Today on Episode #799 of CXO Talk, we're discussing enterprise A I the Leadership Lessons. We're speaking with Sunil Sennen, Head of Data Analytics and a I for Infosys. I've been with the company for over 22 years working with customers on their digital and AI led transformation journeys. For the industry leaders you know across the globe, Tell us about your role.
You have a really interesting title, Global Head of Data Analytics and AI. I work very closely with our clients, CXOS and really, you know, helping them understand. How they can look at data and AI or their transformation and really sift through, you know, all the hype, to then convert this into a meaningful blueprint that delivers value, right? That delivers on the promise of data and AI, not just for their enterprise, but also for the society that they touch in.
And also, you know, really creating what we call data economy around the enterprises, which is a very meaningful way in which you can create value for all stakeholders and bring in a network of entities and partners, citizens and consumers together to then create net new value. And that's something that data and AI holds for nations, societies and communities. So Neil, you're speaking with so many different companies of varying sizes. What do people tell you about AI?
Everybody is curious. Everybody is interested in AI. Everybody knows they have to do something. But can you with a broad brushstroke, describe the general state of the market as you're seeing it? This is the conversation and then the. In the boardrooms that our customers are having, clearly there is a very, very heightened interest in learning about AI and what it means for enterprises.
But I think the key questions that our customers have is how do I translate the potential of AI for my business? And you know how I can reimagine my business, how I can reimagine my business models, what it means for my products and services, what it means for my customers and other stakeholders whom I serve, and most importantly, how do I go about it right? You know there isn't a Big Bang approach to AI. And it's something that touches
the roots of the organization. There are cultural aspects of things, there are processes and obviously there is impact on people that needs to be well understood and articulated for how it will amplify the potential for people. You know within the organization and outside, but how do you translate this really into an execution blueprint so that you can deliver on the value that data and I promised for the
organization. So these are, you know some of the key questions and I would say there are more questions than answers in their mind and that's why they are reaching out to us and we are engaging them on figuring this for them as well as the industry in which they operate. You made an interesting comment just now.
You said that there is no Big Bang approach to AI and for folks who may be younger that have not been through large ERP projects, in quick summary, Big Bang means do it all at once. Take the company live is 1 huge expensive long project and you're saying Sunil that AI is different. Can you elaborate on that? We live in a world where there is a continuous delivery of new capabilities that allows.
Not only the enterprise to learn as to how to operate newer systems such as these, but also the the users, the customers, the consumers to really embrace that and and there is a continuous feedback that then makes this system evolve. But if I if I have to break it down into two or three elements as to why AI systems are like this compared to, you know, the Erps and others that you spoke about. First, there is a clear adoption, you know problem,
right? Which is the trust deficit in terms of what AI systems would tell you versus what the tribal knowledge is? I think it takes certain experience, you know work for the systems as well As for the humans who are interacting or using such systems to then build the trust and the way of utilizing such capabilities for amplifying the potential, getting that productivity that is needed, making that impact on the business and the customers.
But underlying problems are also that of data quality. How do you govern such systems? How do you make sure that the system is operating on ethical considerations that are very important for the society and also making that larger impact? Converting this a I effort into something that can deliver good for the society and for, you know, everyone who's going to interact with it, I think it takes certain.
Amount of maturity in order for these systems to be tuned and really looking at you know how this is working with the ecosystem and then you put this at scale which is very, very important. You know you can't get the value out of these systems by only limiting it to small Poc's or experimentations that are important to get started. But I think the end goal is to then scale it at the enterprise level and that is why the journey goes through the quick.
Iterations and what we call digital is to then take this to business functions, operate it at an ecosystem level and so on with these large traditional software projects. They were highly technology based, but still they had impact across the company. If you were doing an ERP system for example, with these AI systems, there still is impact across the company as you were just describing, but it's very different. So how are these AI systems different from traditional enterprise software?
The enterprise softwares gave a new way of or an automated way of executing right, which is how you could run a process at a global scale. You could standardize a process even though there were specific customizations for how individual reasons needed to cater to local compliance laws and so on, but the idea was to bring an automated system and industrialization and standardization of that process.
What we are talking about in terms of AI is to bring the cognitive capabilities into a system that would interact with humans and the other systems at large. And this has to learn from the data that exists in the ecosystem and within the enterprise and as you can imagine. If you have a bias, let's say existing in the existing data, AI would amplify that. And that's something that would completely distort at minimum and give out inaccurate decisions. But also it would then not be fair.
It's not going to be free. It's not something that would drive, you know, or even meet the ethical considerations with which we all operate. And hence the AI systems need to be governed and need to be looked at differently from that full automation journey to then say how am I tuning this? What business problems am I solving and am I solving it in the ways that are acceptable to the enterprise standards and
also to the societal standards? Would it be correct to say that with these AI projects that they have, they retain the elements of traditional software, but now you have these layers that did not exist before, such as learning from the data as you were just describing? At some level, I think it goes beyond that, Michael, in my view.
When you start to look into the trust deficit aspect of things and how do you bridge that trust between air systems and the tribal knowledge, you know the the story starts to diverge from that of Erp's and other rollouts that we've done. But there's also an aspect of culture. You know, the the culture of data, the culture of insights driven or data-driven decision making.
Is a journey. And you know, as as you would imagine in large enterprises, it is not only an individual who has to get on to a system like this and really understand how to work with the system, but it's also the groups of people, teams, not necessarily in one department, but also cutting across other departments and oftentimes even across other companies. And how do you bring an
ecosystem? To that level of understanding and having that expertise to say how you leverage data in AI and solve problems together, this is where this journey starts to diverge and look at how the adoption and the utility of AI for different business functions would emerge. The other thing is most of what we're going to do in AI is to reimagine the model. You're going to see things that
we have not seen before. And in that sense, it's a great opportunity for organizations to differentiate, to create that discontinuous growth potential and also create new models etcetera. But on the other hand, it's also something that needs to be imagined, tested, experimented
and then put into place. And hence, you know, pivoting AI on what it means for business solving problems is where the starting point is. AI cannot be done for the sake of AI. It's not a system that you're looking to put in place because that's the system that's your end goal. End goal is essentially driving
transformation for business. Getting that outcome that you know the enterprise is emphasizing not for itself or not only for itself but also for you know the, the entities that interacts with and the and the stakeholders that it is serving. Please subscribe to our YouTube channel. Hit the subscribe button. It's at the bottom of our website. Check out CXO talk.com. We really have great shows coming up and we have a very interesting question from Arsalan Khan.
Arsalan is a regular listener. He asked wonderful questions and thank you Arsalan for that. And he says when thinking about the ethics in a I Are there ethical standards that organizations can follow? If not, then who decides what is as ethical? How do you make sure that your competitor's AI is ethical? That they're not cheating and putting yourself at a disadvantage? Any thoughts on this? It's a really thorny it's a thorny topic. You can see this as an
evolution. Of standards and regulations that that you have begun to see, but it's also more that are going to come in. But the, you know, ground level, if you distillate down to two or three things that you know companies can look at. One is you know you have standards around privacy for example. And it's a very, very important consideration to see how you build that relationship. With with your consumers and and partners and employees you know who will have their stakes into
the data that you're processing. Really making sure that you have the permissions or the consents necessary for you to utilize the data or store the data or process it for the purpose that you're stating and for how long you want to do. That is is known and it's something that has been defined in many regulations both within the states as well as across Europe and other geographies even though there is more that's that's coming in on that you know aspect as well.
Most corporates operate with values and standards that they're known for, and that's a good guardrail as well. Most organizations, and successful ones at that, have looked at the societal values, and you know how they have created more value for everybody, not only their own consumers, but also others who operate in the societies in which they operate. Those guardrails apply to AI as well. And that's something that's
known. Most importantly, I think it is also to anticipate and see what kind of you know, regulations you're going to see in the industry, you know around the impact of AI on people. In ways that you know would benefit them if done right, but it would also create negative
impact in the society. Anticipating some of those preparing for that journey and making sure that you're doing the right things from that aspect would put you on the right side of the laws and regulations when they do come into effect, and we know that they will. And I think those organizations and enterprises would find success far more than the ones who don't.
And I think beyond this, there are companies that are working together to lay down ethical standards that can be referred to. We are working on some of these as well. We do help our customers adopt. Some of these processes and standards as we build those systems, how do you take care of biases for example, there are ways to do this and we do incorporate those frameworks into every project and every AI driven initiative that we take
up for our customers. Marketing, for example, is one of the most common area where AI has been applied and we have without exception always held trust, ethics, privacy, compliance and security standards. To each one of those are our projects and those customers have gone about benefiting from the use of AI and share that value with the consumers that they serve. So there are frameworks that you could adopt while working. On the air projects have another great question from Twitter.
This is from Kayla Aragonas and Kayla says what are the biggest opportunities that you predict AI will yield for enterprises, Sunil. For the enterprises, it's going to drive, in our view, 3 theaters of value creation. You know, AI is going to accelerate growth.
For enterprises, this is by way of identifying newer markets, newer segments, newer needs that they can serve or serve those needs in a different way, which is far more valuable for customers or to even figure their play in the industry or across industry value chains. You know one of the things that we always, you know discuss with our customers and and guide them on is that the physical products.
Don't transcend industry boundary, but when you think about data, it does and that means tremendous opportunity and potential for looking at newer ways to create these new data-driven, air driven products and services. You could be a medical device manufacturer, for example, one of our clients and using data that the medical devices in this particular case for diabetes, they were able to really help the other parts of the value chain that interacts with those
very patients. It could be you know, hospitals which are in the same value chain and how do you turn the the bane of the industry which is post facto that you know anything that happens is post facto the the sugar event and using data and AI to predict those events. You could then turn this into a pre facto, which is, you know, really working proactively to help the well-being of those diabetes patients, but also going across other parts of the
industries that touch. Diabetes patients could be consumer products on one side, the physical lifestyle products that can increase the activity levels of these very patients and we all know that has an impact the food industry. You know, as a big recipient of all this data and how you could use this for stitching an ecosystem that improves not only the well-being for these patients, but also in general
for society. So you know, figuring accelerated growth is 1 big theater of value creation, analog efficiencies at scale, you know, you could now really push those economic frontiers to do things at cost that are far lesser if done right. And drive more efficiencies into your operation processes in your field operations and how you
operate your business globally. But most importantly also building connected ecosystems, the kind that I was talking about both in the medical device industry as well as in general where you are creating an economy around you through new data and air driven products and new business models is a tremendous opportunity and the network effects. Of such data and AI, you know, products, services can create immense value, unprecedented value in the industry.
And this is what we have, you know, embraced in our imposis Topaz. Offering that we launched, it's a services brand that brings together all of what we have to offer as in process and the network of partners that we've switched together. The solution investments that we're making to help drive on these three objectives for our
customers. Given the differences between AI projects and traditional enterprise software projects that you were describing earlier, what are the conditions that need to be in place in order to get started in the right way? In other words, what are the factors at the beginning that will drive downstream success? AI should not be done for the sake of AI, and what that essentially means is to envisage and envision what AI means for the business and the industry in which you know the company
operates. Really thinking about the fundamentals of what makes AI successful to deliver those objectives is the very next thing you know data? Is it in place? Is it accessible? Is it available? Does it have the quality that
you could trust in? And if there are specific AI projects on the horizon, you could even start to look into whether this is the data that you want to base your AI systems on. The other thing that I would say is preparing for the journey, you know it's it's oftentimes we see enterprises seeing a lot of surprises as they start.
You know for example there are many Poc's that don't see the day, you know the light of the day because the the impact or the cultural change or the you know, the enablement of people who will be working on such systems is often not thought about and the even costs are not, you know, properly
understood or risk mitigations. Contingency planning to see how you govern such an AI system are not thought through and hence you know, doing this as a tech first project which is just a cool shiny technology that has been used often remains in in that very center as well rather than really bringing it to the
business. So thinking and preparing for the journey, you know you you should have and should look at, you know how AI can change the business, but really breaking it down into smaller blueprints with defined very specific objectives and bring an ecosystem together to to really work on, you know, such a thing. The other thing that I would say is to take a responsible AI design, which is to say that the ethics, trust, security, compliance, privacy cannot be an
afterthought. It needs to be baked in right at the front. Even as you communicate, you know what AI is for your business, that you lay down some of those principles so all stakeholders know what it is that AI is seeking to do or for the business, how they can engage and what are the fundamentals and the underpinnings of such a system that you would operate. Arsalan Khan comes back on Twitter and he says.
Organizations want to do something useful with a I but still struggle with shadow IT. And so he wants to know how using a I affects the organizational culture and makes a I more of an enabler rather than an obstructor. And I think this gets to the dimension of culture and
organizational change. Sunil that you were a leading alluding to earlier, Absolutely. And I think it's a shift in the way in which we view AI. AI is not to displace but essentially to amplify the potential it for example, and this is something that we have embraced at Infosys as well, using AI to improve productivity in software engineering, life cycles in the way in which we test our systems or the systems that we build for our customers, how we ensure data standards or
data privacy across all our projects and so on. There are multiple ways in which you would look at AI and what this does is to shift the work value chain where humans and software engineers in this case would then shift to more complex, more value adding activities and you would have yeah, really amplify the productivity of people by running many things autonomously. The same would happen on the business front as well.
And you know we need to look at AI as a way to change or reimagine the business processes or business functions or business models and embrace this to design those new systems in the way in which we need to do. So I think the thing that I was saying earlier, yeah, for the sake of AI would not envisage
all of this. And I think if we put the right foundation and envision the future from a business lens perspective, I think it tends to clearly communicate the purpose of their AI project and also bring the various teams that need to come together it business and find those champions who can then lead the way to create those systems at scale. As you talk with senior business
leaders and. With boards, to what extent do you think that there is an understanding of the complex impact that a I will have on their organization? Because even when you talk about a I amplifying the benefit rather than displacing humans, the reality is is that there is going to be job displacement. As well. So it's very complex. And so again, to what extent do boards and senior business leaders recognize the depth of complexity on their organizations?
I think there's a great appreciation for the complexity that exists, but I think I would say that understanding that complexity and in what are the ways in which you could manage that complexity and turn this into a positive cycle. It is where the effort and the focus is shifting and that's why we are helping our customers really understand how do you bring those aspects into the things.
For example, we take a responsible layer, design by design for example, that brings that thought process up front in the process, so that you're putting the right underpinnings these systems as you build it, rather than letting it be an
afterthought. That can be a nightmare for the organization and similarly, when you are envisaging your business blueprints, the thought process on why you're doing it and how you want to actually do this, how you're going to bring things together in order to execute on. This is a conversation that we have upfront that prepares the organization to then run such
systems at scale. And there are several examples of this where we have changed existing models, they put new models in place as well bringing new processes and new entities together to do things that were not done before. So let me take a few examples of food and beverage company. We helped them build the AI core that help them pivot to a more off store model to serve their customers and integrate digital partners seamlessly while taking care of privacy and compliance
and some of those other aspects. It became the core about the company whereby they were able to then plug in new partners as they evolved this model and very successfully, you know continue to have that consumer loyalty and in fact built new loyalties on the digital channels which was something that you know they were able to take advantage of. Similarly for a national railroad company, they were envisaging a new ecosystem that
you know they they could create that would improve the yield and the throughput of the value chain.
And this included not only the other partners, the you know the first mid mile partners, but also their competitors who could be part of this ecosystem whereby the entire industry is able to increase the throughput, the economic throughput, but also shift their position from being a commodity provider which is capacity in this case to a value added player right, where you could look at the end to end business outcomes for their customers and be able to orchestrate it in a very complex
web of partners who can then dynamically come together and so on. So there are a number of examples where we have delivered these systems at scale and have worked through the underpinnings of making sure that we're doing this right. We're bringing those micro change management principles which allows the organizations to scale this and then bring the teams together through those
learnings. We have a really interesting question from LinkedIn. This is from Mike Prest who is Chief Information Officer at a private equity investment group, and he says the following. He says as new adversarial A I agents are introduced without ethical limitations to penetrate enterprise systems. Technology leaders often struggle in balancing optimization and innovation within their organizations. And here's his question.
What would you say to leaders who are under pressure to develop a I and the consequences of acting too fast or too slow? And I would just add to that as I speak with business leaders, one of the challenges they face which I think is very much along these these lines is. There's an expectation that they will make these investments, but yet it's a shifting, it's all a
shifting ground. And so how do you, how do you, how do you invest in something that you know you need to invest in, but you don't know exactly what you're investing in because it's all changing? I'll take this in two parts. You know, one is how do you balance the need for moving with speed, but also keeping that purpose and the responsible
design in consideration. I think the first thing is to be able to clearly articulate what division is and what is it that you're trying to achieve and having that translate into a blueprint because that brings the appreciation for what the what are the design and ethical considerations that need to go into this. That would also make all the teams involved in this ready for dealing with that particular challenge and hence not make decisions that might not meet
those considerations. So I think that articulation is very important. The 2nd is to look at not only addressing this on a case to case basis with with each enterprise with each project, but also to develop not only standards that your projects can look at but also you're building your platforms in ways that it has those underpinnings.
In fact for one of the global retailers, we looked at building a privacy first data platform and what that essentially did was that in this particular case when they were engaging their partners, they were engaging various different projects, AI teams internally. Each team did not have to you know deal with the complexity as in you know as if they were doing it for the first time. The lesson learned and the best practice were best practices were baked into the platform.
So for example, we used AI to discover privacy sensitive information which was very useful for every AI project as they were coming out and trying to leverage the data that existed there. We had automated workflows for privacy sensitive information that was not properly masked. So it was not left to the decision of each project to see
what should they be doing. The workflow has baked in rules and and the actors to whom such an approval should go to and that allowed the organization to kind of scale this while protecting the underpinnings that are so very important for doing this. And I think those things can meet the need for speed on the business side because they want
to move faster. But you also have a way to ensure that you're not violating the privacy considerations or not meeting the compliance standards or the ethical standards. So those are a few considerations to keep in mind and work with partners that can build an ecosystem across people, process and technology. Now you just spoke about the retention of lessons and incorporating lessons that are learned into new projects, and we have a question on exactly that topic from Twitter.
Elizabeth Shaw asks how do you take lessons learned from prior AI implementation engagements and use them to support support new client engagements. So this is where this becomes an evolving, you know, practice. There are there are a few ways in which we do this. One is we maintain blueprints that are available to our practitioners globally and this has all the updated standards best practices. You know the lesson learned in these standards.
But more importantly, we bring a community of practitioners together wherein they share the learnings, they share their experiences and and look at ways in which they've dealt with some of those challenges. Many of our customers look to understand, you know, how these things are taken care of. Obviously the confidentiality of each project is maintained. It's only the ways in which we are dealing with some of these challenges that get discussed in the community of, you know,
practitioners. We bake this into our solutions. So any solution that is used by the practitioners globally have these standards baked in as well. And of course for our customers and we are engaging on these projects, we lead with data strategists who are able to engage with the business stakeholders, the Cxos and envision the blueprint or the business potential. Like we say, the biggest problem to solve in this industry is to
find the right problem to solve. And that's where our data strategists come in and they are well versed with the standards and you know the compliance laws. We guide many of our customers on privacy standards or you know, looking at remediations that are necessary in their systems to operate such systems at scale.
So when you are doing this, any kind of such projects building that ecosystem wherein you can push this into multiple vehicles that your teams would use for implementing such projects become important. So best practices industries or the standards document putting this in their training systems so that anybody who's getting enabled on this is well aware of, you know, those standards that need to be invited and then the projects that they will execute.
And also, you know, making that available through, you know, data privacy office or the compliance office is also a very meaningful way so that people know who to go for getting that guidance. And this office can really take on the initiative to make everybody aware, enable them, engage them, become a resource when necessary to to guide those teams as well. On the topic of teams, what would you say is the the team composition that an organization
needs to look for? It's clearly a business first approach to this where you're looking at the business teams really coming together along with the IT or the technology teams to deliver this. But they are like we're saying the considerations of the responsible by design.
So you would definitely have a play of your data privacy or leads, your compliance leads, you know, can audit the project or give blueprints upfront for what the projects need to comply with and similarly looking at the other considerations of data security etc. To bring those experts. So it's essentially, you know, our tribe so to speak that brings together these skills and in an agile passion compose these teams to to address the skills required for delivering
on that project. Like I was saying earlier, it's it's a dynamic composition because you would take on, you know, the business, you know needs in an agile passion and hence building that tribe wherein you're able to pull these resources and create the part necessary for for addressing on this. So it's it's a a more of a cross functional team that you would have to work on your.
We have a very interesting question again from Arsalan Khan. And, he says should given the emphasis on AI systems today, to what extent should organizations be focused on AI? Tools and projects, as opposed to traditional business and digital transformation projects using traditional enterprise software. Over the past few years, we saw businesses embracing digital
businesses embracing cloud. In fact, the businesses that embrace cloud were able to respond to events like Pandemic way better than those who did not. So they've invested in cloud, they've invested in digital. And getting to AI is the very next logical step where you're able to then amplify the outcomes that you can get through this. So it's kind of a continuum on that particular chain, but it also leverages the investments that businesses have made in the digital and data thus far.
It then enables them to get quicker ROI through AI projects, through AI initiatives and that's you know very important consideration even as you look at scaling the AI initiatives to enterprise level, the underpinnings that you have in your digital and data would allow you to scale it at that level. Simple example is that if you are using generative AI for enabling users or consumers to ask questions and get answers, you would want to have the right level of authorizations built in.
For example, let's say I'm a non finance person and I shouldn't be seeing certain numbers. You want to make sure that the generator via system does not give out the information that I'm not supposed to see. And those things are well baked in into the digital and data foundations that most enterprises have, you know laid and that can be scaled to the newer systems as you're doing this. So I think this is a continuum that you build on.
It allows you to get ROI quicker, ROI on the investments that you've already made, allows you to scale at the enterprise level and with the right considerations put in place of responsible by design, you can operate with confidence as well. So it's kind of an initiative, but there is an aspect of experimentation that has to take place with AI.
That's a very important aspect of how you will figure or learn new opportunities that business can really take advantage of and how ready or what kind of data do you have, what kind of data quality you have to really solve. Some of those problems will come through the experimentation punnel, but when you're scaling it, it's gonna go back to some of the foundations that have been put in place. So you're kind of getting from digital cloud to now AI?
That experimentation process or do you find that organizations? Are having trouble with that, or does it seem to go pretty smoothly for folks that are very process bound? I would imagine that this experimentation is just a very different way of thinking the. Enterprise is to think about setting up that experimentation ecosystem. We guide our customers and we do a number of engagements for our customers. We were thinking through the
experimentation. That is not wasteful but is productive and there is a way to think about it. How do you funnel ideas into the experimentation zone? There are ways to do this, you know, through design thinking on one side where you're exploring with business what problems can be solved and in what ways can that be solved.
You could use data to nudge and recommend what areas you could look at. You know, for example data could tell you a trade promotion of certain kind, could improve, you know the sales for other teams and could become the idea for you to experiment on. Or it could be the business teams coming out with newer ideas that they would like to look at because they are hearing those problems in the field or they are experiencing certain bottlenecks in which you know
the business is experiencing you know problems. But how do you then run this through the idea funnel to scenario certain things You could simulate to better understand those things, convert those into real Poc's and you know the the small experimentation projects. But really putting those measure measurements in place whereby you are able to evaluate what the experimentation is, telling the business in terms of what it can get.
And then connecting that into, you know, how you can scale successful ideas when when they need to be pushed to that. But more importantly, also feeding those experimentations back into the funnel. So that next time when the business is looking to do an experiment that has already been conducted by somebody else, one could discover that and use that to then see whether there's a need to do this.
So I think there's a a clear experimentation design that one could adopt, making sure that the whole experimentation cycle is serving the need to innovate at speed, but also gives you the basis on which you could scale those ideas and make this a very productive cycle for yourself. With that, I have to say a huge thank you to Sunil Sennen. From Infosys for taking the time to be with us. Sunil, thank you for being here. I really, really appreciate your time and your expertise.
Thank you so much for having me on your show and it was great talking with you. And thank you to everybody who watched, and especially to those folks who asked such great questions. I always say this, but you guys are an amazing audience. You're so smart. And we love your questions and they add so much to Cxotalk. Now before you go, please subscribe to our YouTube channel, hit the subscribe button, It's at the bottom of our website, checkoutcxotalk.com.
We really have great shows coming up and we'll see you again next time. Thanks so much everybody and have a great day.