Amit Zavery Keynote at IITBAA 2025 Leadership Conference

We can use a few slides, but I think I'd rather we just talk.

Thank you. Thanks, Ash. Come on. How are you? [applause] You can come. How are you doing? Good. How is everyone?

All right. So I mean, the best thing for us to do is I give you a few ideas of what I'm seeing in the market. Maybe spend a few minutes on that. But happy to answer any questions you have, and then we can open it up to thoughts you all have and areas you want to discuss. We can keep it pretty open. Not everything is available to be discussed. So you have a slide deck. So if you look at... you talked about today, Ashish, about the great pivot, and we are seeing a huge amount of change happening in every enterprise I speak to nowadays. Since I've joined ServiceNow, which has been close to 11 months, I spoke to around 300 customers and other partners, and all of them are going and grappling and figuring out what to do with AI, how does it change their lives, how do they change the work, and what are the different investments they need to make. And if you look at the numbers out there, everybody is thinking about AI making a huge impact in their business, but they don't know how to get going and where to start. And we'll talk through some of those trends we are seeing and the worry everybody has in terms of what that means for the business, what it means for their own personal life, what it means for competition, and what it means for jobs and everything else around that as well. So that's definitely one big trend across the board. Second is this idea of how do you get ROI on it, right? The investment going on in AI is pretty large. There's a huge amount of projects going on, but there's a lot of concerns about whether I'm getting the ROI, am I getting the value creation out of it, as well as whether the companies will be effective after all these investments. So those are kind of the major trends going broadly, and we can talk through what that means for you individually, as well as some of the companies you might be in, as well as how we at ServiceNow are kind of addressing some of those problems as well. But the bigger thing around that is, if you look at AI innovation, it is impacting every part of the tech stack, every part of the way business is run. The interfaces are changing. The workflow in terms of how you run your business process is really going through a major, major shift. It's no longer very deterministic. You're adding more and more nondeterministic flows, but you want an outcome which is guaranteed as well. So that's a big change happening around workflow. No doubt the whole tech stack with AI is becoming more agentic. So how do you really manage AI agents with orchestration, reasoning, planning? How do you do security, compliance, trust, safety? So the whole stack is changing because AI has been built into that particular stack, and you're building applications around it. And that requires a huge amount of work around data as well. A lot of federated data out there. How do you connect those things together, and how do you make sure that you have access to the right information in real time, which can impact your AI models but also impacts how your business processes run? And the infrastructure has gone through a major shift, of course. No longer is it just CPUs, but GPUs are everywhere. How do you do testing? How do you do planning? But also how do you do inferencing? The whole tech stack change is going to be a major shift for every engineer out there, everybody who's building new products out there. And we are seeing a huge shift in terms of what kind of training I need to do for my engineers, what product managers need to do, how you design especially across the interfaces and this whole multimodality, as well as this idea of multilingual and ways you interact with the same user interface. And that requires a mindset shift, this idea from going the way we used to build and do things before versus now being an AI-native mindset. Of course, there are a lot of new companies which are coming up starting with this, but most of the other companies who have been in the business or in the software market for a long time have to really rethink how you really build products, how you test, how you deliver, but also how you market, sell, support. The lines between pre-sales and post-sales are going away. Lines between what an engineer does in terms of writing code and somebody else going and making customers successful are going away. You talk about forward-deployed engineering, how the engineers are now in the forefront with customers. That means the shift in mindset is starting to happen for everybody in every role across an enterprise. And this is a big skilling requirement we're seeing, as well as a change in terms of how you will have to get your companies in line to understand what the world will be and ensure that people are doing this kind of work and rethinking how they're going to deliver towards what customers expect now versus how they used to be interacting with software vendors and other providers before. So this, I think, is a big area of work we are doing inside ServiceNow, changing the mindset across the organization to ensure across every role and function we're bringing out this AI-native mindset. And this is a heavy lift for us as well. Just, I'll just leave you with like what ServiceNow... and assuming many of you know ServiceNow, but if you don't, you should think about us as an AI platform which provides you the ability to run any kind of business process for any business functions, be it for IT, HR, customer service, finance, supply chain, procurement, or build your own workflow on the same platform. And the key thing ServiceNow has done, a good job over many years, is to really have this idea of one platform bringing all the different pieces you require, not just data or workflow, but AI into the same place with one data model and one user experience. So it becomes very easy for you to keep on adding more and more functions on top of it. The core thing which has happened over the last few years is a shift towards AI being built into the platform and the idea of having AI agents, which is okay, but the broader thing is around orchestration, reasoning, planning, and how do you manage AI agent orchestration across multiple companies, and not just provided by ServiceNow, but also third-party as well as your own custom AI agents, and the idea of AI control tower so that you can have complete control, but also security, compliance, and trust and safety built in, so you manage the proliferation of AI across a company as well. So that's really how we've been thinking about it, and then connecting all the various data sources using zero-copy adapters and making it easier to do insight-to-action across a business process, across everything a company might be going through. So it's a pretty powerful way of thinking about it. Of course, there are a lot of companies who are approaching the problem today in terms of how you build AI-based platforms and what you need to do to really deliver this AI-native mindset and products to customers. So we are one of the providers, and we work in an ecosystem where we partner and integrate with various other providers as well, a lot of startups and other ecosystem providers. So look forward to kind of learning more about all the things you guys might be doing and where ServiceNow can partner with you all. But the goal is to really make sure that you can add value to customers and looking at customer first and delivering a platform which can make a difference and let customers grow both their top line and improve their bottom line. Right? So with that, we can have a good conversation, hopefully, and answer any questions you might have, but hopefully gives you some context in terms of things I'm seeing and a lot of other changes which are probably top of your mind, which we can then talk through as well.

Yeah. No, it's a good question. So we have gone through many, many pivots, right? If you look at technology shift, and I don't know how many of you have worked at Oracle.

So, I mean, when I started Oracle, which was in '94, it was still kind of starting becoming more of this client-server technology stack, right? And we were evolving to becoming more internet-driven. And that was a huge shift for many engineers, product managers, as well as how you go and build those products. And that took probably, I would say, six to eight years before we could say we have an internet-ready, internet-facing product. Similar thing happened then when we moved from internet to the cloud era, where we had on-premise products which we needed to now make sure that could be a fully cloud-based software, and that took another six to eight years as well. And that was reskilling, making sure engineers understand the value of it. There's a lot of inertia because people said that will never happen, nobody will run their software somewhere else other than their own data center. So the mindset shift required took more time than the technology shift, right? It was a lot of work to be done to get cultural shifts going. Similar things are happening now. If you look at what's going on now with AI, even though everybody believes in AI, the reality is there's still a lot of concerns whether it can really make a difference in terms of enterprise efficiency, and will it be without hallucination or without errors, because you can't expect errors in a lot of the enterprise software out there because you're making real business decisions. But the problem with AI right now is the time shift required is very low, very, very short, right? Things are moving much faster than what we saw with the cloud, what we saw with internet, what we maybe seen with mobile and other trends which happened in the technology shifts. So the big, big thing is, one, the way I've seen it work, and I've seen that at Oracle, we did this at Google, and now of course at ServiceNow, is to really make sure that we are very clear top-down about what our primary objectives are, what are you going to focus on, and ensure that everybody understands that there is no out of it. It's not like we are putting your toe in there and see whether it works or not. You have to really go all in. The same thing happened at cloud with Oracle. We had to really ensure that everybody is focused on it, and there's no way out in terms of not moving to a cloud-based software, and same thing is happening. Really making the top-down discussion, having people who are going to be trained and who will drive the change inside the company, doing assessments and going, making sure that people are learning the new technologies and making the commitment in some releases and product features which are going to be driven by this technology shift, and ensuring those things are delivered on time to customers and getting that feedback loop going faster. So the cultural shift is a big one. I think the technology shifts are easily doable. I think engineers are smart and they know what to do. It's just that a lot of times they don't believe it, or you don't get the right kind of top-down directive in many cases. Said thing that this is important for the job, for the company, and for the survival. If you don't get there, you will become legacy. And if you look at the amount of legacy companies out there on software, it's tons of them. Sad thing is that in enterprise software, software companies don't die. So you might think they're around, but they don't grow. If you look at so many companies out there, you're probably growing at two, three percent, and they're around. But that's not going to be the future long-term if you don't get to the new technology stack. And that's the message which we keep on giving to our employees in my company to ensure that everybody realizes it's a growth-driven mindset and major technology trends to be part of it. And learning, if you don't participate early enough, you will miss out, and catching up becomes even much more trickier because what we start today is not going to be how it will be tomorrow anyway. So you will keep on changing your mind as you learn more, but you have to participate from day one and be part of that community and the whole ecosystem, otherwise you'll be left behind.

Yeah. No, I think there's definitely a huge amount of hype around AI for sure, but it is a technology which is making a difference. So if you look at the first early phases of AI, this idea of Gen AI, or what do you call, like more of giving you content summarization and giving you information in a better way to consume. Great for employee or individual productivity. It helps you cut a lot of the mundane tasks you used to do by searching around and finding information, summarizing it, or helping you even get started with writing some documents and content. Definitely very valuable. Even if it makes some errors, you can fix it and you can get going with it, and it really helps with searching and finding information that is there. It'll going to be around for a long, long time. It'll stay around. But I don't think so that unlocks a lot of the value. The unlocking of value is happening especially around this idea of how do you make enterprise more productive. Right? So going from employee productivity to enterprise productivity, and enterprise productivity happens when you can really now run your business much more efficiently using workflows, but making it more dynamic, more changes happening on the fly without having to rewrite the software every time your business process changes, your rules change, your regulations change. You don't want to keep on doing that heavy lift. Every enterprise has a lot of software pieces, right? I mean, every customer I speak to, especially G2K companies, they probably have 700 or a thousand software vendors which they have various versions of the software lying around, and then you're trying to keep up with that. So the big unlock is happening with AI is how do you now build a business process layer to be more agent-driven, AI-agent-driven, agentic-driven, where your automation is happening on the fly, east to west across an enterprise department, and connecting the various systems together and abstracting out the complexity associated with some of these vendors you might have bought for many, many years. That's always been the holy grail for many companies, they've always wanted to kind of abstract out that for many years. AI is changing that. So we are seeing a lot of value. We have customers like, say, AstraZeneca, who's doing drug discovery. The process of getting data to getting approval and getting a drug out is very long. But if you start taking parts of those processes and start reducing the time for reviewing documents and then approving the results or getting the supply chain coordinated and managed, that can make a huge difference, right? Say we have done it with AstraZeneca, it's usually 30 days for them to order a particular part for a drug for manufacturing after all the data they got based on the experiments, and now they cut it down to 30 minutes. Stellantis is another one which is doing some, given the tariff situation nowadays. Every company is trying to figure out where to buy, procure parts from, and how do you manage those spare parts, and how do you ensure that you're not going to raise the prices and who has the right supply. Again, doing that dynamic supply chain has always been a dream for many companies. With AI, we're starting to see that happen real time in a better way for an enterprise level of productivity where they can improve the top line, manage the bottom line, and really run the business efficiently. So a lot of those examples are starting to come up because of this idea of AI used as part of a solution versus not AI for AI's sake, right? And making sure that you bring it into your product's architecture and think about how that will help you run things more efficiently.

Oh, for sure. Right. So I mean, see, customer service is another part you can think about. And we have a customer called Rossmann, which is a large pharmacy chain in Germany and in Europe. Their big thing is that people who are working in the stores, they, one, don't have enough information, no doubt inventory, but what particular thing might be doing or information they need when somebody asks questions, and searching that information becomes very difficult. And sometimes they don't really have... language issues might be one other thing in terms of finding information fast is difficult, and the users and customers get frustrated. So we started to see a lot of this technology enabling their application where, one, easy finding information, but also second, helping them be much more specific and curating the content to an individual user's query, right? So a lot of this work we're doing with the agentic is the same idea, is like understanding intent and then finding the relevant pieces of information through this technology to give you that summary, and that user who is responding to that back can do it. Another thing we've seen is customer service, where every company, like we're working with UPS, for example, they ship things globally and they have call centers which are, of course, addressing questions in multiple languages. They can't staff up for people for every language. So with voice AI, for example, now we are starting to deliver the idea of doing multilingual, where somebody can be speaking in one language, you don't have to have a call center be able to understand all the languages customers might be from, and be real time still communicating using voice AI as a tool inside a customer service product. Right. So the frontline workers are starting to get much more efficient and also more, I would say, less worried. The anxiety level is pretty high in customer service work, right? Our attrition rate in a customer service group maybe a few years ago used to be probably 30, 40 percent because it's not a great, easy job. Our attrition rate now is like less than 7 percent because we moved to this idea of self-service, a lot of deflection, and then reducing the work an individual has to do in terms of answering every question themselves, and they don't have to be replaced, so you don't have to retrain them. And so your employees are staying longer, the churn rate is lower, but you're also getting much more value out of the people who have a lot of domain. So the combining the context of something in somebody's human brain with AI, that combination really works for the frontline workers.

I think, and one, of course, you use a lot of the years and years of experience to kind of figure out whether this even makes sense, right? So that I think is experience, like a lot of you as you have would be based on your work you've done before. But the biggest thing is really customer outcome, the value creation, right? So if you're not going to build something or deliver something which does not really solve a pain point for a customer, and it really can be not like a dream, but really you can deliver it because the technology exists or we can build it, that's an important kind of discussion, right? So I always really go down to the idea of customer outcomes and the value creation around that customer expectation and delivery we want to do. So finding a design partner, understanding what they are trying to do, and not just kind of do something which nobody really sees importance out of it, right? You can't monetize otherwise, and it takes a lot of investment typically to build some of these things. Sure, we always keep on pivoting once we learn more, but we don't want to go down a path which is completely a dead end when nobody really needs any of the software we might be building. So customer value is always the first part, and then out of that is always the technology, the know-how, the expertise, and can we really get those things done on a real-time basis.