From Pilot To Impact, Building AI That Scales

[00:00:00] Tessa Burg: Hello, and welcome to another episode of Leader Generation, brought to you by Mod Op. I’m your host, Tessa Burg. And today I’m joined by Shashank Kadetotad. He is the Global Senior Director for Enterprise Data Science and AI at Mars Snacking. We’re so excited to have him because while we’re gonna be talking about a topic that’s pretty common on this show, and that is how do we move from AI pilots to scale, Shashank brings a very unique perspective because he himself is deep in data science, and has architected and led AI programs at some of the largest consumer brands in the world. 

[00:00:43] Tessa Burg: And I’m just really excited to hear his expertise, and for him to share his insights with you on how you can start taking those pilots, those, the success that’s sitting in disparate and siloed areas of the business, and begin to drive real business value. Shashank, thank you so much for visiting us today and, and helping us navigate this very complicated topic. 

[00:01:07] Shashank Kadetotad: Well, thanks for that awesome intro- introduction, Tessa. Happy to be here. Uh, the pleasure’s all mine.  

[00:01:14] Tessa Burg: Awesome. Well, tell us, let’s start with hearing a little bit more about yourself and, uh, your career journey.  

[00:01:20] Shashank Kadetotad: Sure. So, um, I currently lead the enterprise data science and AI team at Mars Snacking. It’s been close to four years, uh, with Mars Snacking. It’s also my first time in CPG. Prior to Mars Snacking, I was with Walmart and I was with them for close to four years as well. And my experience so far has primarily been in retail, uh, with Mars being the exception, being a CPG company. 

[00:01:49] Shashank Kadetotad: So it’s been Walmart, and then it’s been, uh, a purely analytics firm called Mu Sigma, and I used to work with Home Depot then again, specialty retailer. Uh, prior to which I was with Amazon, uh, which obviously is, uh, an e-commerce retailer. Uh, so it’s primarily been retail, uh, with around, uh, 18 to 19 years of experience again, primarily in data science and analytics. 

[00:02:13] Shashank Kadetotad: So I’ve seen the, quite a journey from, you know, what data science and analytics was close to 18 years back to now, right? And it’s been, uh, a whirlwind of a change, especially in the last couple of years.  

[00:02:27] Tessa Burg: Yeah, what I think is so interesting about your career is it’s not just about deploying AI. Like, you have a technical background, but you really focus on the organizational structures, the governance models, and the business alignment in order to drive the AI. 

[00:02:48] Tessa Burg: So when you are starting a new AI program, like, how do you help bring people together across an organization? And, and what are… How do you even know where to start when you’re, you’re trying to put together a new organizational s- structure? And the reason why I think that’s unique is what I’m seeing is a lot of clients, they know their use cases, they know their problems and their challenges But there’s– they’re grabbing tech first, and they’re saying, “Maybe I should just use this app.” 

[00:03:20] Tessa Burg: And so I find your background really interesting to be like, “Well, maybe we need to look at the organization first.” So tell us a little bit more about that.  

[00:03:27] Shashank Kadetotad: Yeah, sure. Great question. Uh, it lines up really interestingly with, uh, the way I– my thinking has evolved about this field, right? Uh, so if I were to think back to even, uh, let’s say five to seven years ago, right? 

[00:03:45] Shashank Kadetotad: Um, there’s a very templatized way of implementing AI, data science, and analytics, right? Uh, you have a problem statement. You focus on solving the problem statement, either through advanced analytics, through some modeling. Uh, you know, you figure out what the, what data is required. You figure out what kind of best fit models you have. 

[00:04:07] Shashank Kadetotad: You deploy the solution, and then you iterate on it. I think, I would say in the last two years specifically, uh, what has happened is the creation of models has become extremely fast, right? So something that would take let’s say you know, a few months or even in some cases a few weeks can now be done in a few days. 

[00:04:31] Shashank Kadetotad: So with that in mind, then the bottleneck is not the quality of the solution or the technical sophistication of the solution anymore. I think it quickly becomes everything else surrounding the solution, right? Uh, which is how do we make sure the solution is deployed properly? How do we make sure it is adopted? 

[00:04:52] Shashank Kadetotad: How do we make sure it can iterate with an evolving problem statement? How do we make sure that the organization can actually support it? And as I’ve grown in my career, I think, uh, one thing I have definitely realized is the organization structure and mindset and culture plays an extremely important role in this, right? 

[00:05:14] Shashank Kadetotad: Uh, obviously, uh, in a more junior role, uh, you know, I would say it was all about solving the problem and bringing the biggest and the best solution to life, versus now I think, I think a lot more systemically and organizationally, right? So even if you build the best solution, I keep telling my team this, uh- You know, you it’s like running a 100-meter race. 

[00:05:36] Shashank Kadetotad: You run the first 90 meters really well, which is creating the solution. But then if the last 10 meter, meters, which is adopting the solution, is not up to par, then the last 90 really didn’t make a difference, right? It’s the last 10 that will make or break the race, and it’s where adoption, you know, where the rubber meets the road. 

[00:05:56] Shashank Kadetotad: So organizationally, I can say with this evolution of AI, uh, organizations have to definitely rethink how they are structured, right? Uh, in fact, uh, you know, you have some more native orgs like Amazon, and I have, you know, friends in Walmart who I am seeing them go through this change, and we are doing that in Mars Snacking as well, right? 

[00:06:18] Shashank Kadetotad: Where there is an understanding now that, uh, there is a major sh- paradigm shift in which AI can be adopted, the way AI can be adopted. And it’s really critical to be able to create an organization that can adopt it, right? So you– we, we need to make sure that incentives are aligned. We need to make sure that the goals of the organization, of different departments are aligned to a new structure and a new way of thinking, right? 

[00:06:47] Shashank Kadetotad: Uh, so I would say culture and organization, uh, becomes more important than just the technical sophistication. Because both– most companies now will have access to this technical sophistication, so differentiation will really be, uh, how the organization looks at it, how the organization will continue evolving with this change, right? 

[00:07:08] Shashank Kadetotad: Uh, and how adaptive they can be in this new world. Yeah.  

[00:07:13] Tessa Burg: You’ve worked at companies that I- at a, the surface level looks like they would have very different cultures and very different priorities on how to organize departments within the org. How do you reconcile that? Like, where the… I did get the opportunity to work inside Amazon for a, a short stint, and I have also consulted on Walmart, which is the… 

[00:07:43] Tessa Burg: But just, like, from my external lens-  

[00:07:46] Shashank Kadetotad: Right …  

[00:07:47] Tessa Burg: they’re very, very different organizations.  

[00:07:50] Shashank Kadetotad: Yeah.  

[00:07:50] Tessa Burg: And how do you know what incentives work for which departments or, or even where to start to get that cultural alignment and buy-in from the people on the teams that you’re looking to move?  

[00:08:06] Shashank Kadetotad: Yeah. No, that’s a great question. 

[00:08:08] Shashank Kadetotad: You’re right. Culturally, very different companies. You know, I started off my career in the US here with Amazon, so to me that has always, um, you know, it kind of created the kind of professional I am today, right? In terms of it was my first exposure to how an organization thinks. And, uh, it was a very agile, fast-moving culture, right? 

[00:08:31] Shashank Kadetotad: So to me you’re right, right? It was a very young company even when I joined. I think it was a teenager back then. We used to call it a teenager, right? It can, it has the nimbleness, it can move real quick, uh, and even the kind of hiring practices would get towards that, right? Uh, so Amazon has these principles, leadership principles, and a lot of it has to do with bias for action, right? 

[00:08:53] Shashank Kadetotad: So, uh, being able to discern, and this is something that we used to actively measure for, right? Someone who’s able to kind of identify whether an action is wh- whether an action can be rolled back or not, right? So those are different kinds of decisions. So if it’s a decision that, uh, you can quickly make a quick decision, see it’s not working, and then roll back versus a more permanent decision. 

[00:09:18] Shashank Kadetotad: So the ability to kind of discern the kind of decision you’re making, and if it’s something you can roll back, then just go ahead and do it instead of wasting time thinking about it. So that was very much, uh, the culture. Uh, so when I joined even a company like Walmart or Mars, these are, like, 100-year-old companies, right? 

[00:09:35] Shashank Kadetotad: Uh, and they come with a very different thought and very different mindset. So Walmart, I thought, I thought was for a, for such a large organization was extremely nimble as well, right? Very clearly they, back then I was a part of Walmart E-commerce, and they were kind of making moves to really compete in the e-commerce space with, you know, with the Amazons of the world and several other competitors, right? 

[00:10:01] Shashank Kadetotad: So you could see that culturally, uh, while they held- their principles which was very associate-centric principles, but they were willing to kind of make that change, right? So you would, uh, one, you would start looking at the kind of hires that were being made were, you know, very, uh, different to the standard hires you would make. 

[00:10:22] Shashank Kadetotad: The ability to expect technical understanding, systemic understanding, even at the higher levels, right? So I started seeing, you know, folks come into SVPs, you know, who had backgrounds in, uh, data science or data engineering. So you had people who were more than just people managers. I could see that, you know, they were bringing people who understood what the challenges were and how to really, you know, navigate the complex technical challenges, not just, like, the people challenges aspect. 

[00:10:52] Shashank Kadetotad: So, one, hiring the right people, and then second, also the ability to kind of, allow those people to create their own space and take it in that right direction, right? So I could see the kind of ownership being given to those hires, right? Um, versus, you know, so even now, recently, Walmart just changed, uh, their CEO. 

[00:11:13] Shashank Kadetotad: They have a new CEO, and it’s very widely known, and in fact, you know, the outgoing CEO talked about how they’re facing a new technical landscape, which is why he thinks the newer CEO is more equipped to handle the changing technical landscape. So it’s very well– So it’s very heartening to see such a robust, uh, company and such robust legacy willing to look at how disruptive the future can be and kind of modifying and making these leadership changes to be able to adapt to it, right? 

[00:11:45] Shashank Kadetotad: And that, I think, sends a lot of the right signals throughout the entire organization, right? So, so then it tells you what’s really important for the organization, how the organization’s thinking, and consequently, how should I think about this, right, as an associate. So, uh, while there were obviously some challenges when I first joined Walmart, in the four and a half, five years I spent there, I could see the huge transition happening, right? 

[00:12:12] Shashank Kadetotad: Uh, and it was really good to watch. Really good to see how they embraced that cultural change. And I would say right now, you know, and there were some major changes, right? Uh, you know, there were roles that were totally, completely, um- You know, rewritten, uh, redefined. I was allowed to redefine so many roles, uh, get in people that never- that Walmart wouldn’t have ever hired in terms of, you know, let’s say even five years back, seven years back, who started to form the core technical competency group, right? 

[00:12:43] Shashank Kadetotad: And today, Walmart has a huge technology group, right? Uh, and it’s, I think, directing the organization versus technology just being an enabler. It’s now reached a point where the technology is kind of defining what, uh, Walmart should do next, right? Uh, so it’s a very interesting change. And then coming to Mars. 

[00:13:04] Shashank Kadetotad: Mars, again, 100-year-old company, very brand-focused, owns iconic brands like Mars, uh, Skittles, M&M’s, uh, right? Uh, so, uh, with with Mars, I think it’s been a little more challenging. Most CPGs are not as tech-savvy as retailers, right? They didn’t really need to be in the past. Uh, retail has always been competitive, but, uh, uh, CPGs not so much. 

[00:13:29] Shashank Kadetotad: But I think that has quickly changed, even in the time I’ve been here, uh, right? Uh, we, for example, three years back now two and a half years back, we established an AI lab in Newark, right? The first of its kind. Uh, and the only reason we did that was because, one, there was a huge, uh, change in the la- technology landscape that we wanted to start exploring, and we really wanted to bring this mindset, cultural shift of, hey you know, there are new technologies out there. 

[00:14:00] Shashank Kadetotad: We don’t know yet how it’ll apply to us, but we wanna start exploring. We wanna experiment. We wanna fail fast, learn fast, and then iterate, right? So, um, so I think that was a whole, uh, mindset shift for Mars as well when we first established our AI lab. And, and today, I think we are very well-versed with some of the latest technologies. 

[00:14:23] Shashank Kadetotad: Uh, we have been able to– We have spent a lot of time and effort in training and retraining people. Uh, we have been able to kind of make some intelligent organizational changes right? Uh, and kind of bring the organization a lot closer, uh, versus it was, I think, a lot more siloed even a few years back. 

[00:14:44] Shashank Kadetotad: Yeah.  

[00:14:45] Tessa Burg: Yeah, I… That was such a rich answer, and there were some common themes that even though the cultures were different, what you were looking for is, are we able to be adaptive? And is the sig- is- are the signals from the leadership team saying, “Hey, our expectation is that you, as our team of people powering our company, are going to be able to adapt and move and change as we take on a new and very necessary focus to be competitive.” 

[00:15:14] Shashank Kadetotad: Right.  

[00:15:14] Tessa Burg: And that is… One, it’s fascinating to me that you as a data scientist have been able to have that much empathy. Like, I’ve worked with data science teams and they, and they d- they trust the data, but sometimes they ignore the context in which the data is coming. And what you’ve kind of laid out is context is equally important to the change and the reason for the change itself, and that we have to be aware of that. 

[00:15:43] Tessa Burg: The other thing that you highlighted that is really interesting is that leaders who have a technical background are able to better understand the challenges faster. And I 10,000% agree with that, and I don’t think every company understands the value someone with a data science background brings because you hear businesses talk a lot about we have to start with the data, but there’s a misconception right now that the amount of data matters, that we have to go get more and more data, and not the data features or the quality of the data and how it aligns to that challenge. 

[00:16:30] Tessa Burg: And when you have leaders who, like yourself, can be highly empathetic, can see the context with the challenge, they’re able to better understand how the organization has to adapt, what we’re actually going to do with it here to meet the higher level business objectives. And yeah, so I- What– I don’t really have a question other than those are my observations. 

[00:16:59] Tessa Burg: I think that’s really powerful, and how have you been able to l- to make highly technical challenges more accessible so that everyone does understand sort of the why and the trigger point for that change to happen?  

[00:17:17] Shashank Kadetotad: Oh, uh, first of all, I think that was a, that was a great uh, summary and synthesis, right? 

[00:17:23] Shashank Kadetotad: Now, um, I think to your question that has always been a challenge, right? Uh, I think as a practitioner myself, I can tell you, uh, and I can probably speak for the data scientists or data engineers, right? And say that unfortunately or fortunately, you– I mean, most practitioners tend to fall in love with their solution, right? 

[00:17:44] Shashank Kadetotad: Like- Mm-hmm … you know, like I spend weeks creating a model, uh, my team does or someone else, right, who’s technical does, and, uh, suddenly that’s the best solution in the planet. You know, and there, there is that whole you know, ownership, which is great. But also I think on the flip side, right, uh, that’s where I think a lot of the leadership empathy and objectivity needs to start coming into place. 

[00:18:10] Shashank Kadetotad: Uh, and, and one thing I tell my team is, you know, when we look at, you know, any business problems, right, typically our– the way we would handle it is you have a business problem, uh, you know, you kind of, crystallize it, you know, uh, it’s a very crisp, uh, business problem. Then you figure out what are the metrics, what change, and then from there on you trans- kind of translate it to a technical problem, so to speak, right? 

[00:18:36] Shashank Kadetotad: And then from technical problem, then you, it informs, uh, you know, the hypothesis and then the data and then so on and so forth, right? To finally solve that business problem. So we, we have an analytical framework we follow. But then to your point, right, I think one of the major things, because the challenge again, like I mentioned, is rarely the how difficult a technical solution is. 

[00:18:59] Shashank Kadetotad: Because one, while I would say, you know, few years back it was critical to be able to… It wasn’t easy getting, uh, or modeling a, a very sophisticated solution, right? Uh, and that’s where that kind of expertise, uh, really came in handy. And I think now with the advent of, uh, GenAI, agents right, and the whole LLM aspect of AI that is now becoming so much more popular, right? 

[00:19:29] Shashank Kadetotad: Um, get f- defining and creating a technically sophisticated solution is easier. Like, there’s no, uh, there’s no two ways about that. It’s something that would take me, you know, weeks it will now take me, uh, days. In some cases, it takes us even hours, right? And the idea then is, how do you iterate? So, so the big part of what I think that role is now is not just creating the solution, because that’s always been a part of the role, but it is how do we make sure that the solution is effective? 

[00:20:05] Shashank Kadetotad: Because at the end of the day, the stakeholder is not not buying solutions from you, right? They’re buying outcomes. That’s what I keep telling my team. We’re, we’re not here– If you go and tell someone who’s not technical, “Hey, you know, I created this great regression model with all these variables.” Yeah, I mean, okay, what can I do with it? 

[00:20:25] Shashank Kadetotad: Like, how does it impact the problem or the KPI I’m looking to, uh, solve or hit, right? So that’s where I think that mindset of a lot more outcome-focused than output-focused, right? Instead of talking about in terms of, “Hey, these are the great models we have built,” “Hey, with these models, we are then able to achieve so-and-so outcome,” right? 

[00:20:44] Shashank Kadetotad: So I think very quickly you know, o-one thing that has changed personally for me as well is moving from a super highly accurate model to a model that is good enough, right? Good enough but is robust and explainable, and we are able to iterate it. So if those factors are met, right, and it solves the problem, uh, to a high degree, uh, may not be 100% of the problem, but good enough accuracy to kind of solve the problem, and then we can iterate it. 

[00:21:15] Shashank Kadetotad: You can then deploy the solution more quickly, and then you can iterate the solution with, uh, the stakeholders or the users, right? Uh, so I think that becomes a much more critical play because now you can iterate more quickly, you can build more quickly, and you can make changes more quickly. So I think that is one, you know, major factor. 

[00:21:34] Shashank Kadetotad: Uh, the second thing is, you know, going back to the whole technical leadership point why I think it is so important for business leaders to either bring themselves up to speed in technology or technical leaders to understand, you know, business, I think that has always been the case, right? Technical leaders have to always understand the business they are in really well, uh, to be able to propose and work through, uh, impactful solutions. 

[00:22:02] Shashank Kadetotad: But even now, since the changing landscape is so tech-driven, um, right, it’s really critical for those higher-ups to be able to see the organization in that light. What, what I mean by that is you know, something as simple as, okay we have, you know, a marketing-focused organization, we have a supply-focused organization, you have finance, HR, you know, all these other, other silos and departments. 

[00:22:27] Shashank Kadetotad: But how can I start thinking about with this new technology, with you know, how easy it is to generate insights now, how easy it is for me to build sophisticated solutions? How do I now start answering questions that I couldn’t answer bes- before? What questions can I realistically answer versus not, right? 

[00:22:46] Shashank Kadetotad: It’s also important not to buy into the hype, uh, but it’s also important to recognize that we are getting to a point where, uh, you can easily do things in business that you couldn’t do before, right? And that’s where I think it’s really important for those business leaders to really have that technology, uh, bend of mind to be able to implement those kind of things. 

[00:23:08] Shashank Kadetotad: Because hypothetically now w- you don’t really need a separate you know, data scientist who’s focused on just supply or a data scientist who’s focused on just marketing, right? Now, a data scientist with enough context should be able to kind of build and connect solutions across these kind of departments, which is so much more powerful, right? 

[00:23:30] Shashank Kadetotad: Mm-hmm. And that’s something we have actively focused on, right? Uh, our supply signals giving the right signals to our to our marketing strategy and vice versa, right? So, uh, those kind of things that, you know, a few years back would be very difficult to achieve, uh, is very real now, right? And it’s kind of informing the way we operate going forward. 

[00:23:51] Shashank Kadetotad: Yeah.  

[00:23:52] Tessa Burg: Yeah, I love that. So when you’re working on tech that’s spanning different departments, who– what voices are at the table who’s helping to s- set the design, like to understand the process and then design the solution in collaboration with you?  

[00:24:09] Shashank Kadetotad: Yeah. So I think one is it’s really- And it’s very easy to fall in this trap. 

[00:24:15] Shashank Kadetotad: So I’ll just talk real quickly about what you shouldn’t do, and then we can talk about-  

[00:24:19] Tessa Burg: Okay. Yeah  

[00:24:21] Shashank Kadetotad: … what I’ve seen. Right? Like, I think I think it’s very easy to fall into a trap of building a solution for someone who won’t use it, right? What I mean by that is if the users who are going to use your solution are not actively part of the build you’ve lost them right there. 

[00:24:38] Shashank Kadetotad: Right? So yeah, this, th- through my career I’ve seen this. I won’t name any particular company, but, you know, all the time we’ll hear some some SVP somewhere say, “Hey, this is an exciting new tool,” or, “exciting new technology”, and then everyone rushes to kind of build a product on that technology. But we need to understand we are not building that product for that a- SVP. 

[00:25:04] Shashank Kadetotad: He’s not going to sit and do brand creation or brand briefs or make marketing decisions, right? That … So I think it’s very easy to fall into a trap of understanding who the sponsors are versus who the users are so that’s one, I think, critical element, right? And then the second bit is every time we work out you know, like what the team, cross-functional team should look like or what the team should look like. 

[00:25:32] Shashank Kadetotad: One, it is imperative that, like I mentioned, the users are a part of it, but it’s also imperative that, So let me just take an example, right? When you create a solution, it’s you can, one, create a very point-focused solution for a very particular problem, right? But at the end of the day, as an enterprise team, we don’t want to do that because we are not in the business of solving point problems. 

[00:25:58] Shashank Kadetotad: We are in the business of creating a system that will solve large scale problems. So for that to happen, we then need to not only think about X number of point problems, but we need to figure out how are those problems connected. So how is problem space A connected to problem space B connected to pro… 

[00:26:18] Shashank Kadetotad: Is there a connection? And more likely th-than not, there’s always a connection. How are these KPIs connected? How are these metrics connected? And that will then inform us, right? From a solutioning perspective, can we not reinvent the wheel every time we solve a problem X and problem Y, but we solve this kind of a problem that will eventually can be applied to what may seem disparate problem spaces but are actually really well connected, right? 

[00:26:48] Shashank Kadetotad: And then able to articulate how solving problem A this way has impacted problem B and then problem C, right? Uh, so it is designing and creating that system. And then the third part in terms of who all the participants are, uh, I think other than the tech teams, we make sure right from the start that all the users, we have a good representatives of the user sample who’s there, whether they like it or not, right from the beginning, right? 

[00:27:15] Shashank Kadetotad: And they help decide how the interaction will flow, how everything will flow. Uh, then we’ll have obviously what we would call, you know, domain experts who are experts in that particular area or that particular field or are doing this day in and day out, uh, who will map out what the process has always looked like, right? 

[00:27:33] Shashank Kadetotad: And then the challenge becomes how can we rethink this entire process altogether, right? And that’s where I think agents become, you know, uh, super interesting because, uh, they, they in a big way will change the way of operation and in, in, in general can automate a lot of processes, right? So then the conversations start to become what decisions need to be made by us versus agents, right? 

[00:28:00] Shashank Kadetotad: And what decisions can agents do? These are low impact decisions, so okay, we can have agents do this, or, you know, use an Iron Man model. Uh, we call it an Iron Man model, which is basically an agent plus a human will make a decision, right? The- … decision will be made by the human, but, uh, with the help of an agent, right? 

[00:28:19] Shashank Kadetotad: So essentially there are these kind of aspects, and then obviously, you know, we have something called an AI council, which we have put together. It’s probably more than a year old now, which is essentially every large scale project will go through this council, which has representation from legal, from HR, from finance, from you know, marketing, so on and so forth. 

[00:28:41] Shashank Kadetotad: So everyone’s aware what problem we are trying to solve, and it kind of goes through this council. So we avoid duplication. Uh, we avoid that. There is an understanding of impact on each of these organizations, and each of these leaders know what they’re signing up for, right?  

[00:28:58] Tessa Burg: Yeah.  

[00:28:58] Shashank Kadetotad: And that’s kind of how we take it forward, yeah. 

[00:29:02] Tessa Burg: I love… I think there’s a lot of steps in there that if listeners pause and re-round your answer, they could create a checklist of-  

[00:29:11] Shashank Kadetotad: Yeah  

[00:29:11] Tessa Burg: … what they may have missed that blocked them from scaling adoption.  

[00:29:16] Shashank Kadetotad: Right.  

[00:29:16] Tessa Burg: And where you started is really important, understanding sponsors to who the users are, and getting them involved right at the front, and them being a part of defining and crystallizing the problem you’re solving. 

[00:29:32] Tessa Burg: I- I’ve seen before where sometimes when you only have, uh, the way the sponsor define the problem, and we don’t get into the nuance of how that problem actually impacts those in execution, it’s two very different perspectives. So sponsors are great at vision and KPI and, and what goal we need to hit, and then how we’re gonna do that is equally important, but those are two different roles. 

[00:29:59] Tessa Burg: And I think that’s a really important distinction. The other distinction that you made that is really interesting is that- What we build isn’t necessarily as important as how we build it and, uh, understand the impact of when it goes and gets rolled out. And I love that you’re prioritizing the different types of decision, and I really, really like the Iron Man analogy. 

[00:30:26] Tessa Burg: Because if you don’t, if people don’t understand the impact before you go and try and drive adoption, and they don’t understand that you did separate out the types of decisions or the types of automations, it can feel really threatening, and it can also feel like you’re just kinda coming at this from out of the blue, and, and you don’t have the trust signals that are necessary. 

[00:30:49] Tessa Burg: You didn’t show that you went through some vetting, that you had the right people involved and the friends. I, I really love that focus on what’s gonna be the impact? Let’s make sure we understand that before we just start talking about it and rolling it out and expecting people to use it  

[00:31:08] Shashank Kadetotad: Yeah. No, precisely, right? 

[00:31:09] Shashank Kadetotad: And, uh, one of the things I, I was just going back in my head, right? Uh, I think you covered it pretty well. Uh, like I, I talked a bit about, you know, practitioners falling in love with their solutions, right? And that’s one of the thing that you know, I talked a bit about the AI lab as well. So when we– One of the reasons we started the AI lab was obviously there was this exciting technology paradigm that we wanted to explore and fail fast. 

[00:31:37] Shashank Kadetotad: And very quickly it became a thing where, you know, as you can imagine, everyone was excited about this, and everyone wanted to explore a lot of things, right? And we ended up launching so many pilots and, you know, uh, that’s where it became clear to me, and we have, like, now implemented this, where, you know, if we are going to first, you know, if we are going to explore a pilot, something new, right, something we haven’t done before, first let’s make sure it’s new, right? 

[00:32:07] Shashank Kadetotad: Like, that’s why we have a catalog and, you know, we are like, okay, it’s a large company, so people could have done something somewhere else that, you know, we can learn from. But then along with defining success parameters for even a pilot, right? Like, hey, this is success parameter for pilot and potentially how it can scale. 

[00:32:25] Shashank Kadetotad: Uh, like I talked about, you know, like designing systems versus solving part problems. But, uh, it’s also at what point, you know, defining what the parameters of failure are, right? Like what I mean, that is actively kind of saying that, okay, if I’m not able to achieve X or Y parameter by so and so time, I will retire from building the solution. 

[00:32:50] Shashank Kadetotad: Or we will, you know, create a learning document and let’s move on, right? I think o-one learning has been, you know, people sticking to pilots. Like, things shouldn’t be pilots for long, right? Uh, they should actively be pilot… Whatever, it depends on company to company. But pilots for only so much time before which you should– and you should already have a vision for scale. 

[00:33:12] Shashank Kadetotad: And if it’s still a pilot or, you know. So this forces the practitioner to start looking at these things more objectively, right? And start thinking about, okay, like, you know, realistically it’ll take me X months. Realistically, this is the accuracy I wanna hit or whatever. These are KPIs I wanna hit. And if I’m not able to hit them, I will retire it, right? 

[00:33:34] Shashank Kadetotad: So that saves, I think, a lot of time versus… A- and, you know, the whole fail fast mentality versus, you know, having just success parameters and doing whatever it takes to hit it. Because this technology’s moving so fast, something that I couldn’t do, like, you know, two months back, there’s a product on the market that can do that, right? 

[00:33:54] Tessa Burg: Mm-hmm.  

[00:33:54] Shashank Kadetotad: So then it starts to get into the more complex thing of build versus buy, right? If it’s easier to buy something that will quickly show us the impact that we have tried and, you know, is going to be potentially more expensive to build, so why not just do that, right? So, uh, I think that policy is something that has been difficult initially to kind of put through, but has been so helpful to us, right? 

[00:34:19] Shashank Kadetotad: Where, uh, the folks who propose the pilots are the ones who’ll also discuss and put in paper you know, when they will abandon that project, right? And avoid the whole falling in love with the solution thing.  

[00:34:31] Tessa Burg: Yeah. No, I- that is extremely important. I can relate to that so much. And with all of this change and all the activity and excitement, there’s this reality that kiddos just coming out of college right now are st- are struggling to get entry-level jobs. 

[00:34:51] Tessa Burg: When we– For us that are already in it and our roles are evolving, we know how to evolve, but what would you say to entry-level professionals? What kinds of skills should they be building and leaning into so that they can also be a part of helping to build and scale these solutions in their careers? 

[00:35:12] Shashank Kadetotad: Yeah. No, absolutely. Uh, I think great question. I was actually having this discussion with my own team, right? Clearly even the career, current career paths we are on are evolving now, right?  

[00:35:24] Tessa Burg: Mm-hmm.  

[00:35:24] Shashank Kadetotad: That’s how profound this kind of paradigm, this technology paradigm has been. So I would say one thing that, you know, I have always constantly looked for all the way, and if I look back on anyone who’s been successful in their career and ha- one aspect that has never changed has been I would say three, three areas, right? 

[00:35:45] Shashank Kadetotad: One is critical thinking which I would say, you know, irrespective of which domain you know, it’s immaterial, right? Critical thinking is super, super critical to, uh, I think especially if you’re in the tech area, it becomes one of the biggest shining factors, right? Uh, uh, and when I say critical thinking it’s things like being able to see an ambiguous problem and then distill it and break it down into its components and then being able to solve for each component. 

[00:36:17] Shashank Kadetotad: Now, when you talk about solutioning, right? Uh, there’s obviously, you know, now technology can give you the power to solve these smaller problems without being a super expert in the technology. Uh, but ultimately, it- it’s critical thinking that will take you from taking something that’s super ambiguous into and crystallizing it, right? 

[00:36:38] Shashank Kadetotad: So that, so critical thinking, uh, is, it’s something I would really focus on, and it’s something that I have personally tested for in every single person I’ve hired, right? The second is, uh, I think, uh, you know, this wasn’t necessarily, uh, a- s- it was always important, but I think it has only become more so, uh, with the recent technological change, I think is the ability to bring a vision to life, right? 

[00:37:06] Shashank Kadetotad: Which is, you know, now y- I mean, there’s obviously a lot of hype around this as well. Not everything we see and hear is reality. Uh, but also I think, uh, we’ll be getting there soon. It may not be true today, but it’ll be true, you know, in a year or two years from now. So the idea is the- there’s always going to be a gap and place for people who are able to bring you know, a vision, uh, right, which could… 

[00:37:33] Shashank Kadetotad: It could be a problem statement, it could be an application, it could be you know, uh, like, art of the possible, right? The ability to kind of bring that and articulate that to the right people, uh, be it technical, data engineer, scientist, whoever, right? And be able to do that effectively and iterate and then bring that vision to life, I think that’s never going to go out of style. 

[00:37:58] Shashank Kadetotad: Like, that is always if… It has only become more, I think, critical in this day and age, right? And the third part is, I think, the ability to change and stay curious, right? And that’s not just limited to associates or you or me, but I think to even to companies and organizations, right? Uh, I, I keep giving this example of you know, e-commerce, right? 

[00:38:21] Shashank Kadetotad: Very technology-driven paradigm no one saw coming and then, you know, and that is where Amazon, you know, really excelled and even today it is a behemoth, right? And, uh, a lot of the other retailers were caught off guard. So I think we are in that phase of technology where, you know, this is going to create industries that we don’t know, roles that we can’t see now, right? 

[00:38:46] Shashank Kadetotad: Uh, who knows what the future is gonna look like. I can’t tell you sitting now even two years later what it’s gonna look like, right? But I do know that people who are curious who are diligent and who are adaptable will always continue to do well, right? Uh, so I think that the, the ability to be able to do that, to learn new things and to apply that will never go away. 

[00:39:09] Shashank Kadetotad: So I know these three can seem rather generic, uh, but I can’t stress that, uh, like every single person, not only I hire, I would imagine a lot of leaders who hire, uh, will focus on these th- things. Because, you know, your technology strength, uh, I think, uh, will is critical but, uh, you know, ca- we’re reaching a point real quick where, you know, it can be done a lot quicker, uh, it can be done faster, uh, right? 

[00:39:37] Shashank Kadetotad: And AI is only gonna get more sophisticated. Yeah.  

[00:39:40] Tessa Burg: Yeah. No. I agree, and I think if you focus in college on studying any type of technical skill, like whether that’s engineering, data science I strongly believe everyone should take at least data science courses-  

[00:39:55] Shashank Kadetotad: Right  

[00:39:56] Tessa Burg: … because when you understand the foundation, then when you pair that with critical thinking, you get to what the solution needs to look like, how the process needs to evolve, like way faster. 

[00:40:08] Shashank Kadetotad: Yeah  

[00:40:11] Tessa Burg: I’m excited for the entry-level positions to change too, and I think that’s something else kids need to just be open to. Like, the job you thought you were gonna have is ev- is evolving, and like we’re– like you said earlier, we’re evolving. We’re having to evolve. But if you’re able to take your technical ability and pair it with that m- with critical thinking, with being curious, and show that, then you’re gonna be set up for success no matter what you do. 

[00:40:36] Shashank Kadetotad: Oh, yeah. Absolutely. And, you know, I, I just want to also be clear, right? Uh, s- technical skills are never going to go out of fashion, right? Because I think a lot of the good technical people you met, or you meet, are people who… You know, because technology forces you to think critically, right? So ultimately they are really good critical thinkers, good technologists are. 

[00:40:59] Shashank Kadetotad: Right? A- and to your point, I think, uh, one thing I would say is you know, this is– I, I don’t know if it’s a trend or, you know, if it’s gonna change, but we see we see a lot of people going back to hiring, you know, for traditional data science positions, for example, because now AI is still pretty expensive, right? 

[00:41:18] Shashank Kadetotad: Uh, iterations are still pretty expensive, and, you know, there are bottlenecks in energy and cost and all of that stuff, right? So I… While I think there was a period where everyone’s like, “Hey, no data scientists required anymore,” like, I ca- I remember thinking- Yeah … “Really?” And then, like, you know, I used to see all these posts on my LinkedIn, “Hey,” they are like, “I can wipe code and a model,” blah, blah, blah. 

[00:41:41] Shashank Kadetotad: And then three weeks later, you know, you have the likes of these big technology firms who are now hiring data scientists again, you know, because they’re like, “Hey, it’s so expensive to, you know, just rely on AI to build something.” So I think-  

[00:41:55] Tessa Burg: Right  

[00:41:55] Shashank Kadetotad: … yeah, so that will really never go out of fashion. Like, systemic and technical understanding, for sure. 

[00:42:01] Shashank Kadetotad: Yeah.  

[00:42:01] Tessa Burg: Yeah. I think that’s a great advice to leave the podcast. That’s what we’re gonna end with. Thank you so much for being our guest, and if the listeners wanna reach out or have any questions, where can they find you?  

[00:42:15] Shashank Kadetotad: Oh, they can, uh, they can always reach out to me on LinkedIn. Right? Uh, it’s, uh, uh, my name is Shashank Adetote, and I can assure you it’s a pretty unique surname, so it’s- Yeah 

[00:42:27] Shashank Kadetotad: pretty easy to find. Yeah.  

[00:42:28] Tessa Burg: Yeah, and if you need the spelling, have no fear. You can find this episode on modop.com, M-O-D-O-P.com/podcast, or just search Leader Generation wherever you listen to podcasts. That can be on Spotify, Apple Podcasts. We are on all the platforms. And until next time, thanks again, and I’m really looking forward to doing a follow-up where maybe we can explore, like, how this advice really came to life, not only at Mars Snacking, but for other listeners as well. 

[00:43:01] Shashank Kadetotad: Thank you so much, Tessa. This, uh, conversation was a pleasure. Thank you.