 Live, from Las Vegas, it's theCUBE. Covering Informatica World 2019. Brought to you by Informatica. Welcome back everyone to theCUBE's live coverage of Informatica World 2019. I'm your host, Rebecca Knight, along with my co-host, John Furrier. We're joined by Stephen Guggenheimer. He is the Corporate Vice President of AI and ISV Engagement at Microsoft. Thank you so much for coming on theCUBE. Sure, thanks for having me. So one of the things that we're hearing so much at this conference is data needs AI, but AI needs data. I'm wondering from your perspective, AI engagement, where do you come down on this? What are you hearing, what are your thoughts on that big theme? Well, data is the, some people say the oil for AI. You pick your terminology, but there is no AI without data. The reason that AI is such a hot topic right now is the combination of sort of compute storage and networking at scale, which means the access for developers and data scientists to work with large sets of data and then the actual data. If you don't have data, you can't build models. If you can't build models, that's what is the definition of AI. So you need data. I always, all the coaching I do is about sort of BI before AI. If you can't actually get insight out of your data, let's not try to add intelligence. You can't get insight out of your data. That means your data is not in a good, your data state's not in order. So data first. A lot of architectural work's being done on data. See horizontally scalable cloud gives a nice access to a lot of different observational data sets. Used to be, give it, got a silo, got the data, go get more data, slower. Now data feeds the developer process because SaaS business models have been proven that data and SaaS work well together. So how do we get more, how do we get, what's the sequence of architecture to usability of data so that not only can you just have analytical systems, but where developers can start building their SaaS apps with data? Yeah. I mean, we have this notion, we often talk about sort of blades or feedback loops. There's sort of four or five things most companies do. You work with customers, you have employees, you have a supply chain or some type of partner chain. You run your finance and operations. So the question becomes in each of those processes, there's data. Human generated forms over data or pick your loop. And now you're getting tons and tons of data. The trick now is to make it reusable. Mostly what we've done for years, form over data, take the data, form over data. And what we do is we get all these different databases. We try and create some layer that brings it all together. We build cubes out of it to view and then we get this hopeless spaghetti. So the trick right now, we're working on something called common data model, which others are well or common data service. Let's get the entities lined up from the very beginning. We've worked with Adobe and SAP on the open data initiative. Let's start at the core. Let's make the data layer reusable. We're, you know, databases have become data warehouses have become data lakes. We're heading towards a data tidal wave. And if we don't get the data estate in order to run the line of business applications, to feed all the things we do to use the ML and AI on top of it, we're going to drown in data and not get what we want out of it. So architecturally, I think about the common data model and common data service, both generically by industry. We build accelerators for that. Getting the big organizations like the three I mentioned aligned around that, making it such that any, you know, organization can build from that and then building on top of that. For big companies you have to decide what do I keep and what do I throw out? You know, what do I just give up on and start from fresh? What do I actually clean? Where do I use tools from Informatica or others to help me clean it, secure it? But you got to put all that thought in. You know, we were talking before we came on camera about the internet days and new storied history that you had as Microsoft. And during the internet search was the big application. And search on the internet actually worked really well because they didn't have any legacy. Right. And people tried to crack the code on search inside the enterprise. Much harder problem because of the database things you mentioned. How does today's enterprise get the benefit of SAS as if they were a cloud native SAS with the data? So, you know, the challenge we're hearing here is having a common data model is all great but I just want to be a SAS player. I want to use my data to feed into my business value. How does a company move out of those legacy constraints? What do you see as a- Well, there's different paths that different companies will take. I mean, the good news is if you get your data in order to what you said, then whether you build by or partner for the SAS services you can use that data underneath and you should be feeding it back in and making it such that it's sort of reusable and the pipeline is consistent. The truth is in all this, it's just going to end up infused anywhere. You use the internet, which is a funny analogy because I remind people, you know, when the internet came out, we had internet products, we had internet events, we had internet shows. You don't have any of that anymore. It's just woven into everything to do. AI is going to be the same. You have all this hype right now. You have AI shows, you have, you know, AI groups. The truth is in 10, 15 years, AI is just going to be woven into everything. The data is going to be set up for that. So what's the misconception on AI? Because first of all, I love the fact that AI is hyped up because my kids love it. Machine learning, they learn because they hear about AI and they hear all this coolness. So machine learning goes hand in hand with AI. You feed machine learning. Machine learning feeds the AI application. But a lot of people have aspirations around AI. Some of them are ungettable. And so that's probably a misalignment around the hype. What's your feeling of where the reality is and what's the misconceptions? How should people approach AI? Any thoughts there? I think a lot about the AI journey. The first year we were having these AI conversations, we talked about AI for everybody, just go play. Now the conversation, I call it pragmatic AI. Look, let's talk about, you know, how you want to think about AI. It's going to end up everywhere. So the question becomes, what's your differentiation as a company? And how is AI going to support that? Like any other new technology, in the beginning people just want to play. Just because you can, let's just say, just because you can build a virtual agent, doesn't mean every company should. So the question becomes, first off, be AI before AI, get your data state in order. Second, in a build-by-partner model, what's your differentiation as a company? Where do you want to use either your unique data or your unique skill sets to use AI against that differentiation to help you grow? Otherwise, like expect somebody else to have infused AI into the products you buy, the SaaS services, and they use that. Then build where you want. And then there's, you know, if you think you're going to build a new business based on your unique data or your unique AI capabilities, great, let's have that conversation. We can do that too, but rarely does that become the case. So most of the conversations move from, you know, the hype to, okay, let's get pragmatic, which is why I always come back to data first, because if you're not doing that, you're not setting up for the long run. Let's build for the long run. Then let's just have a business conversation, like how do you differentiate yourself as a business? Okay, how is this tool going to help you? I want to ask about innovation, and particularly because Microsoft is a company that's now entering its middle age, and... What does it say about me? Oh, no. But as one of a famously innovative company, but how do you stay on the cutting edge? I mean, I'm wondering internally how you think about AI for Microsoft's business purposes. What are the conversations around AI? And I would say just one of Saatch's core conversations around this notion of tech intensity. You know, from where we focus and how we think about things, we think about tech intensity against different areas. AI being one of those. Look, AI is this really interesting thing. I always say we're plumbers by trade. We build software plumbing for others. So we do three things relative to AI. We're basically, there's a layer growing on top of the core development stack, compute, storage, networking for AI. So we're building a layer, cognitive services, a bot service, a machine, a set of tools for developers to infuse AI into the things that they build. So that's thing number one. Thing number two is we infuse AI into our own products, into Windows, into Office, into Azure, into Dynamics. You don't see it. We don't talk about it. We don't say Microsoft Windows inking brought to you by Azure AI just works better. Oh, inking works. Oh, FaceLogin works. Oh, you know, it's helping me write a better resume in LinkedIn. That's all AI behind the scenes. The third thing you think about then is how do you actually use AI to run the business better? So how do you think about AI assisting professionals? How do we think about how we do marketing better? How are forecasting sales? So AI is about plumbing. Let's build the platform for others. Let's use it ourselves on our own products. And then let's think about how you actually use it to run the company better. And that's how we think about it. That's pragmatic. Very pragmatic. AI is kind of your getting. Yeah, that's how I think about it. And we, you know, it's interesting because back to the tech intensity point, we get together on an AI conversation with Sachin with the senior leadership team about once every other week. And we'll round Robin between a research topic, the platform and one of the solutions so that you're always getting constant feedback about is the platform doing what we need to build solutions? Is the research feeding the platform? So you're getting this really nice feedback move right now in that tech intensity. Quality data always has been a big part of the data, data modeling in the past. Cloud now allows for data marketplaces. You're seeing sharing of data as a dynamic, almost like sharing libraries if you were a developer back in the day. So data is now being in merchandise in a new way. This is a trend. What's your thoughts on it? Because if this continues, you're going to have more data inputs. Is that, is that true? There are places where data is aggregated and potentially can be reused. We can, Bing is an example. Google would be an example. I know people who aggregate data for different industries, et cetera. It's not an easy business. The rules and rights around data, the GDPR compliance, the rest of it, I think there's a there there, but you really have to, you really have to be in the business for it. And the trick you run into is if you're going to be an aggregator and then a reseller of data, where's that data coming from? What are the rights? What's the security? And then are the people who are providing that data comfortable with the competitors getting the data? Because if you're really going to be a data provider marketplace, the first person it's going to want is the competitor. So I think it's an interesting conversation. I think it's kind of growing and there's some real good work there. I don't think it's as easily to do it at scale for as many people who think they have a data asset as believe they do. But that's Steve's view. That's not a Microsoft statement. Good disclaimer. Steve's view. So I want to hear Steve's view on the skills gap. This is a huge problem in the technology industry. So few people to fill roles. How is Microsoft dealing? What's your view on the problem? I'm glad I work at Microsoft because we spend a lot of energy on that. And I wish there were a single solution, but we have Minecraft for education starting with kids. How do you help? Minecraft is a great tool that teachers use helps kids get started. So that's a tool set. We work at something called TEALS, which is basically our developers teach kids remotely more at junior high or high school coding. We have made investments against this. We have online training. We work with universities. I don't know the perfect answer, but I do know we invest it. We work with Hadi Portovi and his group on code.org. I mean, any place that there's work going on, we work with the military for people coming out of the military service. So we're heavily invested. I'm hopeful that the ease of use of some of the tools, and just from a job area, it drives people, but I don't know the perfect answer. I have Steve's view is I don't know the answer. I do know we try every trick in the book. Multi-pronged attack. I'm a parent of two kids. Like I had my daughter working more on the tech side, and it's hard to keep kids on a track for that. There's no degree yet. Well, Berkeley had a first degree this year, graduated from the school, but this kind of a skills portfolio of different things, depending on the makeup. I mean, domain expertise is critical. If you don't know what you're trying to do, that's gonna get a code. I think we'll get a mix. Because what you're starting to see is the tools for subject matter experts are getting better. We have something called the Power Platform, which allows people who aren't necessarily coders by trade, but want to be able to build sort of apps or services to be able to do that more easily and mix their subject matter expertise. And you're seeing many more people come out of any program, take biology, with sort of computer knowledge at a decent level. AI and ML research, so different area, hard skills gap right there. Steve, great insight. Thanks for spending some time with us. Great insight, some of the skills gap in just overall. Thanks for coming on the team. We didn't talk about rugby, but okay, fine. Thanks. Next time. Oh, those bozemen. We'll track you down. The bozemen cut through us. Exactly. Shout out to them. There we go. Thank you. You are watching theCUBE. We will come right back with more from Informatica World. I'm Rebecca Knight for John Furrier. Stay tuned.