Amit Zavery, ServiceNow President, CPO & COO on AI at Enterprise Scale

Hey, good to see you, Grace. And but thanks for having me.

Yeah, no, no. I've I've been very fortunate to have a chance to be part of a lot of good transformational companies. Uh, you know, when I joined Oracle as a software engineer, and Oracle was really the kind of in the forefront of transforming how applications and technology stacks evolve, right? We were going through the whole dot-com era and then the cloud. So, I got a chance to be in the having a front seat at deciding how we build the right technology, right products, and take a lot of the core capabilities and deliver value to our customers. So, that learning was great. We got a chance to build out a whole middleware business at Oracle, which is the platform. It nowadays we call it platform everywhere else. So, that was we started with like less than 100 people, and I get to go to chance to partner with Thomas Kurian, who's currently the CEO of Google Cloud. And we built this middleware business, and by time I left, it was 6 and 1/2 billion dollars. But it took us some a lot of effort, of course, and be able to deliver that value to Oracle, to our shareholders, and to customers with a great experience. Google Cloud, I followed and joined Thomas Kurian at Google when Google Cloud was just kind of trying to become serious and become bigger in the cloud space. It was when I joined, we were at less than 6 billion in revenue. And by the time I left in six six years, it was 45. So, we got a chance to again take good technical foundation capabilities Google had built, build out the enterprise business, and take it to a really make it a very competitive in the marketplace, and really create a space for Google in that space as well by delivering innovation and value to our customers. Also, similarly at ServiceNow, ServiceNow is of course a legendary company which has been building very critical components of enterprise software. And before I joined, they were starting to look at how they bring AI into the products. And I got a chance to really take it to the next level in terms of how you make it gigantic, how you really deliver automation and and better results for our customers through that as well. And the role I have here is broader in that regards that I have the product engineering and strategy team, but also the operations, which gives me an opportunity to think about what are we building, how we really deliver that to our customers, how they how do you run the company more efficiently, how do we operate and bring a lot of the silos and remove all the silos and bring it into more cohesive outcome, and get to partner with legendary CEO like Bill McDermott. I learned a lot from him, but we get a chance to also rethink how enterprise software will be in the next phase with AI. So, I just been fortunate to do surround myself with very good people, good teams. And have a chance to really learn from a lot of the evolution of technology and bring that to bear for our companies.

Yeah, quite a few things, right? One, it used to kind of be very clear ServiceNow has always been as kind of system of action even though it provides software as a service. So, sometimes things get conflated right now in the industry. I don't think people able to differentiate, especially investors, what are differences between different companies who are providing software to enterprises, right? So, enterprise software is pretty broad segment. From our perspective and what we see differences arises in terms of how we scale our business. AI, one, the speed is just very very fast. Every transformation I've seen, when we did the transformation to web, when we did the transformation to cloud, and what we're doing with with AI in terms of software capabilities, the speed is 100X. Right? It is crazy how fast things are moving. The innovation around it is happening very fast. And second thing I think is customers are moving fast as well. I think the earlier transformation, customers used to be very very slow at adopting a lot of the new technology stack, but they didn't want to make a lot of major changes in their current products or current way of running the business. With AI, there's a huge amount of pressure top-down from leadership in every company, be it the board or the investors and everybody else. So, a lot of customers who are buying enterprise software are getting a lot of pressure as well as interest with AI. So, that speed is great. Second, there's a lot of innovation happening, which is technology foundational pieces like large language model or the infrastructure, which is enabling faster innovation on the upper part of the stack. So, that was not the case that fast because a lot of people ended up having to build all the core capabilities before public cloud became common. Same thing with web, a lot of building blocks were open source, you had to cobble up different things together. It took some time to really build a very solid stack. I don't think that's the case with AI. So, the speed as well as the capabilities are happening already. Second, I think the other thing which is happening is how you now build things. Right? In the in the other transformations, the technology like cloud, you didn't have to build your own data center, which was great. But the people you needed was the same in terms of you need the same number of people, maybe with different skills. I think in the current world, because of the tooling with AI, the coding capabilities, it is reducing the kind of work you need to do hand on day-to-day basis. So, the speed can be faster. You also trying to change how you're building things as well, right? A lot more design, a lot more thinking about how code output without having to write a lot of code. So, that has a big change for a lot of engineering teams now with AI. And how the product role is evolving, right? The product management role used to be writing specs or PRDs or functional specs, things like that. Now, they are pretty much prototyping if not almost live applications which our engineer can take and kind of evolve it very fast. So, that compression of speed, but also the skills and people connecting together much faster from design, engineering, product, operations, and roll-out. It's very very different, right? And outcome might be the same in terms of what you deliver to a customer from the product perspective with AI built in, but how you do it, who you need in there, how quick how quickly are you delivering those things, the iterations around it, and the underlying technology is changing every day. Lot of new capabilities coming out. So, you have to be very smart about what your differentiation is, where the technology you're going to use from third party, how that comes together, and how do you really deliver end-to-end solutions. And that is a big big difference, and that's where a lot of enterprise software is moving towards more consumption-driven mindset, right? You don't have to be seat-based everywhere, right? Like our business, and I would say more than half, if not more, of our business is not seat-based already. A lot of other companies might be completely seat-based. For them, it's a bigger transformation. But it's really rethinking because efficiency will drive less number of people who need to AI agents will be in working with your products more often than humans will be in many cases. So, that changes the game because now a lot of automation is happening through agentic frameworks and agentic workflows instead of people and human really interacting. So, AI is definitely very different at core of every area. It's touching every part of the tech stack, every touch part of your business, and every way you work. So, it is it's a game-changer, and it's a it's a big transformation for everyone everywhere.

I think my view and what we've been looking at and we keep on delivering. One, we let customers come from anywhere. Right? So, is it is it Copilot or Cowork or Teams or Google Gemini Enterprise or could be Slack. Doesn't matter where people start from. The thing where our differentiation and value at ServiceNow comes from is ability to understand what people are trying to do or what they help they need and action that. We have connectivity and capability to really automate that business process on the back end a lot, right? We connect into any system of record, runs runs on any kind of cloud, runs on any large language model, brings data from every different sources, right? And delivers different kind of omni channel experiences. And that I think this any kind of mindset and you can build those applications with any tools. So our goal always has been the central kind of enterprise OS. And the idea of AI control tower is something we introduced last year to showcase that we are taking care of discovering all the things you have running inside your company with any asset, right? Any hardware, any software, any AI and the life cycle around it. What you need to do, what attributes you are running it, what kind of things are making changes to, who owns it, what kind of cost you have associated with that. So we really are the central understanding and the central repository of brain behind operations of a company. So the all these things that Copilots of the world do is really basically getting the information for you. How do you action that and how do you finish the task is where they come to us at ServiceNow. So we're very well integrated into Copilot, for example, right? And Copilot has a ServiceNow interface in it. It can expose everything we do. You still need someone to resolve that issues. And that's where we've been very good at and that's where our differentiations keep on coming up. And that's why we partner with Microsoft, we partner with Anthropic, we partner with Google, we partner with Amazon because they see the value of having that ability to understand what people are trying and need help with. And a lot of the help they need as an employee or a customer could vary across different function areas. It's just not IT, it could be HR, it could be finance. So you need to bring all of those things together. And that's where the workflow we bring in at the business level is the differentiation. We are the workflow company. Now workflow has become a common word for everybody to use and there are different levels of workflows as you can imagine. But our ability to really understand the intent and really execute that workflow for customer really differentiates us. And second thing I think we are evolving to more this idea of autonomous workforce, right? So if somebody has a help they need with IT, we can have an autonomous worker or AI specialist, we call it, level one support engineer. If it is completely AI driven and it's doing full end-to-end work, resolving and getting you the response back without you having to go to individual AI agents. So the day of like orchestrating different AI agents and building you all that stuff is really painful for customers. We have elevated that to this idea of autonomous worker workforce with AI specialist taking care of that problem for them. And that resonates very well because that's nobody else can do that level of understanding. We have the context, we have the data, we have the knowledge and how to really manage that outcome.

No, I mean I think well, see, this this idea of different kind of pricing has been talked around enterprise software for every as long as I've been working here, right? 30 plus years. I think what executives like or companies like is one predictability. And second, hopefully flexibility with it. They want flexibility in terms of being able to have some fungibility of what they're paying you for. As long as you can give them that two things in a structured way, customers are usually happy with pricing. Of course, they want the lowest price, best best lowest total cost of ownership, best ROI. But they understand they don't want unpredictability because you can't plan your business that way. So when you go into this idea that I will give you give you an outcome-based pricing, one, what is that outcome definition becomes very tricky in many customers. Second, who's responsible for the outcome? Is it just the software? There might be other providers around it, right? You might have a system integrator or FDE model or you might have your own IT team doing some of the work. Is it because of software you didn't get the outcome or because of implementation you didn't get outcome or because you didn't have the data which were not available to run those processes or things like that. So it becomes a very difficult thing for a broader conversation when you're really trying to do this as as a scale. Small projects, yes, you can always do that. You know what? I will do this small project, I can give you this outcome. Manageable, that's what a lot of companies do it before and they do it today as well. Once you go to a broader context which is touching many employees, touching many systems, it becomes very tricky to say I will only give you one way of understanding whether you got value. So I think you have to be flexible when you're pricing. I mean, have you thought about what the customer can define and manage? So what we have done at ServiceNow is we have this idea of hybrid pricing. We give you some kind of license we give you licenses which you can fungibly use across many dimensions plus also some capacity which you can burn down. So customers always then get an annual of they know what that annual looks like, what they're going to pay for, what would that cost structure will look like and how are you burning it down and when they're running out of capacity, they can always true up or get more or renew the licenses. And every customer I've spoken to at at least the Fortune 5000 companies, they like that because it's predictable while giving them flexibility. And they can always up and down and go up and down in terms of their consumption. And that will give them the idea of it's outcome driven because they will use less if they're not getting the outcome, they can manage that because they will not do a more contract longer term, right? So outcome is kind of that. Renewals are somewhat in the outcome mix, right? Because if a customer is not getting value, they will not renew your contract or they'll renew it at a lower price. So trying to do it at a transactional level of outcome, it's a very very difficult to think. It's theoretically a lovely thing to have as a concept. In practical purposes, customers get confused. Buying buyer buying centers inside customers changes a lot. Your CIO might change. Whatever outcome you define might not be agreeable to the outcome of the next CIO in many cases. The procurement team might change, they might change the outcome definition. So you end up having more conversations about not the key value of the product. You're having discussions on commercials every time. And that I think is a recipe for disaster if your discussion with your customer is all commercial, not product, right?

I think one, the AI world has clearly need for much more technical conversations. The the technology stack has changed, but also customers become much more understanding of the AI capabilities which are out there in the industry. So the go-to-market for any company, I mean we have done that for some time already, has to become AI driven first, right? One, are we using AI tools to train? Are we using AI tools to message our customers with help our customer understand our value? We are also doing a lot of work to bring this FDE model, right? Where where we can show value to our customers faster than they would be able to get to by doing it themselves, right? So we have black black belt AI engineers who are available to go on with customers to show them first use case. Unlock that first use case because when they see value, then they start seeing doing more and more. But and they'll understand what to do and how to get going because a lot of questions I get from C-suite typically, where do I get started? How do I get going? I Everybody wants to do AI related projects, just they don't know where, how, what and then what what does it skills require associated with that, right? So, the F D model is a very important part of it. So, we are really changing how we think about go-to-market broadly across the partner with my go-to-market leader Paul Fips, who's been very clear about thinking about talking to customers, showing them the product much more technical conversation, the F D model to showcase them one or two use cases, a pricing structure has evolved very drastically to be much more hybrid, and and and then the just kind of making it AI first led conversation with the customers while understanding their current ecosystem. And that's where the openness, the ability to integrate with various different systems becomes a big big part of our story as well. So, the go-to-market has changed a lot. The proof proof of value, the F D thing has really made a difference for our customers as well.

Yeah, I think I think it's we've been doing this for some time and and we see it going forward as well. One the systems are still very fragmented out there. The swivel chair is real. People have to go between different things. The processes are still very painful and you have to do a lot of Bill likes to call our CEO soul-crushing work. I mean, you have to get into a lot of the painful things. People don't really people get excited about but you do it as part of your job. So, what what AI and what the workflows are doing now, especially what we do with the generative workflows, are taking that some of the repetitive pain points out of the system. Where we can connect a lot of the systems together but you don't have to do a lot of manual tasks. As we automate these workflows and automate these processes, you're freeing up resources and time for employees. As I said, the for example, very good example as I mentioned earlier, this idea of autonomous workforce. That's a new workflow, right? It's really the same workflow in terms of what you're trying to solve but how you do it is new, right? So, if somebody's asking for a help when they want to go get a help with their VPN while they need a PTO, few other things, you can ask that request. The autonomous L1 support AI specialist will basically resolve the ticket for you instead of somebody else having to do deal with that case. Uh, human-wise and then you you should take 2 days usually for human to respond and resolve that for you inside the company, now you're getting it done in 18 to 20 minutes. That time reduction for employees is very valuable because they're not waiting around. Second, the the over the the workload on the human agents reduces drastically so they can do a lot of other things and they can figure out how to solve complicated stuff instead of some of the mundane day-to-day tasks. So, that I think is the reality. The work for the workflows are changing into more automation. But the problem with some of the workflows out there where they can just write it with a large language model is not deterministic. It's too probabilistic as they talk about, right? Where there's like no guarantee of outcome, you don't know if the results will be accurate, it might be not the same every time. So, you land up getting a lot of uncertainty in the system. And you cannot run an enterprise business with uncertainty. Even if you're if you imagine your general ledger is off by a cent, you would you believe the financial result of the company ever? Right? So, I think the idea that you can just run everything plain vanilla through AI agents becomes very tricky. You have to bring a lot of structure and context to it. And that's the way the workflows are going towards. And that's what we are delivering to ServiceNow is the structured deterministic workflow mindset using AI as the technology foundation. And be able to now give you an outcome which customers can yeah, this makes sense. This is what I need and I can be flexible enough to get those things happening very fast and make it more efficient. And that's the that's the focus for us now and our generative business processes as well as this autonomous work workforce is really something which customers really like because that gives them the best of two worlds.

Oh, that's a very good question. There are quite a few things. When I joined ServiceNow, one of the big push we put in was to really look at data as a key component of our stack and offering in that space, right? So, we have something we call work for data fabric. Which was basically connecting various data sources and we expanded even further with this idea of zero copy adapters and connected. So, the data can be wherever it is resides but we can use it to leverage and make decisions. And take actions, right? Our business is to really be able to take actions and solve customer issues. So, if we going to take actions, you need to have a lot of information. So, that is where we creating this We bought a company called data.world, which is a data catalog company and governance capability so we can govern and understand the relationship between different data sources. And then put a data catalog to understand the what's happening, when is things are happening, and bring it into a knowledge graph. And then we are delivering something we call context graph, which has all the historical decision-making reasoning, exception handling reasoning, who needs to do what, what happened, how it happened. So, that's the very very valuable information because we run 6.5 trillion transactions on ServiceNow on a yearly basis growing at 30%. But then imagine the amount of information we know about why a business decision was made inside any company. We have 80 billion workflows growing at another 20 plus percent every year running on ServiceNow. So, there's a lot of things going on in our platform and if we can bring that information through our data platform, the work for data fabric, as well as what we do with our data data catalog, the knowledge graph, the service graph, and the context graph, that is a huge value as well as differentiator for us. So, we're bringing that into our stack. We sell the work for data fabric to customers for connecting their different systems together from the data perspective with zero copy adapter we work with uh, data bricks and snowflake and big query and red shift and teradata so that all this information is available without having to move everything around. So, we can do insight to action. And that's where we also recently acquired a company called Pyramid

Technology. They have a semantic layer to understand how the data is represented, how it should be used. So, when you're building these insight-to-action applications and data capabilities, you have that information available to you in a much more structured manner, right? So, those are the ways we've been thinking about our data platform, and I think it's really making a difference for everyone who is using ServiceNow in their day-to-day processes, being able to now get to data insights as well as data decisioning at the business process level versus the data level. See, most of the companies out there talk about, 'Hey, we have the data in a data warehouse, now go run reports against it or ask me information.' But that's again a very... they didn't have any context.

It's a data level. So, I have tables and I have structure, but I don't know why something was done. And if you go up to the business process level, which is where we have the context for the business — why it was done and what was done — that is much more valuable to a customer because they are able to make real decisions and make a difference in their business much faster than they would be by running some report on a data warehouse.

Yeah, our goal has always been... We have an amazing engineering team here, right? And it's an innovation-driven, very modern, very up-to-skill in terms of things we need to do. Uh, so we have been building out, and ServiceNow has been very good at delivering new capabilities, right? We started with IT, now we deliver things for HR, for finance, now customer support, security and risk. Uh, I mean, we keep on expanding a lot of our capabilities on a regular basis. Our TAM has grown significantly. But also, this is the time to also think about speed in many cases. There's only so much we can take on. Our road map is pretty aggressive. The demand for a lot of things we want to do is very, very high. And there might be some capabilities out there with some domain expertise and people who have a lot of the skills. So, if I have to hire one, it takes time.

Uh, whereas I can bring their IP and their people who can really allow us to accelerate our capabilities, also bring some new ideas and mindset. So, I always look at... build is always our default option. And we do, I would say, 95% build. Uh, then I think when you want to go beyond that, there are areas like we want to kind of acquire and bring in for speed and efficiency as well as new ideas. And other things we want to partner. We don't need to have everything in our platform, right? We need to be able to partner with providers where we can complete a solution. So, that's really the strategy we use. Uh, on the build versus buy part, we have been really smart and thoughtful about what do we bring in where we can accelerate our road map, bring in a lot of new skills, as well as increase our TAM. And this is a chance to do that because there is a lot of change happening. And if you can have an opportunity to expand our opportunities in the market and our value to our customers, we definitely have to keep paying attention to that. So, that's what we've been trying to do with our build versus buy, and then the partner as a broader ecosystem play.

Yeah, no, I think this is something which... there are a few things we need to keep on doing, right? One, the tools people are using in the market are the same tools we use. So, it's not a different tool that a 19-year-old engineer... By the way, we have a lot of 19-year-olds helping us in some of the internships and things like that. And of course, we also have a lot of new hires we bring into the company in the early career thing. So, we're bringing in a lot of skills with people who have also seen some of these things. Plus, our engineers are also doing a lot of the same kind of work as other engineers out there in the industry, or even non-engineers nowadays, right? I mean, all these tools out there, you don't need to be an engineer, but you need to be able to have ideas. And I think that's the part where we differentiate, right? We have a lot of domain. We know what enterprises need, what problems we have solved, and we have a lot of context. So, it's not about just building some pieces of capabilities or features. It has to be a solution mindset. And a lot of people are prototyping and bringing pieces of technology out there, but they are missing out in many cases. One, a lot of these technologies are not backward compatible. So, it's not a consumer world where I can just go to a new browser link and say I start something new somewhere else. As you know, enterprise users — one, there needs to be consistency. Second, it needs to integrate and work with the existing systems you have. You can't rip and replace everything, because the IT budgets for most of the companies are less than 5%. So, saving another few cents here or there is not going to make a difference in the company if you're going to break the business in some other places. So, it has to be backward compatible. It has to be compliant. There's a huge amount of work and engineering resources and product resources that go in to make sure your software is compliant — be it SOX or PIAA or things you might need by different government agency requirements in different industries, or you might need something for European regulations. So, this compliance thing requires a lot of work. So, it's not like you're just coding, but also making sure you're certified and then future-proofing it, right? How do you make sure the new features you need, new capabilities you need, can work and can be delivered quickly while working in the same environment? So, there are a lot of things people have not really figured out when they think about building everything from scratch. There's a lot of IP, a lot of domain which you need to take on while you modernize the outcomes for your customers. So, that's how we differentiate, and I think we have been competing with every company every day for many years. I think the same thing we need to keep on doing. As long as we keep attention to what customers need, as long as we pay attention to how the technology is evolving, and we have been aggressive and fast, and we remain paranoid — we have to be aware of what's going on. You can't believe everything we do is always going to be right. So, we have to understand, and that's why some of these acquisitions help. We're bringing in a lot of next-generation thinking as well through those acquisitions, right? Some of these companies are new companies who have done something similar with what other startups are doing, while we, of course, have a lot of data and context around it. So, that's how I think about our competition. We have to remain very vigilant, and we have to make sure that we are building the best products using the best technology and delivering value to our customers. So, we are excited about where we are, and I think it's great that a lot of tools have become very good so that we can also move fast ourselves. Our engineering teams, our product teams, our design teams, our sales teams are using the same tools that 19-year-olds are using out there.

Yeah, I think I don't have a specific thing to go after AI-native unicorns, right? Our goal always is — I always think from a customer angle. It's not about just looking at it from like, 'Hey, what is everybody else doing?' It's important to know that, right? And we also have to out-innovate everybody to keep on making sure we are relevant and we remain important. You cannot sit on your laurels, right? And I think the humble and hungry culture has to persist inside ServiceNow on a daily basis. So, I pay a lot of attention to a lot of the new companies which are emerging. I pay a lot of attention to entrepreneurs. I attend a lot of those events to understand, and so does my team, to understand what's going on in some of these areas. While we make sure that we take advantage of the values we have inside our products and the differentiation we have, right? So, we take this AI-native mindset very seriously inside ServiceNow. Uh, we want to innovate fast. We are building things very differently. We are changing the structure of how we operate as well, in terms of teams and how they collaborate, how they deliver, the speed of delivery as well. And while in the same time, parallel, making sure customers are predictable. See, if you end up making chaos by just giving them every feature every day which doesn't really sit in the context of the business and doesn't work and interoperate, it's a great exciting thing for five days and then it goes away. You have to really do something which is longer-term thinking and longer-term outcomes, and integrates and future-proofs the companies because that's what they're paying us for — to make sure the business runs efficiently and they don't have to worry if something goes wrong, there are people who can fix it and people who can deliver longer-term outcomes, right? And I think people forget that it's not about individual features. See, features were built before as well. There was no difference. You think the software industry is not writing software? You can write any software. You can write ITSM software also, and you can write CRM, you can write software for cloud. That was always there. Build availability was there before. Now, the building has become faster and easier and cheaper. But the need for ideas still remains the same. And what you are going to do, how you are going to do it — a prompt can only be valuable depending on what prompt you give that system. And people who know the domain are the best people at understanding what we should be doing. So, that's how we compete. We pay attention to everyone. And the AI-native companies out there are doing a great job of rethinking some of this stuff. Uh, but let me tell you, nobody is ignoring that inside ServiceNow either. And we are also rethinking a lot of this stuff. And I think we have a very good army of engineers and designers and product managers and people who understand customers which we can really do a great job with as well. So, I feel very confident where we are, while we make sure that we don't ignore things out there.

No, I think the thing is you have to have a good balance across everything. One, we need to make sure our customers are successful with what we have given them before, right? So that they can continue moving and running the business. Uh, on the innovation side, it's all about keeping our road map... we used to probably in the old traditional software industry plan six-month, twelve-month kind of road maps. I think our road map has now become monthly. Because of the pace of change and the things we need to do and the new technologies which are coming out. So as you said, playing with a lot of the new capabilities — of course our teams are doing that, I'm doing that on a regular basis — and what is the art of possible? But showing that art of possible to our teams so they can take that and think about how would we do things differently. Right? This autonomous work that I mentioned to you is a new way of thinking about how do you do level-one support. There's always automation, but automation was in support of a human agent. Now we have a fully automated AI agent. Very different. It is a complicated thing to do, even though this sounds like a simple concept. Getting the right outcome is not easy. So how do we take some of these tools and make sure we are thinking and rethinking every step of the way and changing our user experience, changing our design, changing our integration, changing the outcome? And all that kind of stuff is very important. So the way we balance our thing is that one, really make sure that what we delivered to the customer and what we promised them, we deliver on it. Second, let's make sure that we are doing newer and newer things faster. And so I have carved out enough capacity in my teams to really do that on a regular basis. Uh, third, I think we have to pay attention to everything out there in the industry, like what customers are trying to do, because a lot of times customers might hear five other things. They can't do everything. Like security is a very good example. So when we used to go and pitch about AI to our customers last year, they're like, 'I mean, I agree all these things are great, but I'm worried about compliance. I'm worried about security, governance, how do I get visibility, how do I control?' And that's why we introduced something like AI Control Tower. It came out of customer conversation. Right? Because the AI Control Tower now gives that customer the comfort factor and feel of what's going on before they embark on large AI projects. Because AI can... I mean, you heard about a lot of security issues with OpenClaw. You heard about a lot of things you could do with AI agents, and AI agents' security changes every few seconds based on what you ask them to do. So that's why we bought a company called Vezza to really bring this idea of non-human identities into identity governance, so that millions of agents out there have the ability to be managed and governed through AI Control Tower. So the innovation doesn't have to be like, 'I'm delivering a new feature.' Innovation to me is how do you deliver the full end-to-end solution which takes all the different moving parts and brings it together so the customer barrier to usage goes away. And that's what we keep on trying to think about, and that's what I want to focus our teams on, and that's the investment we make on a regular basis.