How Agentic AI and Semantic Layers Are Transforming Enterprise Analytics

Dave Mariani: Hi everyone and welcome to another version of the AtScale Data Driven Podcast. I’m Dave Mariani, CTO and co-founder of AtScale. And today we have as guests, the HexAware team. So I’d like to introduce you to Rich Williams. Rich is the Senior VP and Head of Partnerships and Strategy. Hey Rich, thanks for joining.

Rich Williams: Thank you so much, Dave. I appreciate you having us on today.

Dave Mariani: Awesome. And then also we have rich. I’m sorry. We have Prashant Dahakar. So Prashant is the is the VP of the cloud data practice. So you’re the technical guy Prashant and you and you you also run the practice when it comes to implementation and the like. So welcome.

Prashant Dahalkar: Thank you, thank you, Dave. Thanks for having us and we’re really excited to have a chat.

Dave Mariani: Okay, so before we sort of get into sort of some of the great things that we’ve heard about and saw, so I ran into you guys at the Databricks Summit as well as Snowflake. So we just came out of a lot of activity around lots of announcements, lots of new technology, and lots of implications for our data and analytics industry. But before we get there, just tell the listeners a little bit and maybe Rich

I’ll give this to you. Tell us a little bit about what Hexaware does and sort how do you help your customers.

Rich Williams: Sure, thanks again, David. Hexaware is a large, I’d say medium-sized global technology systems integrator. We work across all different verticals, and I would say our biggest presence you would see in banking, financial services, insurance, healthcare, life sciences, that tends to be sort of the place where you see more of our work.

We’re about 35,000 people globally. We’ve got probably four or five thousand in North America, another thousand or so in Europe, and then the rest are based all around the world, but primarily based out of a couple of large delivery centers in India, one of which is where Prashant works. The company’s been around for several decades. We have grown a ton through word of mouth. I would say I call us the quiet company. We…

I’d never heard of Hexaware before. I started looking for my next career move about a year and a half ago. It’s a great company. It’s a great culture and it’s a great place to work. I think that always helps too when you’re trying to get hard things done. Everybody seems to be on the same page here at Hexaware. I’ve had a great experience. I’ve been here seven months and I’ve really enjoyed my entire time here, especially working with people like Prashant who are just can-do positive, very…

effective people. So it’s been a pleasure to work with him and the rest of our team. And yeah, it’s a it’s I’ve really enjoyed it here. But Prashant, anything you want to add to that I missed there?

Dave Mariani: Yeah, Prashant, what was your path to running services in cloud at Hexaware?

Prashant Dahalkar: Yeah, so I want to compliment what Rich mentioned. I came from Cognizant, served in Cognizant for 14 years, and then last three and a half years I worked with Exaver. I mean, a few things which I just want to mention is the revenue wise, though it’s close to 1.45 billion in revenue. But we are very nimble, take pride in calling ourselves startup with a lot of sticky clients. Most of our clients are maybe 10 plus years associated with us. So naturally,

We get into the advisory area, trying to lead them into what should be their data and AI strategy and all. So I think along with that, once we started this, the data and AI practice separately started around seven different COEs. And we’ll talk about it in the chat today. But they’re really focused on specific partners, specific tech staff. So that’s the surprise that we take.

Dave Mariani: So, okay, so let’s get into it then. So there’s a lot of announcements, a lot of things to digest over the last couple of weeks from really sort of the two behemoths in this new data and analytics platform market, the Snowflake and Databricks. They’re definitely tit for tat. They definitely are competing with each other. But Prashant, what was some of your takeaways from what you saw in some of those announcements at those shows?

Prashant Dahalkar: Yeah, sure. So I had been to Smurf like first time and I’ve heard all the announcements in Databricks. what I’ve seen is a lot of consolidation happening with the AI coming in. And when I say consolidation, it’s more to do with how do I contain the cost? Because most of the time, out of 10, maybe seven clients will always complain about the cost going up once they go into one of those platforms.

And they’re not able to figure out what is the reason for it. So two things I would say. They have brought in lot of automation around how you take care of your assets that you have deployed so that the cost management is better. And second thing is giving more power to the business user, be it quality of the data or how do you catalog your data so that you know where your data resides, get the best value out of the attributes that you store.

So a lot of self-service driven initiatives in both the cases. And lastly, maybe I want to mention that they also are diversifying in OLTP space by acquiring some of those recent acquisitions that they had. So those are the key things that I hope some.

Dave Mariani (05:49.772)
Yeah, that’s a good takeaways. it’s the old, you know, they both had that, you know, although I think Databricks sort of articulated it.

a little bit better than Snowflake when it came to why Postgres was becoming, hosted Postgres or managed Postgres was becoming part of their stack. It’s very clear that they really are thinking about using hosted Postgres as their OLTP engines. So that was sort of a big, that was a big announcement on both of their fronts. know, Rich, just from like, from the business side,

as a business user, what would some of your takeaways from that you thought would maybe be appealing to people who are trying to get ROI from these investments?

Rich Williams: Yeah, I think it’s, and even as somebody who uses this stuff myself, right, in my personal life, I think just the speed at which you can move now is just unbelievable. And the tooling, the capabilities these platforms are building inside them to allow, as Prashant was saying, allow business users to start to do, make better decisions.

Much faster than they have before and be able to interact with data in ways that they really haven’t been able to do before through just natural language. I the capabilities around this are so impressive now and it’s just come so far. I was even doing some stuff personally over the weekend, know, making some charts and this was in GPT. But even just like, I remember just six months ago when I would do the exact same thing, I had to try so many different times to get the…

chart to, if I, if I wanted to make a small change to a chart, would redo the chart, but it would completely mess up everything again. It would just kind of forget what it was just done. And today it just, every single time I asked to do something, it gets exactly the way I want it. And so I just, it’s amazing to see the progression of the technology so fast in such a short amount of time. That’s really exciting. and I think it’s really exciting for, I think business.

friendly business focused CDOs and CTOs and CIOs who want to deliver results to their business very quickly. Cause now the tools are really there to put the power into the hands of end users much in a different way than we’ve had before. And I think that’s really exciting, especially if you’re a data savvy business user, it’s a very exciting time, you know, cause these tools are getting very close to.

being able to allow the business to interact with data in just ways they’ve never been able to before. on the fly and app, it’s really amazing to see. And so I was very excited. Both companies had great shows. I enjoyed both experiences a lot. But yeah, mean, it’s sort of hard to miss that sort of general tone about just the speed at which you can do things now is, I mean, it’s unbelievable, really.

Dave Mariani: Yeah, the focus certainly was around agents and agentic AI and both Snowflake and Databricks. Databricks, I think they announced Agent Bricks and Snowflake doubled down on Cortex. So definitely core sort of investments on that front. They both also announced a semantic layer. you know, happy to hear that the industry has finally come along.

Rich Williams: Yeah.
Dave Mariani: Around to recognizing how important a semantic layer is for it really enabling this new this new motion of a gentic AI because like you said it rich it’s like people do want to talk to their data

Prashant Dahalkar: Okay.

Rich Williams: Yeah.

Dave Mariani: But you it’s like, you can’t train an LLM on the internet and expect it to understand an enterprise’s core sort of business calculations or vernacular. So you need sort of a bridge. So Prashant, I mean, we’re sort of moving beyond sort of this AutoML sort of focus, know, so AI was sort of focused on AutoML and these platforms to sort of make

Rich Williams: Right.

Dave Mariani: the data scientists more productive by suggesting models and the like. To me that seems like to Ghana sort of like nobody talks about that anymore, right? And everybody’s really talking about agents and gen AI. So what does that mean? Like for what’s that mean for the for for for your customers out there that are looking sorta to just sort of make that transition? What are they doing to jump on this wave of gen AI?

Prashant Dahalkar: OK, so I think in between, I want to make it three generations. One is where you’re doing more pattern-based, rule-based coding and getting the insights of the data. Then we move towards more of general, co-pilot-based insights, or acceleration, would say. More than insights, it’s more acceleration. But now with agent tech, most of our clients, whether it is either

Dave Mariani: Mm-hmm.

Prashant Dahalkar: Think of it as industry specifications being created for insurance or healthcare or banking. For us in the data and AI space, when I’m talking about the tech work that we do in data engineering, every customer, even I had two calls today with different domain-based customers, both of them talking about MDM data cataloging, lineage, data modeling. They want to see how agents can be exploited in

Dave Mariani: Mm-hmm.

Prashant Dahalkar: Order to create the data model. I’m just giving an example. We can create a dimensional model from transactional data set using Copilot or JNI also, any of the models. But now they want to move ahead and see, OK, can I deploy that model? Can I validate that model? I’m talking about just the data model. I’m not talking about the semantic model, just the data model. Now the second piece of it is, how do I create agents for the business user? So a business user gets a big file.

Dave Mariani: Mm-hmm.

Prashant Dahalkar: they should not rush to the IT and say that, you fix my data in this big file and then make it usable for some experiment. So they want agents to sniff through the file to find the pattern to fix the data. So more outcome driven agents. That also helps in modularizing the work. Earlier, was like a big size of code that has to be written to do everything. Now you can just write one agent to do a specific task and get it done.

To answer your question, most of the clients are moving towards asking more use cases where the agents can be created and deployed.

Dave Mariani: Yeah, it’s really, I mean, it’s really pretty exciting when it comes to, know, somebody like, I’m sure you’ve been in business intelligence industry for quite some time and, you know, everything was sort of focused on Tableau in terms of point and click and drag and drop and.

And, you know, and sort of talking to your data was sort of an afterthought, right? It was almost like it’s just a shortcut. But it’s, this really changes everything. And I think that Databricks seem to be being the most aggressive there in their sort of demo and their investments and suggesting that there’s a new way of doing BI now that the, they’re sort of trying to think to really…

talk about business intelligence differently than what we’ve been doing for the past 20 years. So this will be a question for you, Rich, like how is that going to change how businesses do business, right? If you got a whole nother way for these people to ask questions of their data.

Rich Williams: Yeah.

Rich Williams: Yeah, I mean, I think it’s going to speed things up, right? Because you don’t have to, the ability for these tools to do the work that a technical resource had to do before, right? Previously, if you’re a business user and you understand your company’s data, you understand your business context, you have all the questions, you know all the questions to ask. You’re curious about a lot of things. There’s a lot of things you’d like to explore and understand and experiment with in your data.

The constraint was always people and skills, right? You always had to have somebody that could do the work for you. Cause you had to take your questions, translate those into some sort of a technical requirement. They’d get those mapped to some sort of a database and somebody was going to have to do some work and build a pipeline. And it took time, right? And was that, that there’s a lot of friction in that process that would, it would, would take, you know, days, weeks, even months, right? To answer sometimes simple questions. And so the speed at which you could turn interesting, like

Create Insights was affected a lot by that. And that was always a source of frustration in the business too, right? And that’s been this universal business versus IT sort of frustration that almost every company has had to face. Today, it’s very different because if you have a data savvy business user, again, the only thing is they don’t know how to code, but that’s not a problem anymore.

Dave Mariani: Mm-hmm. Mm-hmm.

Rich Williams: because if you can give them an environment with just access to the right data with like, again, the right language, the semantics and the common lexicon that everybody understands, they can really start to do, they can experiment a lot. The amount of things, the amount of questions they can answer in a single day is more than what it would have taken them months to answer before. Because the amount of time and energy you have to put into creating just one single answer.

Prashant Dahalkar: Thanks.

Rich Williams: That is going to be fascinating to see how that works. So it’s going to enable organizations to test a lot of new ideas very quickly and also get results very quickly. So you can see, is this working? It’s an army of engineers and analysts really that are at your fingertips. I think that, and they’re always available.

Dave Mariani: Yep. They’re working 24-7. Yep.

Rich Williams: And they do what they’re told. They’re always on 24-7, they don’t give you any attitude, they work on weekends. They’re amazing teammates to have. But I just think we’re at a point now where it’s really real. Just the last couple of years, we’ve all been talking about GPT and LLMs and JNI and stuff, but it feels like we’re on the cusp of really being able to use it in an enterprise at…

Dave Mariani: You

Rich Williams: for a business user, be able to literally talk to their data. Start again, you got to govern though, right? I think that’s the big thing. How do you govern things still? How do you make sure that, cause I can also see shadow.

IT spinning up again, right? And on the side and you have to make sure everything’s governed. You have to make sure everything is centralized to a certain extent, but you don’t want to have it be a constraint and you don’t want to slow things down either. So I think there’s just always that balance of like pushing the envelope on creativity and speed, but also maintaining governance and security and protecting the firm. And so there’s all, but I think the organizations that are going to really be successful, the ones who’ve walked the fine line between those two things, like they, they, they don’t, they’re not too conservative.

but they’re conservative enough, right? I think that’s an interesting problem companies are gonna have to start thinking about now to really get the value out of this stuff.

Dave Mariani: Yeah.

Dave Mariani: You know, when she said something that he said, like, you know, you don’t have to code anymore. And if you even think about even tableau, you’re still sort of coding because you have to write SQL to do your data modeling. We asked, we basically, even these, these tools are supposed to be great to help business users do analytics. You still have to have a level of technical proficiency and understand data warehouses, even have a connection to a data warehouse, which is itself a big deal. so Prashant there’s it’s, it sounds great, but there’s a lot of stuff that has to happen.

Rich Williams: Yeah. Yep.

Prashant Dahalkar: Yeah.

Dave Mariani: Before we can get to that beautiful world that Rich is talking about. So can you talk about a little bit about, for your customers, how do they really sort of realize this vision of speed? Because you need the accuracy, you need the governance, you need quality data, because the data’s bad, the LLMs is going to surface bad data. So how do you skin that cat? That’s a lot of stuff to do.

Prashant Dahalkar: Absolutely. Yeah.

Prashant Dahalkar: Yeah, I think in our mission statement for our practice, we say that we bring together potential of data and exploit the power of AI. These two are there for sure. But in between, there is a line called discipline of governance. I think that’s where everything starts. So I feel that the foundation of anything is just the data governance, nothing else. So when I say data governance, we are talking about

avoiding any pitfalls around the creation of data silos. First thing is, when you talk about fragmentation or quality or trust on the data, it’s all because everyone is doing their own bit with that same data, creating their own logics, but that logic is not captured anywhere, how that data is getting transferred. I think the whole point is how empowered your governance offices. With a lot of orgs, they do have someone like the chief governance officer for data.

but they are not so empowered because the business will say, no, I need something very urgent. I don’t care about your process and I just want to implement it. I think that has to stop when we say we are actually really data driven and governed organization. I think avoiding data sets, it’s a typical jargon that everyone uses, but how we really see doing is, I can take an example of our own PE firm for whom we are actually building our data platform. So.

Though there are various components that we have snowflake that we have data bricks there we have AWS there and also Azure. So all said and done the governance is done by a purview. Okay so Azure purview is being used for governance though there is data bricks there is snowflake and all are very strong in that area as well. But the key thing is the CDO who is actually brought in there is very clear and how the process have been implemented.

and people are really educated in following that process. think that’s where the whole, you I would say you do the data governance in Excel also, is fine, but you have to have the process which will be followed every day and every time. That’s the main thing. That’s what I think.

Dave Mariani: Yeah, there’s a lot of setup that has to be done. There is no free lunch, right? You gotta have a proper data architecture. You gotta be mature enough on your data and analytics stacks to be able to then put this technology on top of it. agents are great at automating, but your data infrastructure needs to be there. look, I obviously believe the semantic layer does a lot in terms of bringing accuracy through context as well as

Rich Williams: Right.

Dave Mariani: governance, whether you’re doing traditional BI with tools or whether you’re talking to your data or whether you have agents that are acting on that data. So, you know, what are some of the other things, other ways that you’ve seen your customers be able to sort of get to the point where they’re confident enough?

to sort of really pour gasoline on that fire, Rich. So you talk about accelerating, right? But how do you help your customers really get to that point where they have the confidence to unleash the data on a lot more people in the organization than they probably previously have?

Rich Williams: I think one part is you got to continue to experiment and test and try things, right? I think and showcase the wins and the things you’re learning. I think just having a mindset of just we’re going to constantly experiment and keep pushing forward, right? I think it’s just getting one foot in the future for sure. And all the way, because this is where we’re going. So get comfortable with that. Get comfortable with this change that’s coming. think.

companies that I’ve seen that have done well are the ones who talk about it a lot and how it’s going to change things and how it’s going to help people too. I think there’s a lot of fear around this technology still and understandably in our industry. What’s going to happen to my job? Is this going to automate me? Is this going to replace me? And I think that the companies who are out there talking about

you know, AI from a perspective of it’s an accelerant to your career. It’s something to help you do your job even better. It’s going to unlock time for you to do other things. And seeing it as a tool, I think the companies that prepare their workforce like this and start having that sort of mindset inside the organization are going to be the ones that really thrive, because they will face less resistance internally. At the end of the day, organizations are made up of people.

And you need all the people kind of looking in the same direction. And if there’s not good communication from the top, from leadership about how this is going, what our goals are, how is this going to affect you personally, your career, your job, and how it’s going to help you. think that anybody who doesn’t do that is leaving a lot of uncertainty out there. And I think that can end up being a form of creates resistance to, to change inside an organization. So I think just things like, you we were talking about AI literacy, just, just educating people. Right. I think historically.

Dave Mariani: Mm-hmm.

Rich Williams: you again, if you think about the old way, right? If you, you weren’t a technical person, if you couldn’t code, it was like a different language. Totally. It was, it actually is a different language, right? You didn’t know the language. so exactly. So you don’t know how to speak that language. It’s like me talking, trying to talk Hindi or any other language that I can’t speak. I don’t know, right? It’s impossible for me to communicate effectively, but that challenge has been really kind of, it’s been reduced a lot.

Dave Mariani: Hmm? It is. Sequel is a language. Yeah.

Rich Williams: Right? Cause now we can talk about the business problem and allow a lot of the language of technical language to be under the surface. It’s still there. It’s important, but we can communicate better in inside an organization because we don’t have these language barriers that we used to have from a technical perspective. And I think that the companies that, that focus on that and that

This tool is now it’s here to empower you, right? It is you can do things that you could not do before. You can do things now that would have taken you a decade of education to do and learn how to do. And you can do it tomorrow. I think just opening people up to the potential and the possibilities of this, I still, think there’s a big opportunity there still too, because I think a lot of people still think, it’s just a technical thing. It’s out there AI. What’s it going to do for me? I just said, even at the conferences, I was talking to people who were around the conferences, right? Not super

technical people, they don’t understand. I’ve just five minute conversation with them, kind of opens their eyes like, wow, really that’s gonna, that’s not the way I was thinking about it. I was more fearful of it, but this is more of an optimistic view. I just think that that messaging from leadership can have a very positive impact on their company’s ability to.

push forward and try new things. So yeah, I think it’s just the messaging, the training, the literacy, get everybody up to speed on what this stuff is and just keep experimenting. Try it, use it in your life and your day-to-day job. How can this make you better at your job? No matter what you do, there’s something out there that you can use to improve how you do your job and become more productive. I think those are the things that if I was a leader in an organization, those are the kinds of things I would be doing.

proactively, obviously getting all the technical skills, as money as you can and get the right environment set up. But there’s a whole bunch of organizational stuff you have to do as well to be able to get the most out of it.

Dave Mariani (25:24.513)
Yeah, the difference with this revolution for me at least is that this seems like there’s a lot of emphasis in the C-suite about adopting these technologies, right? It’s a very top-down, as opposed to like, never heard like a CIO say, we need to have a data warehouse that is elastic. It’s like, we need a cloud data warehouse that’s elastic. That didn’t come down from the top.

Rich Williams: Yeah. Yeah.

Rich Williams: Right.

Dave Mariani: But you definitely hear in the C suite is like, hey, what are we doing with this with Jenny I and LLMs? How are we using it? And we need to use that. Prashant, is that what you’re seeing in your customer base too? Is it different or is it, am I seeing something?

Rich Williams: Right.

Rich Williams: Yeah. Yeah.

Prashant Dahalkar: No, no, absolutely. In fact, I want to qualify further that within Hexavit itself, right? So we always say that we are disrupting ourselves first. And as you said, exactly, it’s coming top down. Keach, who is our CEO, he has mandated everyone in our learning goals. Now, 20 % of our goals are around learning. And everyone has to be certified on one of these stats, either ChenAI or doing something on Agent TKI.

So that’s how we disrupted ourselves. And then of course, you then create your use cases, either it is vertically aligned or specifically cutting across the verticals. So those use cases were defined. that’s how we created a library of agents which are already ready to be deployed. Now, when you asked about clients, I interacted with many clients who either asked us to provide the roadmap for how you got trained or how you’re doing the mass trainings for all your people.

The second thing is I have an example of an insurance client who, you know, they already created their own private LLM almost a year back, but there is no adoption. Hardly they see like initially started with 30 % very enthusiastic, you know, workforce going on to that and, you know, asking questions or getting the work done. Later on introduced to even less than 10%. So one is, you know, getting literate on all these things, but it has to be a continuous process.

Dave Mariani: Mm-hmm.

Prashant Dahalkar: And it also always has to come top down and that what we always see.

Dave Mariani: Yeah, it’s like there’s always a question of like, how do you improve ROI in your analytics investment. I always say, look, it’s like, are people using it? It’s like, if your usage is up and to the right, you know you’ve done something right. And that may drive your costs up and to the right, but hey, there should be more people getting value out of data if they’re asking more questions of the data. So I think that’s a really sort of good benchmark for whether you’re being successful is whether people have adopted it, like you mentioned.

Rich Williams: Yeah.

Rich Williams:
100 % agree.

Dave Mariani: Let’s close this out with a little bit of like, I’d love you guys to put your hats on, they’re future hats, and where do you see this going? So I’ll start with you, Rich. What’s next? How do you think things will be different in a few years from now? I’m not gonna say five years from now, because that’s way too long in our cycle. Couple years from now, how do you think things might be different?

Rich Williams: Yeah. Yeah.

Rich Williams: I guess, you know, I wish I could predict a future accurately. But I know, I know. I do think it’s going to be the same like the way I think about it and the way I talk to people about it is and I gave this example actually at your happy hour the other night at the the joint happy hour you guys did two weeks ago. There was a woman there who was not she was a journalist and she wasn’t from our industry and she was going to ask him the same thing. My take is

Dave Mariani: We’d all be much richer and probably not on this podcast.

Rich Williams: I feel like I gave her the example, I’m like, you know, before 2000, you really didn’t use the internet and now you use it all the time, right? And I kind of think of it the same way. It’s like, if you’re not somebody who is using AI right now, I think that’s gonna change. I think in some way you’re gonna be using it in a couple of years. I think of people in our industry, you’re gonna be using, we’re already using a lot, right? More than we probably have, but.

the average person, I think it’s going to trickle down there too. And it’s just going to be part of daily life the same way going online is. And I don’t know exactly what that’s going to look like, but I think interacting with agents and, you know, bots and things to help you in your life, I think it’s going to become ubiquitous. So I think that’s one thing that’s going to happen.

You know, I’m also very interested in other things that have implications on this whole space. You know, one thing that’s true about the AI space is it’s very hungry for energy. It consumes a lot of energy. And I’m just curious of what are the implications of all the things that happening around the world too on energy supplies? Because these things use a lot of energy. And so how will global energy impact the AI revolution?

Dave Mariani: Mm-hmm.

Rich Williams: because it’s upstream from AI. need a lot of energy and CPUs. You got to have the chips, right? You need to have the chips. So what’s the global geopolitical landscape going to look like in the next couple of years? How’s that going to affect AI? I think there’s a real question there we don’t know the answer to. That’ll be interesting how that plays out. I do think it’s going to be everywhere. I think it’s going to obviously change people’s lives in massive ways and many people good ways. I think it’s like any other major technology shift. There will be people who, you know,

to change their skill sets. have to upskill and they have to learn new things. I also think I believe there will be an explosion of like new startups because you’re unlocking really smart people who don’t have the technical skills to build stuff themselves. But now they have tools that can. And so I think you might see a lot of new really interesting stuff to come out from people who are very creative, very intelligent. They have a very cool solution for an interesting problem.

but they could never, cause you can start these things with no cost. You don’t need a million dollars of funding anymore to build something. You don’t need months of engineers time. So these are the kinds of things I see like a big picture. I think it’s really interesting. I’m really curious to see what the world’s going to look like in three to five years. you know, my job probably won’t change very much. You know, it’s like, I needed to automate emails and those kinds of things, not automate, but like how to, how to just write emails faster. Cause that’s a lot of what I do. Right. but it can’t automate me. I don’t think yet.

and that’ll be shocking if I it definitely can it’s going to make me more productive yeah yeah same same yeah

Dave Mariani: It can make you a lot more productive for sure. It does already for me. I’m addicted to these tools. It makes me better. How about you, Prashanth? What’s your takeaway?

Prashant Dahalkar: Yeah, so I think I will qualify with some tangible things like for example, let’s say a pod for data implementation is 10 people, example, 10 tools that will come down to two roles for sure in coming days. I would say it should not be far away, maybe less than a year. So 80 % of that workforce is reduced. But how do I make sure that the revenue of tech server keeps going? So that will grow.

Instead of creating 10 data lakes today, I might make 100 data lakes at that time. So that way I can see it. And apart from that, think more or less the new roles will be created because of this now. There’s a lot of responsible AI stuff going on. Now there are forums being created just to make sure that we are really taking care of those things. So AI literacy has to work. One very classic example I will quote, Dave, before we close.

We have top two legal clients also with us. So I have seen them, are doing analytics on how our entire journey. So the moment they hire someone from the legal college, how they perform on their matters, that’s being analyzed. But at the same time, in India, I’ve seen, because in India there are so many cases and the judges are less structured. So there are no judges to even work on it. Here, the problem is we can…

Dave Mariani: Mm-hmm.

Dave Mariani: Mm-hmm.

Prashant Dahalkar: automate and use AI into maybe analyzing the last cases which has happened in the past and create some, let’s say, inferences around it. But you cannot use AI to give the judgments. That has to be completely human driven. I think we’ll have to balance it out. Maybe productivity gains from this has to be more. And then you have to really responsibly use in tight areas. That’s what I see.

Dave Mariani: Yeah, I like that. That’s good. It’s like, look, the human doesn’t go away in the equation that ultimately, know, AI is the helper, but human needs to make that last final decision at the end of the point. And someday that won’t be true, but at least for our lifetime, I think, I hope it’s true.

Rich Williams: Yeah, I think so too. Absolutely.

Dave Mariani: Yeah, it’s human plus machine. All right. You guys are awesome. Thank you, Rich. Thank you, Prashant. It’s been a great discussion. And to everybody listening out there, stay data-driven in this new AI agentic world. Thanks, everybody.

Rich Williams: Thank you.

Prashant Dahalkar: Thank you.

Rich Williams: Yeah, really appreciate it.

Rich Williams: Thanks so much Dave for having us. Appreciate it.

Prashant Dahalkar: Thank you, bye bye.