Saurabh Gupta — President, HFS Research[00:21]
Good morning, good evening, good afternoon, wherever you’re listening in to today’s episode of HFS Unfiltered. And look, I wanted to have a very special conversation today on how to AI, because I think the why of AI is very well understood. The need and the value proposition for AI for enterprises is very clear. The what in terms of the technology and the solutions is also evolving at a ridiculous speed right now. But where we at HFS see the challenge is how, how to make it real. That somehow still remains a black hole because we’ve actually done some research with IBM and over 90% of enterprises are still struggling to go beyond the pilot purgatory when it comes to agentic AI. And I’ve got a fantastic person to debate this with, Thiru, who’s joining us from IBM, who leads a lot of enterprise AI transformations with clients of IBM. So welcome to the show, Thiru.
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[01:35]
Thank you, Saurabh. Glad to be with you and especially having a conversation that is near and dear to my heart. And this is where we are spending a lot of time.
Saurabh Gupta — President, HFS Research[01:44]
So, Thiru, let’s jump into it. When you think of what I was talking about, how to AI, it’s still a struggle, right? No matter how you splice this data, only single digit enterprises are able to move to AI at scale, especially agentic AI and gen AI. You know, machine learning, you sort of perhaps, perhaps cross that chasm. From your perspective, what’s, what’s missing? What capability gaps are missing? Why, why aren’t we able to move beyond the single digits?
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[02:20]
Yeah, great question, by the way. So a few, few nuggets that I see. First, you know, we go back to how it all started. The ChatGPT moment, right? One day we all woke up to the excitement that ChatGPT brought in the consumer space. But when our clients try to replicate that success, you know, they come to work next day and say I want to have that same success for my enterprise in my day job. There’s a number of things that come in the way, right? And this is where the first part of the how really happens because there was a notion that what made it successful in the consumer space of the AI is the same as enterprise space. There is a clear realization among our clients who have tried it that the enterprise version of the AI success is different and requires different things. Things like, you know, enterprise-grade security, for example, right? When you put something out there, it needs to be secure because it’s on core systems. The other factor would be ability to integrate with enterprise core solutions, SAP, Oracle, SuccessFactors, using very standard mechanisms like A2A, agent-to-agent, or MCP server mechanisms. That’s critical. Ability to scale, ability to perform on a global organization. We work with a lot of organizations that are globally deployed. Ability to also get at the data that it’s at the root of it and its quality of it, right? I would be the first one to say, you know, data quality problem was a problem before I started my career and it will be there when I retire. So it’s not about boiling the ocean, but it’s important how we get at it. And we’ll talk about it in the coming minutes as to how the data would be there. So there is a number of factors that are critical for success for AI implementation in enterprise space. And that remains to be one of the key things our clients need to make sure that they realize it, they pay attention to it, they plan for it in their engagement. One answer among many that we have found that improves the odds of success is going and subscribing and relying on a platform approach. Our clients have like two, three different options. They could go to, you know, some of our biggest clients with very unique IP, like L’Oreal, for example, are going after creating their own LLMs, their own orchestrator, for example. That is applicable to large, multi-billion organizations that works. Many of the organizations are going to the more, you know, pointed solutions from dot-com companies, you know, $70,000, RFP reviewer, that works for some. But for the vast majority in the client base, in the middle, they need a platform. And that platform kind of takes care of the enterprise concerns that I had talked about. IBM has one. We call it the Enterprise AI Advantage. So we have a platform called IBM Consulting Advantage. On top of that, we have developed Enterprise Advantage as a solution offering that comes with both the plumbing that is required to make AI implementation successful. It comes with AI forward engineering. It comes with use case development, identification and prioritization. And it comes with maintenance and support of these solutions in a standardized way across the globe for the organization. So to me, paying attention to these enterprise needs in AI and relying on a platform, IBM has one, there are others as well, would be a big factor in how you ensure a higher percentage of success. Not to say that it’s going to be solving all the problems, but it solves some of them and improve your odds. Hope that helps.
Saurabh Gupta — President, HFS Research[06:41]
I think that’s exactly what we’ve also seen, Thiru. We call these enterprise debts, you know, and every enterprise is struggling with, has a huge amount of data debt, has a huge amount of process debt, has a huge amount of cultural or people debt, and has technical debt. And AI is not some magic that will solve for these debts. You have to, you know, I keep joking that, you know, there’s a limit to how much you can borrow on a credit card. At some point of time, you have to pay the credit card bills. And I think we are in that stage. The other question that I wanted to ask you also, Thiru, is people are obsessed about cost reduction, and rightly so. We are in troubled times and economy isn’t doing well no matter which geography you look at it. But sometimes if you’re just obsessed with efficiency and cost, that also can be a detriment to value creation. And how are you advising your clients on what is the business case for AI? Is it just sort of keep doing things cheaper and faster or is it something else?
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[07:56]
I think the single largest question that I get from our clients is many of the boards are asking the leaders, you’ve gone ahead and bought a bunch of licenses, for example. You jumped into the AI bandwagon. Where is the return on investment? Our client C-Suite are grappling with this question. So it’s very timely. I kind of say there is three or four disparate things when it comes to return on investment. Okay. Just like everything else, clients need to be very thoughtful about getting into AI functions and enterprise AI investments. First is there is a group of AI opportunities that are, I call it CFO ready, near term. These are low hanging, very well understood use cases that you can implement and get an immediate return on investment. And I’m talking immediate is four to eight weeks. You start to see the return on investment very, very early. And there is a number of examples of that. Then there is the intermediary period where it’s six to 12 months to sometimes even 18 months. You’ve got initiatives that are more, what I call it as the competitive differentiator. Meaning if you don’t do those things today and invest in ahead of time, you’re going to fall behind. And AI as a competitive advantage will become a competitive disadvantage because all your competitors will move ahead. So there is a second group of AI initiatives that are very targeted to this group. Then there is the third group, which I call it the moonshots. These are things that you need to invest in now as a business leader to prepare the organization for what’s coming, right? Because AI is evolving, machine learning become AI, AI become Gen AI, Gen AI become Agentic AI. Now we are talking about voice AI, proactive AI, autonomous AI, and physical AI. It’s going to continue to move. If you don’t invest in the future on these moonshot ideas, you’re going to not be serving your role as the officer of the company to prepare the organization forward. Now in these three categories, if you look at it, several, on the short-term CFO-ready things, absolutely, you’re going to look for immediate and direct line of sight to return on investment. This could be a cost reduction play, totally. I’m subscribing to that. Then when it comes to the second group, which is the, what I call it as the competitive differentiator, the 6, 12, 18-month play, that’s not really a cost reduction play. That’s actually a revenue growth play, if you ask me, right? We did certain use cases for some of our clients. And as a byproduct of that, they realized the AI produced some information that was hard to produce, and it took longer. And that actually was available now. And it actually produced some competitive differentiation of, oh, only my company sells these products, right? None of my competitor has actually this product. I could actually produce more of it, price it higher, etc. So there are some meaty examples that we leaned on and learned about. That actually tells me it actually increases the revenue. It’s not just a cost reduction play on this. And then the third category is also not a cost reduction play. That’s an innovation play. Companies need to innovate. When you go after these moonshots, this is the R&D type of investment in there that actually produces new and fantastic ideas to emerging opportunities in the market that were never there because AI wasn’t there. So if you think about it in these three categories, the cost reduction is really on the short term immediate play, revenue improvement and ability to sell more, and then innovation is on the other two categories. That’s how I advise clients to think about it.
Saurabh Gupta — President, HFS Research[12:19]
That’s a very helpful framework. It’s almost three horizons of AI investments, and potentially it can create a flywheel effect of investments. Your cost reduction can fuel the investments by Horizon 2 and Horizon 3. That’s very helpful. But the other corollary of that, Thiru, is that AI is also disrupting several industries, and our industry is one of those that’s getting disrupted as well. You know, you’ve seen the SaaS apocalypse that’s starting to happen right now where, you know, I’ve never seen software stocks go down that dramatically. But every IT CIO client that we have is asking, let me look at all my software contracts. Let me look at all my IT services contracts, you know, and is that, what are you advising? How are you thinking of the disruption that’s happening in the IT world, you know, both from a software and services perspective?
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[13:28]
I think this goes back to your first question of how do you do the how correctly, right, for organization? It’s closely related to that. Because the how correctly is we advise our clients to say, throwing AI as a technology on top of old business processes is not going to land results, right? And there is no quick answer also. That’s really, you know, the first realization. With that realization, what we say is, you know, we have a tool called Component Business Modeling, cbm.ai. What this is, as you may have seen, it takes the industry by industry and puts them in a, you know, two-by-two kind of box where different functional areas are laid out. We do a heat map, and then that heat map further goes down to use cases and agents that are high value on the areas that the client plays in within the industry. We kind of did that to ourselves in our client zero example where IBM took our finance, procurement, HR, legal, all functions and really looked at the workflow and said, what are we doing now? Right. Two things. How many software pieces that we have subscribed to, disparate things, overlapping things, and how is our workflow working now, right, business workflow? And we got at it together and said, let’s simplify the software stack, let’s figure out what our business workflow looks like for an AI world. And then the culmination of the two led to an overhaul of what we have today in our system, which, you know, as famously talked about by everybody, saving four and a half billion and saving half a billion every year in that exercise, right? Fortunately we did it to ourselves, we’ve got the IP from there and we are able to leverage it to the clients. To me, that’s critical to be able to look at it at an enterprise level, each line of business, as you know, AI is not an IT thing, it’s an enterprise thing. So every line of business could benefit from it. So we look at the disparate software and the disparate functions that it is serving. And in parallel, we look at the business process and combine the two and say, here is how a human plus digital organization of the future would look like. And these are the software stack and the business processes that will support that, right? Because, you know, back in the days, we used to do coding and development. We would do business analyst testing. We don’t intend or aspire to do it. Agents will do the coding, right? Agents will do the testing. We, as the human in the loop, would do more supervisory functions. So the org needs to look differently. The software functions like, you know, telemetry and their ability to see what’s going on are critical as opposed to the software of the yesteryears. So to me, the answer to the question is that holistic approach using a tool like cbm.ai, which then systematically drives a business process reinvention and drives the need for the new type of software, is how I think we should have organization go about. And the last part is also skills and org structure. We believe there are specific skills that teams need to acquire intentionally, and the org structure need to be different. It cannot be still the same, you know, SAP, Oracle, you know, SuccessFactors, custom development, AMS type of org structure. It needs to be a dramatically different org structure for the AI world where asset-led and AI-based development is front and center of it, agent development. And that serves as a services organization for different lines within the team. Say centrally the assets get produced and being consumed by everybody, right? So there is actually a method to madness that we teach our clients.
Saurabh Gupta — President, HFS Research[17:41]
You know, in the last 10 minutes, we’ve talked about so much. You know, we’ve talked about data, talent, process, IT, org structure. It sort of starts to become overwhelming, right? And so the question for you is, what’s your one piece of advice? Where should somebody start on this journey? Because this is not a slam dunk by any stretch, right? Based on our conversation, it’s not like you press a switch and suddenly you are AI first. So what’s your practical advice on where should, what should a leader do to get started on this journey?
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[18:24]
Yeah, great question again. So we’ve learned a lot in the process in the couple of years. And I’ll share those key two or three learnings that will guide the leaders in how they go about it. Every software project, IT initiative, business initiatives have their own personality. You know, if it is SAP, you put a SAP leader, that’s an SAP transformation, S4, there’s a pattern to it. When it comes to AI, this is a top-down function. This is a board thinking. This is the functionality of a CEO to send the mission downwards, not a bottom-up exercise. Because it lends itself when you come from top down, right? As opposed to individual departments and managers and employees driving it, then it gets fragmented and lost in there. So the first thing I would say is the realization that the CEO is emerging as the central figure in driving the success of AI transformation in an organization. So that is very critical and the board need to be very supportive of it. And then that kind of goes down to the line of business VP sponsoring and then the teams. That’s number one for success. Second is the platform approach that I talked about. There is many different ways to go about this. I would recommend organizations to look at a platform like Enterprise Advantage that IBM is offering. There are other platforms from other vendors. A platform has a few benefits. One, it takes care of some of the common things that is not your core business, but the companies like IBM takes care of and keeps it up in an evolving pattern. So the platform approach also ensures that you get to focus in a rapid way, whether it is the first agent design or a 50th agent for your organization. But there will be some standardization. It’s not like the team one is creating in a certain way that doesn’t interact with the team two. This is important in an enterprise setting because you need to have maintainability. And if you cannot manage what’s happening where, we have in our platform a kind of Grafana type of dashboard where we get to see what teams are using the platform and what is being developed, which LLMs are being used, how much tokens are being consumed, how many users are online. And we also get to see behind the scenes how decisions are made by AI agents. This is very critical for the governance, control, and management of this, right? And the last piece in all of this is security, right? There is a consistent lack of security piece that I’m sensing because of how we jumped into it at first on AI. The traditional security guardrails are a little loose. Like somebody is going to hack a bunch of agents very soon and everybody would wake up. But security need to be foundational in anything that we do, not as an afterthought, would be the three things that I would advise clients to say, go about this in this way for success.
Saurabh Gupta — President, HFS Research[21:58]
I think that is fantastic, Thiru. So in fact, the platform point, we came up with this concept called services-as-software a year or so ago. And our thinking was very aligned with what you are articulating because I don’t think it’s, you can keep hiring a bunch of consultants, whether internal or external, to do this. All these services, the lines between services and software are blurring and you need a more non-linear approach and thinking to build AI and leverage AI. So I think completely aligned with this. This has been a fantastic conversation, Thiru, but before I let you go, I’ll ask you one question. If let’s say you had one wish, and there’s only one wish, that could come true, Thiru, what would that be and why?
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[22:58]
I think the one thing I wish is, you remember we were on a few different journeys now, right? We were on changing our traditional systems like IBM as a company became IBM.com in the dot-com era. Then came the cloud. There’s a lot of questions about is cloud real, here to stay, et cetera. And then cloud is very real. I think I want everybody to realize AI and enterprise transformation using AI is very real. Once we understand that and accept that as a technology evolution, you can never fight technology, right? It’s like fighting CD-ROMs being an evil thing. It’s like it’s here to stay. We’ve moved on from CD-ROMs. So my biggest wish is for people to really embrace the fact that this is not a challenge, but this is an opportunity for us to thrive. And once you get into that mindset, I think there is greenfield right across. That’s my wish, Saurabh.
Saurabh Gupta — President, HFS Research[24:03]
Fantastic. So artificial intelligence is real. So on that oxymoron, Thiru, this has been a fantastic conversation. I think your guidance is very practical and very down to earth and something that our viewers and our readers can actually follow. So thanks a lot for sharing your experiences with AI.
Thiru Venkatachalam — VP & Senior Partner, Enterprise AI Transformation Leader, IBM Consulting[24:32]
Thank you, Saurabh. Thank you for the opportunity. Looking forward to collaborating with HFS more and we’ll see you in the marketplace.
