UiPath Fusion 2025 | Lalitha Kompella, Capgemini & Eric Lim, TetraPak

Clips
More from UiPath Fusion 2025

Lalitha Kompella

VP & Global Head, Intelligent Automation, Business Services

Capgemini

Eric Lim

Head of Hyper Automation CoE

TetraPak

play_circle_outline Introduction of Lalitha Kompella and Eric Lim at UiPath Fusion conference in Las Vegas.

play_circle_outline Overview of Capgemini's intelligent automation practice and its mission to enhance human potential.

play_circle_outline Enhancing Automation Success: The Critical Role of Collaboration Among Technology, Processes, and People with Real-World Agentic AI Solutions

play_circle_outline Critical factors for successful agentic AI adoption include choosing the right processes and data.

Info
Transcript

Lalitha Kompella, Capgemini & Eric Lim, TetraPak

Lalitha Kompella

VP & Global Head, Intelligent Automation, Business Services Capgemini

Eric Lim

Head of Hyper Automation CoE TetraPak

In this UiPath Fusion 2025 segment, theCUBE’s Dave Vellante and Rebecca Knight sit down with Lalitha Kompella, VP & Global Head, Intelligent Automation Business Services at Capgemini, and Eric Lim, Head of HyperAutomation CoE at Tetra Pak, to unpack how organizations are evolving from traditional RPA to agentic automation. Kompella shares Capgemini’s automation philosophy (using technology to unleash human potential) and explains why success starts with picking the right processes, building a strong data foundation and orchestrating people, bots and AI agents... Read more

explore Keep Exploring

What is the context of the live coverage of UiPath Fusion in Las Vegas and who are the guests being introduced? add

What role does the speaker hold at Capgemini, and what is the company's belief regarding technology and human potential? add

What role do humans and technology play in the ecosystem of agentic AI, and can you provide examples of how it can solve real-world problems? add

What considerations should be taken into account when choosing processes for automation in organizations? add

bolt Powered by CUBE AI

Lalitha Kompella, Capgemini & Eric Lim, TetraPak

search

Rebecca Knight

>> Good afternoon, everyone, and welcome back to theCUBE's live coverage of UiPath Fusion here in Las Vegas. I'm your host, Rebecca Knight along with Dave Vellante, my co-host and analyst. I would like to welcome Lalitha Kompella, VP and global head, Intelligent Automation Business Services at Capgemini. Welcome.

Lalitha Kompella

>> Thank you, Rebecca.

Rebecca Knight

>> And Eric Lim, head of HyperAutomation CoE at Tetra Pak. Welcome, Eric.

Eric Lim

>> Thank you for having me here today.

Rebecca Knight

>> So I want to start the interview with having you both talk a little bit about your firms and your roles. Obviously Capgemini needs little introduction, 350,000 employees across 40 countries. Tell us a little bit more about it.

Lalitha Kompella

>> Absolutely. So I run the intelligent automation practice for Capgemini, which is, like you said, more than 360,000 employee base. And so it's interesting for you to know that Capgemini is also my customer, not only my employer in my current role as automation data. And so what this large organization means, we have several segments of transformation and our fundamental belief is that technology should be applied to unleash the ultimate potential of humans. So that's the purpose with which we drive, and that's the same purpose of automation practice that I run.

Rebecca Knight

>> So we're going to dig into that.

Dave Vellante

>> I love that.

Rebecca Knight

>> For sure.

Dave Vellante

>> Reminds me of Michael Dell's mission.

Rebecca Knight

>> Exactly.

Dave Vellante

>> And look at him-

Rebecca Knight

>> His vision, exactly.

Dave Vellante

>> Works for him.

Rebecca Knight

>> Eric, tell us a little bit about your role in Tetra Pak.

Eric Lim

>> So I'm the head of HyperAutomation CoE in Tetra Pak. Tetra Pak, we are actually present in over more than 160 countries. We are one of the leaders in the packaging and processing services globally. Actually we are global, but we have a local presence. And in my role, what I actually do is to try to enable our internal units as well as our customers to be able to actually achieve business outcomes that are excellent.

Dave Vellante

>> Okay. So let's get into it. We're going talk about agentic here. So let's talk about agentic. What does it mean to you? You ask multiple people what a agentic is. You will get somewhat different answers, but what does it mean to you and what does you think it means to your customers and the industries that they participate in?

Lalitha Kompella

>> Yeah. It's a very interesting question. Everybody wants to talk about agentic AI in every UiPath or any other conference. I think definitely it's hyped about a lot more than it should be, but the technology from my understanding and my experimentation learning perspective has the potential to finally bring a solution that can reliably replace the work of a full-fledged individual who is doing that job manually, provided we wire it correctly. But it may be long way to go there, but the technology promise means first of all, end-to-end process automation experience, end-to-end seamless coordination between organizational processes, people, clients, and employees. And it's possible because of the ability of memory, perception, reasoning learning, which gives the ability to act and be decisive, which was not there in traditional automation technologies before. So here is what agentic AI means to me and to my customers that we can take them not just about the current level of automation, which makes far greater benefits.

Dave Vellante

>> Eric, you have hyper automation in your title, which is so interesting. And so what does this mean in manufacturing? Does it get us into things like robotics? I mean, we think about that, but let's start with what agentic means to you in manufacturing.

Eric Lim

>> That is a very good question. Actually, in Tetra Pak we're having automation initiatives for many years. So we have been working together with UiPath and with as well Capgemini for many years putting in place all the different automation initiatives. I think what is interesting now is that as we move from computer vision to AI and now agentic AI, we are evolving with the people and also with the processes. I think what is important is to understand that it's an ecosystem where we have to work together all together. Y

Rebecca Knight

>> And you make a great point that it's not just the technology, it's not just the processes and the tasks, it's the people too. And you need to bring everyone along. Eric, can you talk a little bit about some real world examples and problems that you believe that agentic AI could be the solution for?

Eric Lim

>> That's another good question. Actually, in current world, the human resource, the people are the very most important parts. The robots and agents are actually strategic enablers to help us deliver the outcomes. So what we would like to do is to enable our people to do more higher value works and to be able to do that agentic AI introduction of AI agents help to reduce the time needed to perform some of the tasks that either can be done more efficiently or can be more productive. So this enable us to actually work together with the robots that we have already in place. With the AI agents, for example, in some of the places, like for example in warehousing our service supply network, we have actually reduced the number of hours required from days to actually hours. For me, that is pretty amazing.

Rebecca Knight

>> And you said with that time saved, they can focus on more higher value. What are some of the real world examples of the higher value tasks that they can focus on?

Eric Lim

>> So one of the real life examples is that we have enabled some of our employees to transit to different roles. So for example, one of our key business experts was working previously in a warehouse environment performing more of the tasks in the warehouse itself. Now he has moved to become an analyst. He's doing an analyst, working on analysis of the different type of tasks that's automated. So for me that is a transition to higher value work and he has become also more efficient, effective for our organization.

Dave Vellante

>> Lalitha, yesterday in the keynote, Daniel Dines, of course they were talking about the MIT study. Everybody is talking about that. He said the major... He gave four reasons why agentic was failing to meet expectations. He said one is they don't have a strong automation foundation. He said the second, which was very interesting, I think was kind of controversial, the ambitions are not high enough. You're not going for big enough winds. And then he said the third is security, fear of rogue agents. And he said a lot of hype in the market, a lot of confusion. So two-part question. Do you agree with most of those? I won't hold you to number two because the CFOs are going to say, "Well, hold on. Let's not take so much risk."

Lalitha Kompella

>> A little risky.

Dave Vellante

>> But I still like his ambition. He shoots for the moon. But what are you finding are the critical factors that are allowing clients to succeed in Agentic?

Lalitha Kompella

>> So we have to see enough success solutions still really remark on this, comment on this, but I can tell you what's working in the early stages of adoption. Picking the right process and not agent washing everything, process for courses. You don't want to take a bot's job and give it to an AI agent. At the same time, you don't want to apply an AI agent for a very complex problem, which actually requires human intelligence. But traditional automation, people are so pampered by technologies like robotic process automation. They really like to think bottom up. They really like to just put the next big flying thing into their automation solution. That's what is causing this problem, right? But what's working is when you pick the right use cases, the right processes, which will benefit from this technology. And when you put the right data foundation, which is the biggest issue in customers organizations, nobody prepared for a technology today achieve this level of an ambition. But unless they prepare the data to be consumed by AI agents properly and their infrastructure is ready, and the people are ready to accept it, it's not going to fly. I agree with everything he said, but there's more to it at the ground level. These are the points that I'm observing. I just want to add to one more example.

Dave Vellante

>> Please, yeah.

Lalitha Kompella

>> We did HR process transformation. I presented that yesterday in one of the forums. The one thing that really worked our way is picking a process which has not reached its final potential because of robotic process automation, but it could take a big jump because of agentic automation. And number two, we built that data foundation in parallel so that this solution can be affected. Today, we are able to answer the million queries of 360,000 employees of Capgemini because of these three or four success factors that we have incorporated into the problem-solving.

Rebecca Knight

>> So it really comes down to, as you say, picking the right use case.

Lalitha Kompella

>> Absolutely.

Rebecca Knight

>> So are there any rules around that? Because I mean I think that there is this, on this one hand, you need to be ambitious, you need to think big, and yet it also needs to be narrow enough to really be able to know probability, this is what the impact will be.

Lalitha Kompella

>> Absolutely, yeah.

Rebecca Knight

>> We solve this.

Lalitha Kompella

>> The picking the right process has been a historic challenge. People have made mistakes even with robotic process automation and especially in the world of AI, picking a process which is somewhat less deterministic, which means not all parts of the problem are known. There is some decision making involved is a first step, but do not go to the other extreme of completely probabilistic analytics-oriented, research-oriented processes, which only can dwell on human intelligence. The balance is right in the middle. That's how the processes have to be picked. And it's always either RPA or agent decay. It's not A versus B. It is really not about which tool wins, it's about what is a mix of these tools that can deliver the seamless end-to-end automation problem. These two mindset changes will really help a lot.

Dave Vellante

>> How do those apply to manufacturing?

Eric Lim

>> Yes.

Dave Vellante

>> What's consistent? What's maybe nuanced in manufacturing?

Eric Lim

>> So maybe let me share a little bit about our journey. We started off with finance and techs actually. It was actually the support of our leadership who had the vision, who wanted to make sure that we have the chance to change and together we're a team of a fearless individuals who are willing to go embark on digital journey. That's how we started. We started, we tried with POVs, proof of values with POCs. We tried with both smaller initiatives that has impact as far as bigger ones. So of course we learned along the way and we actually enhanced approach. So after we started with that, we went on to other different business units like services. We went on other business units like sustainability. We have started on that area as well. And a lot of different areas, for example, in packaging solutions. So one secret sauce, I would call it a key ingredient is to make sure you have a very reliable implementation partner. That's why I'm here with Lalitha here today besides UiPath tools that we have.

Lalitha Kompella

>> Thank you.

Eric Lim

>> So I think it's important to make sure that it's a change journey. So the tools and the changes you're making are able to stick and also make sure that they have the impact in the industry. So because some people think of manufacturing as something which is repeatable, that's maybe not much room for change. That might be a wrong perception actually. There's a lot of opportunities for change. Make sure you pick the initiative that has a lot of impact. I think that is very important. And secondly, strong leadership. And thirdly, a very good team that consists of implementation partner as well.

Lalitha Kompella

>> In fact, he said something very important. The processes in their current state are not directly suitable for this advanced technology. The processes need redesign to be adaptive and to be AI native design from the get-go because then we can fully take advantage of this powerful technology, which is what Daniel meant when he said that they are trying something too small, too little or trying something too complex.

Dave Vellante

>> And he also talked about his first point was not a strong automation foundation. And that's something that you understand well, because you've been doing this for a while. You've got experience with previous automations in RPA. And so my question is what's the dovetailing between the traditional RPA, traditional automation and now this new agentic? Where do they complement each other?

Lalitha Kompella

>> Yeah. So actually every agentic automation is... But a system of agents, it's a compilation of different types of agents. Just like a team is a mix of different persona roles and responsibilities. The trifecta is defined by AI agents, RPA agents, I love to call them agents. They're doing very reliable job today and human agents. Without the mix of these three, at least for now, in the early phases of adoption, agentic AI will not be successful. They really work with each other very well. And the whole problem comes together with repeatable rule-based work being done by RPA agents, intelligent and more orchestrating work done by the more intelligent agents, AI agents and humans in the loop telling them what to do, controlling them and managing the exceptions that fall out.

Dave Vellante

>> The first time I've heard that RPA agents, I like that because I think people sometimes put RPA in a... They denigrate it a little bit, but they're trusted doers. Right?

Lalitha Kompella

>> Yeah, they are.

Dave Vellante

>> They're deterministic. It works, right?

Rebecca Knight

>> And as you say, trusted I think is a really critical part of that is that their colleagues already know that they are going to deliver what they say they're going to deliver. And that trust is so critical-

Lalitha Kompella

>> Critical, absolutely....

Rebecca Knight

>> as we go forward

Lalitha Kompella

>> And then a solution, if 70% is not handled by these trusted RPA agents and it goes all the way to AI agents, I will be worried. Right?

Dave Vellante

>> There definitely are human analogs too, right?

Lalitha Kompella

>> Yes.

Dave Vellante

>> When you think about it. I mean, it's like some people in the organization, they can do this. If you ask a fish to climb a tree, it's not going to be successful.

Lalitha Kompella

>> No. And even human evolution, I think we weren't this intelligent like thousands of years ago and now we were evolved into very intelligent people, hopefully. And so same is the case with these digital workers.

Rebecca Knight

>> So describe some of the most powerful wins that you've already seen. I mean, Eric has already talked about the saved time, that has enabled workers to focus on more higher value tasks, but what are some of the things that you're also seeing in terms of downtime and yield and cycle time?

Lalitha Kompella

>> As benefits, you mean?

Rebecca Knight

>> Yeah. Because you are a customer as well.

Lalitha Kompella

>> Absolutely, I am. I think definitely the level of autonomous or straight-through processing has drastically increased. So for example, in the employee queries or self-service, right? What used to take probably minutes or sometimes even for a more complex requirement, hours has come down to a few seconds now. We have seen 80 to 90% reduction in the average handling time. So that automatically boils down to productivity improvement and you don't need as many people to deliver the same work. But you had one interesting question in the previous point. What would those people focus on if their rot work is given away to these machines? Take the case of HR. Today HR needs quality time to be spent with employees understanding their career goals, how to reskill them. In the backdrop of AI agents taking over their work. Do they have the time? Do they have the techniques? Do they want to just simply sit and answer the millions of questions coming from employees? What is the real human employer-employee relationship? That's where all these machines are coming to play. That's the real benefit.

Rebecca Knight

>> Because I would also think with HR, yes, there's a lot of tedious, boring tasks that HR representatives do, but also being that human touch point, being that person who is, "I'm here to hear what are your dreams and aspirations and how can I help you? What can the organization do to support you?" That's also a critical component-

Lalitha Kompella

>> That's a critical component....

Rebecca Knight

>> of human to human. So we're going to need that.

Lalitha Kompella

>> We are going to want to reverse that ratio of how much time they spend in operations versus how much time they spend in employee relationships, employee development, and this applies to every industry. So that way potentially there is no industry which cannot be disrupted by this technology if wired well.

Rebecca Knight

>> We had Craig Le Clair on yesterday and he's written a new book called Random Acts of Automation. He's giving a talk today too. How do you think about that and making sure that you're not just implementing AI for AI's sake and saying, "Oh, we can do this." So why not? How do you think about that thoughtfully and empathetically in terms of your workforce?

Lalitha Kompella

>> Look, I'm a technologist. I have seen at least three to four revolutions of technology in the world. This is my 34th year in the industry. So your Brad Pitt, so that doesn't impress me. So the impression will come only when they actually make the impact. So in my practice, we have very well-defined rules and the expertise to decide which process to pick and do not implement agentic AI or AI agents just for the sake of it. The whole industry calls it the agent washing. And I also can clearly predict when is the hype wave and when it is going to taper and become one more advanced technology in the full stack of solutions that we offer. As a technologist, I have the balanced view, but we also need to promote the growth of greater advanced automation solutions for our customers. So we'll always sell the tipping point solution and not the traditional solution. That's just a way of growth and it doesn't mean that that is all we want to do.

Dave Vellante

>> I think you're right. You had a comment, Eric?

Eric Lim

>> I had, because I was thinking about the excellent question you asked. In Tetra Pak, we want to protect what's good, so we begin with the end in mind. We want to be able to introduce all the automation with the AI initiative that helps our business and our customers. That is important. So when you begin with the end in mind, you make sure you have that objective in mind. You'll be able to ensure you put in the right initiatives in place.

Dave Vellante

>> And the point about agent washing is so right on because I always ask, "Okay. Well, let's talk about the data. How do you harmonize the data? What's the role of the BI system? How do you govern these? What's the agent control framework or what these guys call orchestration?" And then a lot of times you get the answer, "Well, we're working on that."

Lalitha Kompella

>> And the last but not the least, if I launch agentic solution tomorrow to my customer or to my internal customer, are the employees ready to work alongside it? Do they have the skillset and the mindset to know how to complement it?

Dave Vellante

>> Right. So the technology is evolving, but then the human aspect of it from adoption standpoint is still... And that's why I've always said, "This is going to take the better part of a decade to play out."

Lalitha Kompella

>> Easily.

Dave Vellante

>> It's not going to happen overnight. It won't be 2025 is the year of agentic marketing.

Rebecca Knight

>> Well, it's where Lalitha started this entire conversation.

Dave Vellante

>> Marketing should lead the adoption. I think these guys actually are so focused on orchestration. That's good.

Lalitha Kompella

>> That's right.

Dave Vellante

>> Go nail that slice. Beat Switzerland.

Lalitha Kompella

>> I love that, yeah.

Dave Vellante

>> Let somebody else deal with the data layer and so forth.

Lalitha Kompella

>> No, I agree with you. The most important element is to orchestrate the end-to-end process. The biggest reach from RPA to this is, RPA needs to do bottom-up assessment of opportunities. This one is a hundred percent top down end-to-end process.

Rebecca Knight

>> Mic drop. I love it. That's a perfect note to end on. Lalitha and Eric, thank you both so much for coming on theCUBE.

Eric Lim

>> Thank you.

Dave Vellante

>> Thanks, guys.

Eric Lim

>> A really fascinating conversation.

Lalitha Kompella

>> Thank you for having us.

Eric Lim

>> Thank you.

Dave Vellante

>> You bet.

Lalitha Kompella

>> Great questions.

Rebecca Knight

>> I'm Rebecca Knight for Dave Vellante. Stay tuned for more of theCUBE's live coverage of UiPath Fusion. You're watching theCUBE, the leader in enterprise tech news and analysis.