What's Up with Tech?
Tech Transformation with Evan Kirstel: A podcast exploring the latest trends and innovations in the tech industry, and how businesses can leverage them for growth, diving into the world of B2B, discussing strategies, trends, and sharing insights from industry leaders!
With over three decades in telecom and IT, I've mastered the art of transforming social media into a dynamic platform for audience engagement, community building, and establishing thought leadership. My approach isn't about personal brand promotion but about delivering educational and informative content to cultivate a sustainable, long-term business presence. I am the leading content creator in areas like Enterprise AI, UCaaS, CPaaS, CCaaS, Cloud, Telecom, 5G and more!
What's Up with Tech?
Agentic AI: Transforming Global Business Services into Growth Engines
Interested in being a guest? Email us at admin@evankirstel.com
The strategic evolution of Global Business Services marks a revolutionary shift in enterprise operations. No longer just cost-cutting centers, GBS functions are becoming powerful growth engines through the integration of agentic AI and unified data platforms.
Behind this transformation lies a fundamental change in approach. While early GBS focused on moving simple tasks offshore, today's advanced operations are handling core business processes and contributing to strategic outcomes. As our expert guests reveal, this shift is enabling unprecedented efficiency – with automation rates soaring from 40% to 80% through AI implementation.
What makes this possible? The emergence of agentic AI stands at the forefront. Unlike traditional automation that follows rigid rules, agentic systems can evaluate multiple scenarios, communicate across specialized sub-agents, and make context-aware decisions. We explore real-world applications already delivering remarkable results: autonomous procurement systems handling millions in transactions, perpetual KYC processes in banking, and dramatic reductions in process turnaround times.
The data challenge remains critical. Enterprises struggle with fragmented systems and siloed information, but leading organizations are tackling this through configurable pipelines that harmonize data from diverse sources. Rather than attempting total standardization, they're creating purpose-built connections that power specific AI use cases.
Perhaps most importantly, this revolution isn't replacing humans but empowering them. By handling routine tasks and providing AI-driven recommendations, these systems free people to focus on higher-value activities like relationship building and exception handling. The workforce is evolving from task executors to AI trainers and process partners.
Ready to transform your operations? Discover how building a next-generation GBS hub with the right technology platform can position your organization at the forefront of this enterprise revolution.
Discover how technology is reshaping our lives and livelihoods.
Listen on: Apple Podcasts Spotify
More at https://linktr.ee/EvanKirstel
Hey everyone, fascinating show. Today we're diving into one of the hottest shifts in enterprise operations how global business services is moving beyond the sort of cost-cutting and efficiency mode to becoming a strategic growth engine powered by agentic AI and unified data platforms. Really amazing guests today industry insiders and experts and allow them to introduce themselves and their missions and organizations. We'll start with Shiddich. How are you, sir Nice to connect? How would you describe your role these days?
Speaker 2:Yeah, of course, I'm Shiddich, based out of London, part of Infosys retail, CPG and logistics business. In my role, it's a primary responsibility to help solve the customer challenges using technology, using consulting services and, more recently now, some of the new AI platforms that we'll talk more about shortly.
Speaker 1:Yes and Praveen, if you could introduce yourself to the folks listening and watching.
Speaker 3:Van, great Thanks for having us. I'm Praveen Kogbial and I lead the global sales and solution consulting team at Infosys Edgeworth. Van Edgeworth is a product subsidiary that was set up about 14 years ago with the sole intent of creating world-class platforms, digital capabilities, to help our customers solve transformational challenges. And what an exciting time to be here. We're in the midst of the largest mega trend trend, so our platforms are being leveraged by global customers in helping them scale and adopt AI and drive business outcomes that are meaningful to their business objectives and goals.
Speaker 3:So very excited to be here. Would love to share more and know more about how the industry is shaping.
Speaker 1:The big topic, of course, is AI, and I have to ask Praveen I mean, large global enterprises have been automating processes for years, so why, until now, hasn't investing in AI really shown much results? What's been the missing link in this long AI story that we've been participating in until now, from your perspective?
Speaker 3:That's a great question, ivan. Look. I think we see this as it's a trend that is bringing together a lot of capabilities. So, first of all, there's been so much more investments by the industry in AI in the last five to seven years, by the industry in AI in the last five to seven years. We've had traditional AI machine learning for years and we have leveraged that in certain sub-aspects of our industry in understanding, predictions and suggestions, et cetera. But I think consumer AI and the investments thereof in core capabilities has really helped make this technology fairly usable for enterprises to scale with. And, of course, as with all things, the industry drives a certain agenda. Obviously our expectations rise and so on and so forth. So that's where I'm at Now. Why, Look?
Speaker 3:Automation had great value in the initial phases of shared services setup. But look, at the end of the day, they were deterministic. They got to a certain element of productivity and straight through processing and that's about it. That's all you can do with root based and that's where humans take it from. There, right, we're thinking, we're reasoning, we're collaborating Now with the latest technology, especially agentic AI. All of that is possible with software. So we just see this fantastic value that is possible where humans are orchestrating the software and that software and agentic software is going to drive business operations. So tremendous value to be derived. Much to be done ahead.
Speaker 1:And Shatish tell us about some of the bottlenecks that are being solved. At the moment. We see fragmented data, silos, fragmented systems. How are you looking at this new opportunity to tackle true enterprise-wide transformation with AI and agentic and other approaches?
Speaker 2:Yeah, thanks, ivan. So first of all, I just want to answer your previous question in terms of what are the trends and how are we trying to accelerate? So, look, what we saw initially is the shared services was just cost arbitrage and just moving certain workloads from finance HR. Then it became legal and procurement in finance HR. Then it came legal and procurement. But now what we are seeing is core operations, be it order, be it booking, and some of the downstream processes are being also moved.
Speaker 2:These are really voluminous processes and GBS now has almost 30% of the overall workforce for some of the advanced enterprises. What that has led to is, as you rightly asked, there are certain bottlenecks because the tasks which are being brought from front office, middle office, to back office don't necessarily have a system of record supporting them. They've been tasks been done in Excel files, manuals and macros and whatnot, and that has become a huge bottleneck. Now, of course, with good quality data which is available and by leveraging platforms, which we'll talk a bit more you know where we have leveraged the data, identify the friction points within the process and start automating and eliminating. What we have seen in our experience, with Praveen and me are jointly working on, is, with deterministic patterns, we could achieve in the range of 40% automation. Now, with cognitive and AI and even agentic AI, we can push the envelope up to 70, maybe even, in some cases, 80%. But critical is having data, having harmonized processes, and, of course, we need good change management. So that's some of the recent transformation that we're driving together.
Speaker 1:Amazing and enter EdgeVerve AI next. Praveen, Maybe you'll define what that is for the audience exactly and how does it tackle some of those challenges that you just told us about regarding enterprise-wide transformation.
Speaker 3:Ivan, I like to put this very simply by first of all saying a lot of AI in the industry today are these amazing gilt-headed hammers looking for new nails? Are these amazing gilt-headed hammers looking for new nails? And that's why so much of the industry's initial foray into AI and the POCs have not gone into production. What does it take for AI to be scalable in an enterprise, in production? The answer lies in the context of the enterprise.
Speaker 3:The context of an enterprise is its data, is its applications, is the pipelines that you need to correct to make sense of the information that exists. It's processes, it's workflows, it's capabilities, it's experience layers that people have defined over years and operators are working on. Layers that people have defined over years and operators are working on. If we don't bring all of these elements into the mix of helping generate an insight with a poly AI layer I've got machine learning, I've got document AI, I've got agent AI, I've got generative AI At a poly cloud level, you don't want to get locked in one or the other. There's Azure, there's AWS, there's many other global capabilities that are out there. So, bringing all of this to help power an insight which leads to an action that completes the loop of how business realizes an outcome or an output is the single thing that they're unlocking with the AI Next platform.
Speaker 3:So it is about bringing a data layer, a poly AI layer, a process layer and an experience layer to power end-to-end transformation of business processes of the enterprise or the IT processes of the enterprise. That, in short, Evan, is kind of what the AI Next multi-stack platform stands for.
Speaker 1:Wow, amazing and Shittich. When it comes to global business services, this back office function, as you described it, has been important but not exactly strategic to the business.
Speaker 2:I think this might be changing how so, and how do you see it evolving into a strategic lever for growth and revenue and other opportunities?
Speaker 2:Absolutely so. As I said, initially it was more of like task outsourcing and moving a part of the process, but more and more I'm seeing a deeper penetration of the process to GBS and I shouldn't call it back office, but it's like a value office now because we're able to do far more higher value processes, optimize the process, sometimes even contribute in terms of reimagining the process. So they're really partners now and that is what I have seen. In fact, I've seen the process owners, process partners, working very closely with the markets. They're able to optimize business attributes or performance attributes like turnaround times impact the NPS in a very, very positive way. So as we bring in more automation, ai patterns, processes become optimized. We have seen turnaround times as an example, one of the parameters reduced by 98% and that would lead to a massive impact on the front line of the process from a customer experience perspective, turnaround times and, of course, lead to higher revenue because of repeat customers and happy customers. So this is something we are seeing is a new trend with GBSs.
Speaker 3:Fantastic, and Ivan, if we may add to Shatij. Sorry I didn't mean to intrude, but look, we thought we'd also give you a bit of a flavor of what we've seen kind of globally and without getting into the names et cetera, just talking through some of the customer needs that we've seen evolve from the GBSs. So, interestingly, the GBSs have started putting out a techno-functional requirement of what they expect as capabilities to help drive this innovation and transformation agenda. That should be chosen effectively. So you know, here's a very large global beverage producer and they actually put out a very specific request for helping them build out a platform, an AI-first platform, to drive their agenda. So they were looking at, you know, the whole architecture of our platform and how it would be relevant for them.
Speaker 3:Here's another insurer who is looking for actually setting up a healthcare provider, setting up a global capability center in South America and, an interesting twist, you've heard about BOT deals you want to build, operate and transfer at some point in time. But they said I have another T I want to build, I want to operate, I want to transform and then I want to transfer, maybe right. So it's like a BOTT model that's emerging. We saw another one in the financial services space, which is like a regional mayor major player in UK. You might think that it's only the big institutions that are setting up GBSs. That's not true and many of the mid-tier enterprises are also setting up mature GBSs and are looking for partner support, techno-functional support, to help them drive their operations agenda and their transformation. So we're kind of seeing quite a nice blend of industry verticals as well as global scenarios to help drive this change.
Speaker 1:Amazing. Let's talk about the data challenge. I've had probably a thousand podcasts and every CIO, cto I talk with has a real challenge with their data when it's coming from, you know, disconnected sources or silos. And, of course, ai is only as good as the data that's fed to it. So, praveen, how can you help? What are some of the solutions to the data challenge that every enterprise is facing?
Speaker 3:I like to call this the. Where did we learn how to manage data? Even in one of our capabilities of the platform? We arguably have the largest demand sensing network in the world, where we're collecting supply chain data from 5 million retail touch points at a store and SKU level for our CPG customers. Yesterday as of today morning, 7 am, that's where we learned how to manage data and what does that mean. That's where we learned how to manage data and what does that mean.
Speaker 3:We said look, we cannot let, we cannot dictate how we want the data. We've got to build capabilities that people will send data in any which way and form and we can help bring it together, assimilate, transform, harmonize. We brought that same mindset to the enterprise. That said hey, I want to build pipelines within your enterprise and I know you have legacy applications, you have modern erps, you have things on the cloud, you have independent homegrown applications. How do I assimilate all of this data together in a very configurable way to create pipelines that are pointed towards the use cases that are helping power those AI insights? So don't try and solve world poverty and hunger. Go at a use case level, but make sure that your applications, which are disparate, and have their own data structures, but use that to drive the data to drive the data Brilliant.
Speaker 1:We're also seeing agentic AI come to the forefront. It sounds like science fiction, but we're seeing practical use cases even emerging today. Maybe I'll ask Csitic, you know, how do you define agentic? What does it mean for the global services landscape and how is it different from technologies we've seen before, like RPA very powerful or even all the LLMs we're seeing emerge?
Speaker 2:So look, I know the Agile AI Next platform has a library of agentic AIs and we are releasing a huge series of additional agents as we speak. What it does agentic AI is mimics what your human agent does. It A deterministic process would look at a fixed set of rules, fixed set of data and execute it. However, what an agent-tkai would look at is, okay, look at an agent's daily tasks and maps, and it does not necessarily follow the set flow. It can take multiple scenarios based on what an input has been provided, what are the different scenarios, and it can take innumerable paths.
Speaker 2:For example, if I take a process, let's say an ordering process, I can split that into 10 different steps and which may require three or four agents. It can be an email agent, there can be an ERP agent, and so on and so forth, and then I combine this to form my worker agent or my use case agent. So that's how I'm able to stitch and drive and those agents can actually talk to each other and actually form a use case on the fly as well. And what we envisage and this has been talked about in the industry the future is agent network right. Let's say, today I have an agent for one of the retail customers, it should be able to talk to one of its suppliers and communicate and take and drive decisions, of course with a human in the loop, if the transaction is critical. So that is what the future we are driving towards. But, praveen, would you like to add?
Speaker 3:No, that's so true, shetha. And look, it's a very exciting space event. There's some caution in terms of how to set up this. Multi-orchestration engines, agentic ecosystems you have to have the right guardrails. So at Edgeworth, at Infosys, we're working very hard on, you know, defining the right guidelines and the guardrails. I think we were on the forefront of some of the earliest companies doing ISO certifications in this space. We put out some responsible AI guidelines and we're just coming up with responsible agentic framework guidelines. You need to put the right guardrail saying what are you allowed to do? Right? An agent going rogue for sending up maybe some extra SKUs of an order might not be that big a deal, but an agent going wrong and lending somebody more money than necessary. So you've got to find the right guideways and the right human in the loop interventions, and that's why it goes back to the platform event. This is not just about another hammer called agent TKR. Again, the context of the enterprise is its data and its processes. The context of the enterprise is its data and its processes. So when autonomous or near autonomous ecosystems are making decisions, for some critical ones, you need to put it out to humans to make the call, and that is something that the whole platform will emerge. But extremely exciting times. Now where are people doing it right?
Speaker 3:I hear about so many examples and so on and so forth. It is science fiction in what all you could do, but a lot of agentic ecosystems that are existing today are there today. They're operational, they're real. I'll give an example. Here's a bank and it does a lot of KYC processes and we've used document AI to help them understand their corporate customers.
Speaker 3:But a big utopian agenda in financial services has been can I do perpetual KYC? Because these things usually get done from time to time for audits and people are always scrambling, there's not enough manpower, et cetera. But if you could do it perpetually, now you've got these maker agents which are constantly looking at fresh information that's coming. They present it to the checker human who takes the final call, and now far much more data is available in a much more autonomous manner. I love a slide that actually Shethaj uses. It's called a half-robo, half-human. That's the way we see these agentic ecosystems in GBS emerging. We think crystal gazing. 50% of the GBS will be agentic and humans collaborating and coexisting and the humans orchestrating what the agents are doing and controlling it. That's the future we see it.
Speaker 2:And if I take one more example, ivan, you know in the CPG industry this is for a European customer. So the challenge they have is they do very high volume of procurement For large volume, large categories. It's easy. You have procurement experts or some industry tools, but the challenge they have with tailspin, less than 50K, 100k procurement, and that were high volumes. So what we have managed to achieve is make it completely autonomous.
Speaker 2:And that is where the agent TKI comes into play and this is literally human-less. So a procurement or a buyer internal buyer can go through a curated catalog, make a request. Tender gets issued. The suppliers maybe there is an agent on the other side responds to the RFP. The RFP gets scanned. The suppliers maybe there is an agent on the other side responds to the RFP. The RFP gets scanned by the agents and recommendation is made of the top three suppliers and then a bot triggers a negotiation through a chat interface with the three suppliers and even awards completely unanimously. That is something already in production and we are doing more than 100 million worth of transactions through this solution. So yeah, two powerful examples Back to you, ivan.
Speaker 1:Fantastic. So I love the analogy of empowering humans. What is the impact on the human in the loop here? Is it employee experience? Is it maybe customer experience, when it comes to the customer interacting for service or support or otherwise internal, external. Maybe talk about some of the notion of AI working with humans or giving human superpowers. How do you see that evolving?
Speaker 2:Yeah, I can go first. So, look, you know we have taken a slightly unique approach and maybe I'll explain a bit about the stack that we have developed. We believe, you know, we are one of the first ones to have a fully integrated AI stack. As Praveen explained, there are seven components, right from discovery to execution. We've got a human in the loop, like an endpoint capability. We've got an ability to ingest data. We also got RPA built into it. But what makes it very unique is we have a case management solution which is fully integrated. It very unique is we have a case management solution which is fully integrated.
Speaker 2:So when an inbound workload comes, let's say a booking request comes, uh, our platform will identify look, if the, if there are enough inputs and input enough data is available, it can be executed autonomously. It will do it and be working towards 70 percent. So then there is no human required, but let's say there is an exception. So then there is no human required, but let's say there is an exception. There is some special requirement, some notes like a specification, even those we try to automate where possible through LLMs by getting the intent out. But if let's say there are some special requests which requires a human intervention. It goes into a platform. It identifies which agent is active, which agent is skilled Maybe it's a VIP user, vip customer and allocates tasks.
Speaker 2:The agent gets assigned and that's where human in the loop comes in. They get presented with some pre-populated information, like a copilot approach. So it's not just a transaction information. They also get recommendation in terms of how to deal with the transaction and the entire environment that they have. So they don't, they never have to go back to their core systems or enterprise system. That interface that they see is only like it's only shows them the relevant information, relevant attributes. It makes the agent's life far more easier. It reduces the chances of errors, improve the training and onboarding process and it's much more faster. The turnaround times are much faster and now the team lead the process, lead the center heads can actually monitor. This never existed before. It's like a ticketing and a case management. Real-time performance monitoring Right up to the agent level, process level, center level is what we have enabled and, of course, make the job of the agent far more easier through this process.
Speaker 3:Praveen, you want to add yeah, thanks. So, Ivan, it's quite the question, because it is so topical in everybody's mind.
Speaker 3:First of all, I think we're just enriching work for humans, you know a lot of the grunt work that can be done, a lot of data preparation, information is all presented as insights, as suggestions. We're enriching human workforce and so we're allowing for mind space to do more value-added aspects like research, relationship building and so on and so forth that can help it truly enrich end customer journeys. So that's one thing that comes to mind. The other thing that is very interesting is that we've talked about this democratization of technology for a while. Right, you know, cios and CTOs really want to democratize capabilities so that they have a broader innovation adoption agenda within enterprises. But you know what, till?
Speaker 3:Somebody said the best programming language is now English. I think it was Karupati who said that. About three years ago it was not so easy to democratize, but today the AI Next platform, for example, could help you prompt retrieval, augmented back kind of REG cases so easily. We could help you configure the agents yourself with, obviously, the right guardrails and testing protocols, et cetera. So we're truly making a lot of this technology democratized for wider access, with limited deep coding knowledge but deep contextual business knowledge, and that's going to be very early days. But that's going to be a really enriching environment for humans in the workforce so that they can orchestrate and design the software and the agentic ecosystems for the work.
Speaker 1:Amazing, so you know. Final question here, looking ahead what do you think a you know next-gen GBS hub will look like in the near one to two-year future? Given there's no like off-the-shelf product that you can just buy, there's no playbook or rulebook you can just follow how do you get started on building something that's scalable and secure and future ready for your GBS?
Speaker 2:I can go so. So I mean the futuristic GBS you want. You know if I take a very simple framework people. You know you don't need people who in the past you know where you can just get to do repetitive tasks because those will be automated and you'll have deterministic and cognitive patterns. What you need is people who can deal with exceptions, people who can empathize with the customers and really elevate the end customer's experience, but also partner with the markets end customers experience, but also partner with the markets. People who really understand AI, people who can train AI, because you will need these people to literally be partners in this journey. So the whole skill set and the mindset required will be dramatically changed.
Speaker 2:If we look at the technology, we never had a platform for GBS. Gbs was always like the poor person in enterprise everybody will focus on the ho's head office and the markets and gbs will just be left to do things on their own on the side. But now, with gbs volumes and the value that they're adding, I always see like there'll be two parallel platforms system of records for the enterprise system, but system of action or intelligence for GBS. And that's what our focus is. And process-wise, it's a massive opportunity to transform process. So GBS is going to be, and is becoming a partner.
Speaker 2:We are one of the large customers, so they've gone through first wave of optimizing and automating the processes. So they've gone through first wave of optimizing and automating the processes. Now they're focusing on reimagining the process, experience-led and driving from the upstream right up to downstream. But GBS is part of it. So they are now true partners in transforming the business because there's such deep knowledge of the process and the pains within the process right, so they can really bring those insights out. So that's what I see the role of GBS. Praveen, your view, great thoughts.
Speaker 3:Shatish Crystal grazing event. I really think that you know, first of all, where to get started. I mean, the Annex platform is built for GBS. That's a great place to start. But also I see this as a way that the global business shared centers want to upgrade themselves and how they're driving innovation. We believe and I've got a couple of examples- I want to tell you about.
Speaker 3:We believe they will also example this is a very critical compliance function of an investment bank, a storied investment bank, and you know this is like literally, when trades come in, they need to be compared with the trade guidelines that have been built over the years that manage the risk frameworks of how much you can fund, et cetera.
Speaker 3:You now have a generative AI solution on AI index platform in production, hosted on a cloud, with some NVIDIA reasoning capabilities to be able to really power the insight on a near real-time basis as trades come in, for them to make the right decisions about that. It's a new application that was created and the CTO said you know what? This is a really elegant way to get AI into a high-risk compliance function with an application that is built and the application is not about the way we have traditionally thought about it. Oh, you know, I'm going to build out an application that's going to be around for five, 10 years. No, it's here and now. It's got to be done quickly, but it has a shelf life and a lifetime. With some new AI innovation, that might not be long. So we also see the GBS organizations really also enabling a creation of these applications, but on that common stack platform of AI. Next, as we said, you know, across the data, the poly AI, the poly cloud, the process and experience layer.
Speaker 3:So, that's another innovation agenda that we see for GBSs.
Speaker 1:Well, that's a mic drop moment there, so one to reflect on and thanks so much for joining Really insightful, informative discussion. I can't wait to catch up in six or 12 months and hear all of the amazing stories and anecdotes around customers and what they're doing with you on Agentic and beyond. Thanks so much for joining. Appreciate the time and the education.
Speaker 3:Thank you. Thanks for having us, Evan.
Speaker 1:And thanks everyone for listening, watching and sharing. Take care everyone. Bye-bye, thank you.