 Live from Orlando, Florida. It's theCUBE, covering ServiceNow, Knowledge 17. Brought to you by ServiceNow. We're back, this is theCUBE, the leader in live tech coverage, we go out to the events and we extract the signal from the noise, I'm Dave Vellante with Jeff Frick. Farrell Huff is here, she's the general manager of the service management business unit at ServiceNow. Great to see you. Yes, great to see you, thanks for having me. Awesome, you're welcome, awesome keynote this morning. You have your baby, which is ITSM we know, but at the financial analyst meeting and you represent today's keynote, you represented more than just ITSM, which is good. But let's start there, so awesome keynote, a lot of energy, so much meat in Jakarta. Absolutely, we have been busy for sure in our IT portfolio. In ITSM, we really spent a lot of time and energy and giving back to our customer base and making sure that critical capabilities and features in ITSM have a lot of depth behind them as well. So making sure service level management, solid, service catalog, which is 99% adopted across our customer base, servicing over half a million to end users, that making sure that that's solid, and then additionally making it really easy for new customers to join on to ITSM as well by giving out of the box best practices in a guided setup format, like a wizard format, that they can within just a couple of hours stand up a brand new incident management process prescribed by ServiceNow and feel confident in what they're getting. Yeah, so I didn't realize the number was that high in terms of adoption of service catalog. What do you see for CMDB? I mean, when we first started following ServiceNow, it was mixed, because it kind of gets political. But now, today, when you talk to customers, it's like, oh yeah, that's a big initiative of ours, or we're already there. What do you see? Absolutely, I don't have the exact percentage in front of me, but I believe that it's upwards of 70% adoption in our customer base, and that is a difference from where we were in the past, for sure. Just like the main spring of innovation was once you get there with service catalog and CMDB. You get all your assets in there, you get all your services defined, it's go time. Then your operating leverage is huge in terms of when you bring out new function and the impact on the organization, the business impact can be really enormous. Absolutely. And best practices out of the box is a huge, huge cue. Everyone we've talked to, they're smart enough now to know customization is bad. Keep it to a minimum, keep it to a minimum. Do config, but not customization so that all those upgrades are easier, easier, easier. So to come out of the box with an integrated best practices workflow, great, great solution for the customers to get up and running quickly. It is, and they're asking for prescription, and we're going to give it to them. We've got our own services arm, we have a partner community, we know between all of us in this huge ecosystem what's working and what's not, and we're going to put it in the product and make sure our customers, existing and new, get best practice out of the box. So kind of three areas you talked about today, service management we just touched on. Yep. We didn't talk about the surveys, but it's cool, it's a nice little feature you guys have added. Oh yes, that's right. So you had new and improved surveys. Operations management, so that's the ITOM piece, right? And then business management. That's right. So give us the high level on ops management. I will, yeah, sure. So we announced this year that we're putting out the cloud management platform. And the adoption of cloud is long past its tipping point. We're seeing cloud being adopted everywhere and cloud resources are extremely easy to procure, stand up and use, and IT may or may not know about it. And that becomes just a huge problem in terms of cost and even in terms of security and compliance. And when we're able to, we made an acquisition roughly a year ago, the ITAP team, and this is basically the next generation cloud management platform where now you're able to have a cloud portal where a end user can go and consume and just like a service catalog, they're going to have a service catalog of cloud services that you've already provisioned very easily with the drag and drop interface that accounts for all your policy already in those services. And so it makes it very, very easy for the business to continue to operate at the pace and the scale that they need to, but for IT to make sure that we're, we have the consistency and the compliance that we need to protect the business overall and manage cost. All with a really great user experience at the same time. So we're thrilled to be able to put out the cloud management platform. And then the second major thing that came out in the IT operations management space was around service mapping. When we went to market with service mapping, it was for all on-prem services and mapping out what that looked like. This time around, we're just, you know, book ending it and kind of closing the gap and saying, okay, let's look at what's off-prem. Let's look what's in the cloud. So you get a holistic view and are able to discover resources in the cloud and on-prem as well. And you've got that holistic view of your services mapped going forward. So I have to ask you, so we're always asking, oh, you know, when service now gets into HR, it's like, oh, does service now compete with Workday? No. And when service now gets into security, it's like, does service now compete with, you know, FireEye, et cetera? No, no. Now, when you talk about this multi-cloud, sort of mapping, visibility, there's a lot of talk about, sort of, we call it sometimes inter-clouding and inter-cloud management. How far do you go into that? I mean, can I actually orchestrate across clouds? Is it just giving me visibility? Well, it's not just, but how should I think about the positioning of service now in that space of cloud management? You know, we're out there to create flexibility for our customers and we'll start to make it happen that you can orchestrate across different clouds, regardless of what they look like. We're not totally there yet, but that's the direction it's going. Well, nobody's there. Yeah. You know, this is a jump ball for the industry. And so, and it's got to be a huge market. I mean, everybody's doing multi-clouds. In fact, somebody told me today, David Floyer told me in Europe, there was a mandate in the banking sector that you have to have a second source for cloud. Oh, really? Yeah, I don't know the context, but good news for the cloud vendors. Yeah. Good news for somebody. Exactly. So, okay. And now what about, are we done with options? That was operations. Yep, we're done with that. Business management. All right, on the business management side, the big news is the software asset management. We were able to deliver another new product this year and that's really going to put a lot of power back in the hands of IT. You're no longer caught on your heels with a software audit, realizing you're out of compliance. We struggle with visibility and understanding where are all these software assets? Who are they allocated to? Are they actually using them? How much is it costing us? And when we're able to have visualization to that because it's on the ServiceNow platform and we understand where all those items exist, we're able to go in and very easily reclaim licenses or reallocate them. And to me, that's found money. And I just love that. I think that's going to be great. And guess what? You want to find your sourcing for your next IT project. It's right there. Right, right. And you're getting humble. That was the thing where the biggest roar came up from the crowd without a doubt. It was super, super well received. We were talking to CJ this morning about how it works and you got the platform. The platform comes out with all these features and then the business unit take advantage of those features. Now, of course, he described it differently. He said you start with a customer and then you figure out what to put in the platform, knowing that the business units are going to take advantage of it. When you think about intelligent automation, you gave an example of predictive maintenance today. So that's a use case for that so-called AI or just deep learning, machine learning. So talk about that a little bit and then I want to get into the DX continuum piece. Yeah, absolutely. You know, when we're sitting on this data set that our customers have and they want us to take advantage of it for them on their behalf, we're able to go back and apply algorithms to those data sets to say what's the norm and did it have a good outcome? And all that data is in there. We're able to model it now. You're not having to go do that and export that into some other system to try to figure out with some advanced analytics what's that looking like? You're able to be able to say very clearly, listen, here's what the normal pattern of behavior is and establish that for everything else going forward. So it becomes really clear where outliers exist and what suspect events or suspect alerts look like in your environment and then you can fire off a process to say, look, this looks like a problem. And with certain signposts associated to it, go ahead and automatically open up that incident. You apply it to change management where you're talking about predictive maintenance. Something has enough failures, automatically schedule a change window or decommission it, fail it over, back it out, move it out of the way so that it's not causing a problem anymore. We put so much on humans to do for so long because the technology wasn't there to allow us to do it. Well, it's time, it's here now. And so we can take some of that burden away. You know, I just had a thought. We talk in this industry so much about consumerization of IT and trying to mimic consumers. Fred Lutty talks about it all the time. Well, you just described, I thought about an experience of an iPhone user and anytime you do a migration, my wife just migrated from an Android to an iPhone, what question was asked? Is it backed up? What you just described is proactive. I mean, way beyond, is it backed up? You're at the point of, we're going to just eliminate any possibility of a disruption. So, I guess my question there is IT, is enterprise IT finally not only catching up, but in some regards surpassing this consumerization trend? Hey, I think there's an opportunity to leapfrog all the way and I'm behind 100%. I do, I think exactly that. And why not get way out ahead and over our skis with that and over deliver and show that yep, we can see what's coming. We're sitting on all this data. When you choose to go to the cloud and you're in all that data is accessible and you're on a single platform, it's all intermingled. You're not having to stitch together in a created data lake that's got all these different integrations pulling data and then try to sort it out from there with some data scientists or some business analysts looking at it. You're now able to lean in way more with your operation and really start to take care of it and truly own it. Yeah, I was just saying, my favorite part of your keynote today is kind of teeing off what you said, which is using machine learning and artificial intelligence on relatively simple looking processes that are painful, cumbersome, and horrible, like categorization, prioritization, assignment to take the first swag, let the machine take the first swag at that stuff and take that burden off the person because it's tedious, it's cumbersome, and it's painful. So it's this really elegant use of machine learning and AI which is talked about all the time on a relatively, again, simple looking activity that just delivers tremendous value. Yeah, I'm really, really excited about that part because there's a lot of mystique and I don't know what the right word is, maybe misunderstanding potentially which can lead to mistrust of AI and machine learning and what's really going to come of it and when we're able to say, using supervised machine learning which is the model that we're going after with the auto classification, you can work with customers to be able to let them tune the level of accuracy that they are comfortable with and so you're building trust right away with a really simple example of auto classification or auto categorization that is so frustrating for both parties, the person who's filing the incident and for the person who's going to be supporting and fulfilling on that incident as well and I just love that fact that we can start to dip our toe into this pool and wade in and create trust along the way so we don't leave anyone behind or create mistrust in our user base that we're just trying to get rid of them in some capacity or pull the wool over their eyes. We're not and we're going to be really transparent about it in the way we do it and I think that's phenomenal. And it's dynamic, right? So it continues to learn and if you have Spotify, you have a playlist, I like this, I don't like this, the playlist hopefully gets better. That's right because it took your input. Correct, right. And so taking input from the end users is going to then help train that system over time. So many questions for you. Okay, so the auto classification piece, that comes from the DX continuum acquisition, correct? So explain that. I know you guys re-platformed everything but what did that give you and let's get into auto classification a little bit. Well, it gave us some incredibly talented smart engineers and some really great intellectual property in terms of algorithms that we are able to now apply when we re-platform something, we are making sure that it works in the ServiceNow platform stack and that it is going to be available and pervasive for every application that gets built on top of the platform. Okay, so you said before, we're not just building a data lake, which I want to talk to you about that too because a data lake, as we know, turns into a data swamp and it's just a mess and then you got to really do a lot of heavy lifting, right? It's not good. I don't like that. So scary critters. You're auto-classifying at the point of creation, I would presume, more use of that data set. So how does that all work? How is it being applied? Where do you see customers getting value out of this? Explain that a little bit. Well, really I see in the ITSM side and the IT space and in the ITSM side specifically, anything that you're trying, anything that you've got to do, apply a drop-down field to, whether you're an end customer doing it through a service portal or you're an IT worker too. Like let's help those guys out. Why not? Anytime you need to fill out a field and through a drop-down mechanism, it's one discreet set of values, that's a candidate there. Now you want to have a large data set which is why incidents, incident category or assignment group or what skill set might be required to work. That particular incident works because there's tons and tons and tons of incidents out there. So we have lots of examples around what it could possibly be and then that's what the data model would be built on. You're not, this auto-classification is not meant for the obscure or the random or the infrequent. So when we're talking high volumes out of service desksies, this is the perfect setup to apply it. So how will it work? I'll have a corpus of data with a bunch of incidents and I'll just sort of tell the machine, go classify this and it will do some kind of process. You're going to take, you're going to have a set of data, a portion of the records you're going to use for the training model. The other portion you're going to leave behind almost as the control group and you're going to go apply the algorithms to that training set of data and it's going to start to learn and you're going to tell it what fields you want it to learn from and pay attention to and spit a model out on the other side on and it's going to crunch through all that data and it's going to give you a model on the other side and you'll look at it and see if you agree and then you're going to take that model and you'll apply it to that control set and you're going to look at what level of accuracy came out on the other side and you'll decide with that data set whether or what accuracy level you want to have. For me, I really probably 70% accuracy will work for me on password reset because in all likelihood, what's it going to be? But maybe for a VPN issue, I want 90% that you'll be able to start applying accuracy by category to then tune in exactly how you want things to work to make sure you get that good user experience. And then you'll continue to train that model and iterate. Absolutely, and you'll be able to train it as often as you like. I mean, on demand, like, yep, I want to train it again. And when you have a service desk worker who goes back in and recategorizes because yeah, that wasn't quite right, that's just the same thing as clicking the like button, thumbs up, thumbs down on Spotify. You're right, you've just given it feedback. When you train it again, it takes that feedback into account. And then that incident, the subsequent incidents they get the learning, they get the learning. There's not magical learning that happens. In this particular case, the technology's not evolved to that state. There's no unicorn back there that's doing all the learning for you. It takes feedback and it'll take some tuning. But hopefully in being able to make the feedback mechanism very easy, the tuning happens naturally, therefore the model gets better over time. Well, that's a great use case because it's relatively narrow. And you have tons of data and it can be implemented right away. And as you said, even if it just helps you partially down the road, it's better than zero down the road, especially in these repeatable processes that have to happen over and over and over. It's like, oh, please shoot me. This is the work that machines are supposed to do because it's mundane and repeatable and, you know, let me thank you. Let me get to solving the customer problem. Okay, so when we first encountered ServiceNow, we did our first knowledge, it was from 2013, and it was at the height of the big data, sort of hype cycle. And so we would ask, of course we asked, well, what about data? What about big data? And the response was always, well, we got a lot of data and we're looking at that. But now we're here and you mentioned earlier, it's not some data lake that you're processing is offloading your data warehouse. So what are you doing in that space? So it's not a data lake, it's a corpus of data and you're basically applying these AI and intelligent automation models to, can you explain a little bit about how that works? Well, first off, we won't do anything, you know, we have to have our customer's permission to be able to use their data. They show interest in machine learning services, then they will give us permission to leverage their data. And, you know, all customer data is separated to within their own instance, within their own database. There's no co-mingling of data. So there will be no data lake whatsoever. But what we are able to do, and it's on a personal level, which I just love, because that's who we are as a company, that we're offering personalized supervised machine learning, personalized auto classification. We're not taking all the data of all of our customers kind of aggregating it up and then building models against that and then saying, oh, I think this model would pertain to you. And then it's only 25% accurate or even relevant. We are building a model very specific to you and working with your data set and we have access to it with your permission and we'll go build that model using the training set as we described and then, you know, go test it out and then help you go redeploy it. So you're going to, you know, we'll pull that data into a central instance, help retrain it and then move it back into your instance so that model is always constantly tuned and then you get to decide when you retrain it. So who's we in that example? Do you have a team of data scientists that do this? This will be in our platform team. It's a platform service. You don't need data scientists to, you know, I would say on the customer side, maybe if they were wanting to interpret some of that data or do something with it, maybe they'd have a data scientist. This is just tried and true engineering and being, having a good service model behind it. It's just the central instance. Do you, I'm sorry, interrupt it. No, I was just going to say, the, you know, through our acquisition DX Continuum, those engineers are building those training models and we'll keep them up to date, but they're not literally, you know, turning a crank when that data comes in. So it's a model that they apply, it scales, it's part of the service. Now you iterate that over time. That's right. But it's the. And you can build out other training models. So we just talked about, you know, auto classification for incident, but this can extend in other areas as well. Well, I was going to say, do you think it's an opportunity for the ecosystem that has specialty expertise around, pick your favorite topic here? You know, we were talking to, something about oil and gas earlier today that, you know, they know what the model is way beyond just simple correlation to take an incident flow and predict that, you know, I think the example is the well caps going to break or whatever. You know, do you see that potentially as an ecosystem contribution as well around more specific use cases? Well, I think that would be super cool. You know, if we had customers of similar ilk, whatever that looked like, wanting to collaborate and share and crowdsource something for a greater good that wasn't competitive. I think that that would be amazing to be able to do that and we would be able to facilitate it. We don't have any current plans to do that right now, but I could absolutely see it. Well, we've talked about the ecosystem now for years to see it just burgeoning. And awesome story. Thank you for coming on theCUBE and doing the brain dump on us and educating us. Thank you so much for having us. Great opening line. Exciting time to be in IT. That was your opening line the key night. I know you got the excitement. That's time to be in IT. I mean, oh my gosh, it's fabulous. You're exploding. All right, Farrell, thanks very much. All right, thank you. Good to see you. All right, keep it right there. We'll be back with our next guest. We're live from Orlando. Be right back.