S3E32 - Mari Shankar, CTO of Telarus

Speaker A: Welcome to the Keeping It Real podcast. This podcast features honest and transparent conversations about what it takes to manage corporate technology and expenses today. Each episode listen to true and personal stories from IT professionals who are working hard to make sure their business stays connected and competitive. I'm, um, your host, Jethro Castillo and let's meet our next guest. Today I am joined by Mari Shankar, Chief technology officer at Telaris, a, uh, leading technology services distributor. Mari brings deep expertise in delivering complex solutions, especially around Salesforce infrastructure and service delivery, and excels at aligning technology with business strategy. Mari, thanks for being on the podcast.

Speaker B: Thanks Giotto. I'm excited to be here. Um, really looking forward to our conversation today, so thank you for having me.

Speaker A: How did you get started in the world of IT and tech and how did that love sort of grow and, and become what it is now?

Speaker B: Um, I wish I had a great story like what interested me, but it was basically like, you know, when I, when I was looking at my engineering courses that I wanted to pick, computer science was brand new in India. I graduated in India and computer science was a brand new course, you know, engineering course that was offered by Fuel Institute. And I was in one of those institutes. So it felt like something that different that the institute was offering. So I was like, let me explore. Typically we go for like mechanical, uh, engineering or civil engineering was the trend when I was graduating long time ago. Um, so computer science was brand new. They were experimenting with different things at the university and I was like, this is a good way to expose myself to something different. Um, not with any graduation or anything. It's just was like interesting to explore that. So I started exploring that and that's how I got started in the engineering field. Um, what I learned and what I do are like a little bit of a more, um, things have changed because learning is very theoretical, very, you know, you learn languages, you learn how a computer works, Microsoft works, all of those things. But I don't deal with any of those things. Right. I don't Google with zeros and ones. So that's how I got interested. And when I graduated I started as a consultant developer. So basically a developer work for consulting company, mostly working with Microsoft technologies. So I have built websites, business applications for bunch of uh, companies that our consulting firm usually partnered with. It was the early web era, HTML, ASP, you know that, that uh, era, uh, that was 98, 99, 2000. And everything was moving, it was moving fast. New things were being all developed and rolled out. So it was actually changing how the business Worked. So that's how I got started on it.

Speaker A: I started my life when I was 98. So, you know, same same path. Me and computer technology, I guess.

Speaker B: Yeah.

Speaker A: In your role right now, specifically, what's kind of a big challenge that you're facing and how are you choosing to tackle it?

Speaker B: People look at all the innovation happening. We all love to talk about AI, cloud, automation, but if you think about it, none of them work without clean data, right? For me it's mastering the data, uh, ecosystem is very important and that has been the hardest challenge that I see. Not just with them, um, you know, not just right now, but in pretty much in a long time in my career after my development days when I was consulting, I go to organizations, advise them on what they should do and all of those things. You know, this is how it should implement a contact center. This is what you need to do. Your um, policy admin system should be like this and how your contact center should be set up and what tools you should use and help them implement it. But what I found out was the success or the expected outcomes happens only then they have a really good data. So getting the data right is the hardest part and that is the challenge that many of us are running. The issue is that data lives typically everywhere, right? It lives in your systems, most probably multiple systems, not just one system. Then it lives in spreadsheets. Uh, you must have something like Google Drive, SharePoint something running it, right? Then you have vendor, platform that you're using and all of those things are open. So before you can do any of these, use these tools or things that are being innovated right now is that you need to get this data in shape and that takes a, uh, long effort. People discount how much effort and how much it is important, how foundational it is for their success with this, even solving a business problem. Um, so that's the biggest challenge I run into. You can't measure the success accurately or you can't prove a success point or ROI on it. Project is because of data. It's not about the technology, it's not about the tools, it's not about the automation. All those things, all those are great. But if you have a consistent, clean connector, reliable data, then everything falls into place. Um, so that's, that's how I feel and for me it's not about adopting the next best tool in the market. It's making sure you have the right data, uh, related, reliable, clean data so that the tools can be applied on top of it and make it effective.

Speaker A: I Always come back to this really easy example as a non, uh, fully tech person where it's like you have a dirty house, it's cluttered, it's messy, it's disgusting, it's gross. Right? The value of the house, not like objectively, numerically, financially is low because it's dirty, it's a mess. And you're like, oh, actually I know how to solve it. Buy more things, more nice things to raise the value. Not realizing that the reason the value is so low or why it's hard to understand why you can't increase the value is because it's so messy in the first place. And the thing is, is that we think that moving backwards or slowing down or stopping is bad for progress. Right? But I, there's uh, a really obvious adage which is like if you're taking three steps in the wrong direction, it is more efficient to take three steps backwards and then take three steps in the right direction than it is to continue taking three steps in the wrong direction. You are, you are, you're going too fast in the wrong direction, which is worse than going backwards and then going forward in the right direction. And so yes, in my opinion, for that house, you take the time to clean the house, even though you're not moving forward and getting more things for the house, like the newest fridge or like the Roomba or like a smart lighting, uh, system that's not going to clean the mess of your house. Then when you finally clean the house, you'll realize how much you already have that can solve the problems that you think your house has. And then you can also get rid of the things that are actually wasteful for your house. So then you save more money because you're not burning energy. And then you don't need to buy more things because you realize you have the solution in the first place. And then after all of that is done, you find out where your gaps actually are and then you can find the right item to solve the solution of your house. Right? That's easy to understand, right? That's the same thing in tech and it in.

Speaker B: That is so true.

Speaker A: Right.

Speaker B: My favorite example is you buy a Ferrari and then you start putting, you know, the real quantity petrol in there. It's not going to give you the performance that you want right now. Uh, in your deep petrol, it, it's not going to crude oil. Whatever you put in, it's not going to give you the performance you're looking for. It's going to get stuck somewhere. M. That's how it is. Right. You get all these things, cool tools built or cool applications built, but then the data coming into it is not that great. Then anybody who's using it is not going to have great experience. Right. So I think data is the foundation. Data is the foundation for all projects, all things that an organization does. It helps us when the data is right. Right. Everything just happens together very well. Like you make smarter decisions because you can trust your data. Otherwise you're doing the, you're looking at it and saying, yeah, I think this is current, but I don't know if the data is correct. Right. So it's, you go back and forth and uh, we actually have done a, you know, that was one of the reasons why I joined Teller Assist or Revision Orca, create a single source of truth. And we have done that and we have been on that journey for uh, more than a year and we got that now all under control and it's single source of truth. Now it's, now it's about how do I get better ingesting data, um, and get, start capturing better data. So that's a project that I'm working

Speaker A: on currently and we're going down the data train now. If data is foundational, why is it not? Because you mentioned, and I agree that we see this issue across many, many, many organizations. If we think that this is common sense and we believe that data is foundational, why is it not a priority for many organizations to get clean data? Where are. And it's not just like, oh, they're not thinking about it correctly. They're not understanding. There are probably many different routes, uh, or different reasons. Right. But why is it not a focus of every single organization to clean up the data? Especially with AI, generative AI being so big too and it taking in data. Why is this idea of cleaning data not a priority?

Speaker B: It is hard to do. Not that easy. It's very hard to do. Um, even with generative AI, yes, that is helping us a lot in a lot of ways to improve our data quality. But it is hard to do. It's just uh, the history of the data, how good it is, you know, just being able to figure out what is wrong with the data taste. And you know, you also mentioned about the shiny new tools being there. It's always nice to have some of this shiny new tools too. Right. If I go to the M, you know, for example, if I say that I'm working on data for the next three years in my organization, that means we are not able to deliver tools or think, you know, uh, Capabilities that they want to run there. So I think there's a balance you want to strike. You don't want to go full fledged saying that I'm going to focus on data for next two years or three years because it's not a, uh, done and said and done through. Right. You are going to do it. It's the hardest thing you will do. Even if you put in all the guardrails, there's always going to be this edge, cases of things that are going to go out of, make it wacky and you got to keep on working on. So it's a consistent, it's like you know, keeping you, you used a good example. If the house is dirty, you cleaned it up. Okay, fine, you got, you understood its data, you cleaned it up. What means that you have to keep it clean regular. You just can't just do it once and say I'm done now, I'm done for the next 10 years is not how it works. Right. And that is, that is a lot of effort and people do want, um, the standard, um, it just, it's an ongoing kind of like a battle that you have to keep fighting, keep, keep on top of it. Right. It's always needs to stay the focus has to be there. Some level of effort has to be invested in it to make it better and better.

Speaker A: I appreciate you saying that because, uh, I think it challenges my, I, uh, guess all or nothing thinking, which is like, oh, let's just spend all the time fixing the data and then we can do the stuff on top of that. But it's like, well, but if you are already a big business, a robust business, a growing business, you also need to keep up with the times to be able to provide value to your stakeholders and your customers. Uh, so you can't, you kind of have, you can't be perfectly, you can't perfectly stop everything to focus on this because then you stop providing value. And then you can't even do that in the first place because then your company just doesn't make enough money. And so you need to have this balance and you actually do need to be a little inefficient. You have to deal with dirty, you have to work with dirty data with new tech while cleaning up at the data at the same time. Uh, I can see by saying that out loud why that can sound kind of exhaust like a chore to do. Right?

Speaker B: So it is hard. I mean data maintenance, data, cleaning it up and making it reliable requires not just fixing the data that you have, right. Anything new coming in also needs to be taken care of. So you want to make sure that you are building tools around those tools so that your ingestions, how do you get the data, how do you store it, how, where do you store it? All of those things have to be thought through and that's going to be an ongoing challenge with even when you bring in new tools you want to make sure that all of these are being done together. And as you said it goes both together. Right. And it will always be there.

Speaker A: When we talk about data, what does that mean?

Speaker B: So from our perspective data, you know, when we talk, when we say data we we are an organization as you know, technology, service, distributor, uh, so we can start from coating to pricing to you know orders and commission payments and all of those things. All of these are data for us, right? Starting for a sales cycle starting with talking to a customer or talking to an advisor and then figuring out what they want and backing that from that point to making sure that we are paying the right people the right money. That's all part. So every one of these is power to buy data. Behold and these are data that sometimes we create, sometimes that is sent by our vendors, um, sometimes our suppliers send our data. Like here's what you have, here's what we service. There's a lot of ah facets to this data um, which basically powers our organization.

Speaker A: What are the factors that cause data to become messy or unclean?

Speaker B: Um typically it is usually how you in ingest the data. That's where it starts. Right? So when you start when that I consider it's a static part. How do you ingest the data? Because for us we have multiple ways we ingest it. We create. We get it from vendors, we get it from spreadsheets, we get it from um different teams in the organization creating it. We get it from PDFs. So there are so many ways we get that information and there are overlap between all of these information. Trying to figure out what is the overlap. Um and making sure that there's only one version of it exists. That itself is hard thing for us. The second step of it is like for people and people are looking at PDF and translating it into something in the system. You know typical mistakes happen. Um what I call as Apple, you might call it as green Apple and trying to figure out all of those mapping are harder to do and these are all ever all we industries and things change frequent. So I did, I had the new review and I call it start calling it Green Apple. You might decide that I'm going to go to the faecal 81 or fuge, right? I mean, you might change it tomorrow and I would not know about it. I would still be calling it because, hey, this is a green apple that Jethro is giving me information about. So it's just those kind of things are, you know, things that you work with. But that's where the data starts to come in. And that's where I think the most of the effort is to get. How do you get clean data that is reliable? Right? I mean, you want to make sure that, hey, whatever you get inside into your systems is cleaned up, ready for consumption by what, whoever needs it. Whether it's our advisors, whether it is our vendors, whether it is even our, including our internal team members, we want to make sure they have the, that's quality of the deed.

Speaker A: This is not a question that can be answered very simply, um, because it is the crux of the whole conversation we've been having. But you understand that you have dirty data and it, and it can be because it's just coming from many different places and it's coming in different ways. Uh, not just different places, but different ways. Right? It could be a PDF, it could be directly in house, it could be, uh, through email. And then you have to manually input that. Obviously the ingestion is different. How do you then look at that whole mess and go, okay, let's start to clean it up. How do you clean up dirty data?

Speaker B: There is, there is not a one answer, right? I mean, each one we are looking at it and different sources, different ways it comes on. We are trying to figure out what is the best way to consume that information. How do we automate some of these things? And this is where you were mentioning generative LLMs and AI tools come into play. It's like helping us understand what actually the intent of the data is and trying to capture that and help us clean it. We are reevaluating tools around those things and incorporating some of those into our practices online, both proactively as well as reactive. Some of it, you can do the corrections as the data comes in. Some of it is like, since it comes from multiple sources, you might have overlap, you don't know exactly that. Uh, time you want to get all the data, put it in there, then we go the reactive route, saying that, hey, let's build some level of sophistication in there so that it can recognize these on a regular basis and tell the informer saying that, hey, you have these data that is not good, or being able to Flag it to somebody who can look into it and say, you need to go look at, take a look at the data and say, why is it not we don't have full information, we have duplicate information or a potential duplicate, whatever that. So we, we focus on both the ingestion and the after ingestion also. But it's, it's an evolving one, right? Different sources, different ways. You will have to tackle it. Not a one solution that I can put in place. I wish it was that easy. I, uh, really would hope so. I would be, you know, so, so happy if that was the case. But that's not the case. And it's just, you work through that multiple streams, multiple solutions.

Speaker A: Let me try to simplify it like this, based off of what I'm hearing from you. And then you just let me know where I'm missing or if it sounds good. The solution. I'm simplifying the solution of cleaning data. What the goal is, is to take all of these different data pieces from different areas and put them into one consolidated place, first of all, so that everyone can. So that the data is not siloed and separate, but all in one place. Now that addresses the idea of storage and where it is and its visibility. But we then need to also address its authenticity and its integrity, which means we have to check for duplicates, we have to check for accuracy, we have to check for all of those things and so get it all into one place, see it for what it is, believe that you have all of the data there. Right? That's another issue. If the data is hidden or somebody's not implementing the data. But then you need to make sure that the data is good and trustworthy and reliable. So then you need to do the things to be able to clean it up and make it consistent. Which means like if it's green apple vs Fuji vs red apple vs a more specific Apple, uh, you need to ensure that everyone is on the same page and that it is standardized so that people know what is the methodology to m. To map those things. And so then there's that. The second piece, the after ingestion, right? Is it actually good data? And then once it goes from all of these different places to one place and then it's clean and it's good, then it goes out to its relevant places to say, this is the data you need to do your job or make your decision easier. Yada yada yada. Right? That's kind of how it ideally would flow. Perhaps challenge me or affirm me.

Speaker B: No, that's absolutely right. To an extent. Uh, when you say about a single location or single storage mechanism, I say differently. I say single source truth, right? It can be in multiple databases or multiple systems, but, you know, always there is one and only source of truth for that particular data attribute. Right? And that's what I wanted. So, yes, you want the data to come into it. You ingest it from multiple sources, you take it to where, set of locations. I don't want to say one location or one storage mechanism or one database, whatever, however you're using, whatever platform you're using. But let's say you want to ingest it from multiple sources. You want to go to a particular data platform where you can do some of those cleansing activities before that it is being consumed by others. And that could be one platform, multiple platform. It could be database, it could be data warehouse, it could be any of those things. But it's, you know, depending on the source, depending on your applications, you have to figure it out. But the goal is to have. If we call something as, uh, booking, you want to call it booking consistently across the board. And the definition is very clear. And there's only one definition, and there's only one place where I know this is so simple.

Speaker A: Yeah, yeah, yeah. Got it, got it, got it. Mari, we're obviously running up on time, so I have two more questions for you. The first question is convince me, but I guess it's convince the audience and convince anyone who's listening why cleaning your data is important. What is the ultimate value you get from having clean data for your organization? Uh, essentially, what is the ROI when you say data?

Speaker B: Right? If we get the data right, uh, when it's consistent, connected, trusted by your users, everything just comes together along very well. The tools that you build on top of it perform better. The insights get clearer and sharper, and you can rely on it. Um, since you have a reliable data at that point and you typically, um, see or ROI on the business outcomes are real at that point in time. So data is just not one part of the organization. It is the foundational power segment in your organization, right? It powers how your business process works, it powers how your business is running, and it also powers your insights to make informed decisions for future. Hey, where do I need to focus on? Is my sales doing extremely well? But my, you know, the processing is taking longer. All of those things you find out if you have good, reliable data. When you don't have the data, you're guessing, um, I hit my sales numbers or not? Did I hit my you know, service goals, did I hit my, you know, processing times or sle's. All of those things become harder to do if you don't have the data. Right? Um, so I think data getting it consistent, clean, connected, and reliable is very important.

Speaker A: I, uh, actually think I came up with something too, a little bit. Kind of, uh, it's kind of like I woke up and I'm tired this morning. That's the problem. Okay, why am I tired? I don't know. If I keep track of that data, how much sleep am I getting? What did I eat the day before? Who did I spend time with, what were the things I did then? If I wake up and I'm tired, I can look at all of those data points and figure out what is the reason that I'm tired and then make a better decision to solve that problem the following day. Right? That's the punchline, right? It's, it's one thing to have a problem. Data is only as important as its ability to solve the problem. Right? Which means you need to have a problem for the data to act upon, and then you need to have a solution that comes from that data. And so this is why data integrity is important and why looking at data is important in the first place. So that if you have a problem, you can figure out what you need to do based off of the data to come up with the best solution you can possibly come up with.

Speaker B: A great example.

Speaker A: I got there. I got there eventually.

Speaker B: I mean, we, you can talk about it many ways, but data is critical. Um, it's like our nervous systems, you know, that powers us. Um, as you said, um, data is foundational. Everything we do in technology, one way or other, relies on it. Um, somehow you will be using it, whether it is analytics, automation, AI, or even just simpler business metrics that you're tracking and reporting on an Excel sheet. Data is important. Every decision, every insight relies on data.

Speaker A: Here's my final question. This is your soapbox. Now, what are kind of the most important lessons, pieces, uh, of advice, um, that you kind of want to share with the audience to close today.

Speaker B: So Jatra, you know, I've been in tech for a long, uh, long enough to remember when building even a app, small, ah, app or you know, took weeks you would write, usually specs, debug, test, apply. Everything was done manually. We had to type it out. I can remember the days I used to type out requirement documents and all of those things. And fast forward to today. It's incredible where the LLMs and Agent Eki, what used to take days, can now happen in hours and there are instances, some things can be done even in minutes. Right. I have watched LLMs write code, complete code blocks, generate APIs, design my database schemas. I even work with it or co create helped me co create architectures um, faster than I ever could thought it was possible. So it's just the incredible pace at what we are moving from a technology innovation is both exciting and uh, sort of scary for me. So it's not just any more autocomplete kind of a feature. Right. It's co creation. You know the LLMs and agentic AI have gone to a point where they are basically your teammate. Uh, these models understand context, reason through problems and adapt as you refine ideas. So when you layer an agent AI on top of something, um, it's like having another engineer as part of your team. So it's just the scale at how far we have come, it's said all those things. It all still depends on the data. Good data means good decisions. So the cleaner the better data, uh, the smarter view, the system speak. So what excites me most is the barrier between idea, when an idea is in my mind and showcasing it as a product has never been thinner than what it is today. Uh, it used to take me a long time to get the idea into look at paper. You know, we used to do wireframes. Then do you know to use other tools to mock it up, get feedback and then go from it. Now I can go from like idea to kind of like a sort of ready product that I can prototype and help users use it. It's incredible. It's hours, it's hours, maybe days where it would used to take weeks. Uh, I can build it instantly using here. Uh, so AI is turning literally our imagination into exhibition. This is exciting, right? Uh, it's just exciting and I think what is going to come over the next few, I say few years but then I also realized these things are coming like weekly. It's just going to be mind blowing. It's my opinion and it's awesome. It's awesome on where we are in today.

Speaker A: Well Mari, thank you so much for this conversation. I admittedly had a lot of fun. This was a very, I kind of forgot we were recording for a second. I, I, we were, we just had a normal conversation. Um, and I really appreciate um, I think your wisdom, um, your thoughts, your intentionality behind your answers but also just uh, the love and passion and excitement I can tell just exuding from you about this. Um, um, it it makes me excited. It makes me interested. You got me riled up and confused and stressed, but also happy at the same time. So just want to say thank you for an excellent conversation. Thank you so much for coming on the podcast today.

Speaker B: Thanks Shuster. This was really fun. I enjoy.

Speaker A: Well, take care.

Speaker B: You too.

Speaker A: We hope you enjoyed this Keeping It Real episode and thank you for listening. Make sure to hit subscribe so you don't miss a future episode. This podcast is all about celebrating and connecting professional professional heroes who work hard at it every day. To continue the conversation and learn how Vitecom can help you better buy, manage and pay for your technology, visit us@, uh, vitecomsolutions.com until next time.