somebody pick it up and respond to them within 15 seconds. Now what happens is that average handling time is about 11 minutes and we believe generative AI can cut that down by 50 percent because right now it takes us a lot of time to troubleshoot and figure things out.

With generative AI, right, we can analyze all of the data in the past and, and almost kind of create a life script based upon the customer who is and what they're saying, and so we believe that service insight thing can cut down very, very significantly and also get, can get automated in a lot of cases. That could be tens of millions of dollars worth as the full solution for something like that rolls out. So we're very excited about that. 

We're also incorporating it in our products in simple ways. I'll give you an example. We've created an optra product, something called the AI assistant. And so what that does is it's like a how to, right? So in the past we would tell people, Oh, go, go and attend like some kind of training. Nobody has time for training these days. And so you have a novice, they're using the product and they just basically ask the, the, they can literally talk to the product and say, Hey, how do I do this? And because we've embedded the AI assistant in the product, the generative AI assistant will actually tell them the steps to take to be able to do this. 

We've also added something called the coding assistant and that's a very interesting thing in the product as well. What that does is if people have requests for reports that are not already prebuilt, they can actually say, how do I get to this kind of data and it'll generate the SQL automatically for them. So from English language, it will essentially generate SQL, which is Generative AI actually is very good at doing, and this means that they don't have to go to a professional services or, or go to their IT to figure out how to get these new pieces of insights that they didn't, you know, that the product didn't have.

So they can pretty much generate those so this product and data assistance can essentially make our products much more valuable to our customers. So these are some of the examples from a process and product perspective that we're seeing applications of generative AI. 

Evan: Thanks, Vishal. Appreciate you sharing some of the, those wins. I think it's good to give people some real examples of how AI is having impact today. 

I wanted to kind of maybe move to like the future, right? So if I, if you forgive me, I'd love to ask you to kind of take out your crystal ball. When you think about, you know, how AI can transform your product, your services, your offerings, or maybe change how the company operates. What do you think like, you know, five years from now, like if we, if we did kind of version two of this episode, five years from now, and, you know, Saam was asking, Hey, Vishal, what has been like the big three wins you guys have gotten from AI, you know, now that we're sitting here in, you know, 2029, like, what do you think, what do you think is going to be?

Vishal: I think fundamentally, my hope, Evan, is that what AI is able to do at the crux of it is to level the playing field for any of us to achieve our ideas in a better, faster way and achieve our full potential.

That to me is the ultimate power of possible, right? That we live in a world today, right? Where because I'm in the U. S. I have an unfair advantage and that's good because I'm in the U. S. but may not be so good when you're not in the U. S. Right. And so I think about this really from a perspective of a true democratization that with generative AI, with AI, right? Just like internet made the information available to everyone. I think generative AI will enable that use of information to create that advantage for folks, so the friction in the world goes down very significantly. You're able to achieve your ideas in a more seamless way. So that I think is what the opportunity is. 

Now you and I know that technology can be used by good guys, it can also be used by not so good guys. And so we've already seen, for example, and it comes for fishing, you know, people coming up with generative AI led fishes, which now many more people click on, right?

And so there'll always be a race, I think, as well, between the good and not so good, right? And so we'll have to stay on the, on the front end of it to make sure that we're guarding, you know, how these technologies get used, how do we make sure that they get used for, for the better, and, and how do we put, you know, enough guardrails Around these things so that the way the trust can be there so that way the misuse doesn't happen, and so I'm cautiously optimistic, Evan, but we'll all have to work also towards making sure that we can minimize the misuse of these as well, which is inevitable 

Evan: Alright, we do have a limited time, so one thing we'd like to do at the end is do a bit of a lightning round just to get some of your quicker hits or kind of the one tweet responses. So, we got maybe like five or six questions to, to throw in this lightning round. So, um, Saam, you want to kick it off for us? 

Saam: Yeah, absolutely. So Vishal, maybe to start, how do you think companies should measure the success of a CTO or CIO? 

Vishal: By based on innovation, right? What are the new things, either new things or existing things that they can make better?

Evan: What's one piece of advice you wish someone told you when you first became a CIO or CTO? 

Vishal: Network more. I wish they told me that. Now I do more of it, and so like now I'm the co chair of Iwanta CIO Council, and I find that when I talk to 200 of my other peers, the challenges around people, around technology, around change management, around all of these are so similar, that there's a lot to be gleaned from talking to them. And this is where I think, having a show like the one you're doing is very helpful.

Saam: Vishal, what do you think most people underestimate about AI? 

Vishal: I think people underestimate the accuracy aspect of AI, right? So they, they think more about how do I get a project done, and they don't think about how do I actually measure accuracy on an ongoing basis to know whether or not I can really trust it. And so being objective about it by having a set of metrics will let you be objective about it where you're neither, you know, leaving a comparative advantage on the table, nor are you completely putting yourself at risk.

Evan: Maybe, maybe switching gears to more of the personal side, uh, what's a book you've read that's had a big impact on you or your leadership or how you think about the world? 

Vishal: You know, it's so interesting. We think about all the new books, but, if you read this one around seven habits of highly effective people, I find that it is amazing how many things are still true, for what were said in it, right? It's like the world hasn't changed. Like on one hand, we're talking about AI and all, there's no AI, there's no technology, there's no cloud at time. But the same things still apply. I have read hundreds and hundreds of books, but some of the ones that are written like 50, 80 years ago, it's amazing how true they still are.

Evan: What's an upcoming technology that you're most excited about? 

Vishal: I'm very excited about Edge AI. And I, and the reason I say that is I think the, the benefit of Edge AI is it can act real time, right? And so if I think about our physical world in which things are happening, Edge AI will let you, you know, if I think about any process we're in, If it can help me make it more efficient because it's looking at people and things and telling it how to optimize it, that's a better experience for all of us, right? That's more real in the environment we are in versus, you know, just Initially, we used to think about just decisions, right? So we're kind of past the decision points now to say, how do we make our experiences better? So I'm actually very excited about Edge AI. 

Evan: Vishal, I wish we had more time, but I really appreciate you taking time and joining us on the show. Thank you so much, and great to see you again. 

Vishal: Great to see you as well, Evan. 

Saam: Thanks a lot, Vishal. 

Vishal: Thanks, Saam.

Evan: That was Vishal Gupta, Chief Information Officer and Chief Technology Officer at Lexmark

‍Saam: Thanks for listening to the Enterprise Software Innovators podcast. I’m Saam Motamedi, a general partner at Greylock Partners.

‍Evan: And I’m Evan Reiser, the CEO and founder of Abnormal Security. Please be sure to subscribe, so you never miss an episode. You can find more great lessons from technology leaders and other enterprise software experts at enterprisesoftware.blog.

‍Saam: This show is produced by Luke Reiser and Josh Meer.

See you next time!