17. The Analytics & AI Landscape

Speaker 0 00:00:02 Welcome to data points, a podcast by InterSystems learning services. Make sure to subscribe to the podcast on your favorite podcast. App links can be [email protected]. I'm Derek Robinson. And on today's episode, I'll chat with product managers, Carmen lobe, Benjamin Devo, and Thomas Dyer about the overall analytics and AI technology stack at InterSystems and how these products fit together. <inaudible> welcome to the data points podcast by InterSystems learning services. On today's episode, we have a bit of a round table discussion for you with three guests that have each appeared on prior episodes of data points, Carmen lobe, Benjamin Dubbo, and Thomas Dyer. These three are all product managers or specialists for different product and feature areas within the analytics and AI part of the InterSystems technology stack in our conversation. We'll talk about what technologies exist in this area, how they fit together, how some customers may use these features and more. Alright, so Carmen Benjamin and Tom, thank you all for joining us on the podcast today. Why don't we start by introducing yourself? I know you've all been on the podcast before in past episodes, but, uh, Carmen, we'll start with you. Speaker 1 00:01:24 Sure. Good to be here again. Uh, yeah. My name is Carmen Logan. I'm a product manager on the, uh, database platforms team here at InterSystems. And my focus is really on the kind of business intelligence reporting, um, aspect of things so that, that set of users. Cool. Benjamin. Hi and thanks for having us. So my name is Benjamin <inaudible>. I've been with him for 10, almost 11 years, and I look after, uh, within the product management team to look after data management subjects. Cool. Yeah, my name is Tom Dyer and I'm a product specialist and I work in the same team, Carmen and Benjamin, but focusing on machine learning and AI within our database platform product. Nice, Speaker 0 00:02:13 Great. And so thank you all again for joining us. And of course we're, we're still doing these remote interviews after all this time, but hopefully before long, we'll be able to have a real round table in person again, um, in our office. But so, uh, starting with you Benjamin let's, uh, we're gonna be talking about our AI and analytics offering, uh, within the InterSystems technology stack today. So, um, let's start out with just telling us a little bit about, give us a little overview of the AI and analytics offering that we have. Speaker 1 00:02:38 It'll be one big overview today. It let's, let's start with the beginning and maybe a little bit of a history because in a different life, I would have been asked to be the teacher because we really evolved quite a bit over the past, say four or five years. So, and up, um, that has been, uh, an activist, uh, an active process because we really started thinking about analytics and AI in a different way. So previously again, say up to four or five years ago, we typically work closely with, uh, development shops with application partners that develop their applications and we embedded select, um, analytics capabilities inside of our platform so that they could take advantage of that in their application. So I would, audience was largely developers that then built something for their customers and those, those customers that went all the way from, from hospital information systems, lab systems to, uh, outright carpentry coordination systems. Speaker 1 00:03:36 Uh, but now we've, we've really started to cater more directly to, um, to a much broader audience of end customers that includes, uh, fill developers, but also data scientists, data engineers, uh, analysts, and also, uh, totally non-technical people. So that means that we've tried to really get under under the skin of these, these people and, and try to understand how they think what they need and make sure that we, we have the solutions or we have the technology in our platform to serve those people's use cases. And that very often meant that we went, uh, higher up the stack that we have to offer more, uh, visually appealing, uh, more, um, clients, uh, kind of tools. And you could think of that as camping to stock. And so that's, that's really how we've been evolving our platform from a couple of, uh, embedded technologies that were really closely embedded, uh, close to the data at very high performance, very robust to a more broader set that is still of course, uh, made to be very robust and very dependable, but then we're also catering to a lot of its user types directly. And we think that also benefits these application partners because they now also have a broader set of options to offer to their customers. Speaker 0 00:04:53 Right, right. Nice. And so when, when you look at kind of building that out to be for the users really end to end an offer, everything that they need, um, does that mean that our partners and customers really kind of have to stick to the selected solutions that we offer or is there more that they can choose from? Speaker 2 00:05:08 Oh, they, yeah. They don't have to stick to what we offer at all. Um, they can, they can, you know, use our choices, particularly we have some customers who like to, you know, have one organization that they work with and have one set of support, you know, support number and that kind of thing. But if you've, if you've followed inner systems over the last years, you know, you've probably heard the term freedom of choice, which, um, is something that we've, we've used in a number of ways, but certainly the goal was to provide our customers with access to the tools that they want to use. Um, so it's kind of a, we call it a native plus open approach, which is having somethings that are embedded in the platform developed here at inner systems, like our, uh, InterSystems Iris business intelligence, our OLAP engine and NLP tool, which are embedded into the product, but also to, to make sure that we can support other tools by supporting key standards like MDX and PMML and EMA so that we can be inter-operable with those tools. And, um, so yeah, so that's, that's really the goal there. And I think as, as Benjamin said over the last year or so, we've kind of, uh, looked up or moved up the stack, I guess, to, um, looking at how the business users are interacting with our technology, something that's, that's still relatively Newell new, but things like InterSystems reports where it's getting information all the way to the end user about, you know, what about what's happening with their data. Speaker 0 00:06:41 Right, right. Nice. And I think, um, you know, us in the learning group have experienced this as well. Like as we have progressed our technology stack over the years, right. It, it becomes, it can become a lot to keep track of like, what is, what is in each umbrella of the entire stack. Right. So, um, I know each of the three of you kind of have your products and features that are in your specialty areas. So, um, Carmen starting with you kind of, just for the users that might have been exposed to these different features and products within InterSystems, they've seen it on the learning side or the corporate site, um, what features fit into this AI in analytics umbrella, at least within your area. And then we can also ask Benjamin and Tom the same question. Speaker 2 00:07:20 Yeah, that sounds great. And yeah, I guess we have made things a little, a little more exciting for you guys. Haven't we, by giving you more to talk about? Um, yeah. So I mentioned at the start that, you know, my, my focus is on kind of business intelligence and reporting and that sort of stuff. So, um, so I think of the kind of users that I think about our analysts and data modelers, um, all the way to the business user, who's interacting with something like Tableau or power BI on the desktop. Right. Um, and so these people are often, um, they're kind of labeled as non-technical, but in some ways they have to, you know, they have work to do to understand the data and be able to visualize it in a way that makes sense to their end consumers and the people who are making business decisions with it. Speaker 2 00:08:07 So, so the things that I look after are, um, InterSystems Iris business intelligence, or BI formerly known as deep sea, um, InterSystems reports, which has now been available for about a year, um, third-party, uh, integrations with, uh, power BI and Tableau and that sort of thing. And then very soon I'm very excited about that with, with our 20 one.one release, we will, uh, be including adaptive analytics, which, um, is, uh, basically a, a layer in between InterSystems Iris and those BI tools to allow you to do some common data modeling and, uh, query acceleration. So, yeah, so that, that's kinda that's, those are the things that I look at. Speaker 0 00:08:51 Nice, nice. That's great. And we do it. I think we had recently some adaptive analytics coming out. I know some of my colleagues worked on that in the learning groups. That's great. Um, Benjamin, what about yours? Speaker 1 00:09:02 Well, I I'd like to think more about a second category of users that are not, um, that are typically more, um, more technical folks. So it's really data professionals, uh, task data scientists, data engineers. So people who are really very much hands-on with the data itself and are, um, the end consumers of the, of the insights. So maybe they, they, they built things for the personas that Carmen was, uh, was talking about. For example, could be that they're building a predictive model or they're building a particular subset of the data, data Mark group, world, or a data pipeline that feeds the data warehouse. So these typically rolled up sleeves kind of folks. They would, it would be using capabilities such as our, uh, our interest in some Cyrus, text analytics capabilities, the Apache spark connector to access data that's stored on Iris from Apache spark. Speaker 1 00:09:55 Um, some of the stuff that we've been adding recently to our SQL engine, also this, this, this group of users. Um, and if you, uh, if you happen to have watched the session that we presented that, uh, virtual summit last year, about our integration with Alteryx, that's another example of, uh, an integration or a combination of products that cater specifically to these data professionals set up. Um, although, um, we specifically work with Altimax to, um, offer non technical people or less technical people, also the ability to get really hands on with, uh, with complicated data. Speaker 0 00:10:34 Right, right. Nice. And so it sounds like in that answer, you sort of, um, begun the transition to what I'm about to ask Tom about, which is sort of a, that AI and ML part that we haven't really gotten into quite yet. So I think Tom, that's your area, tell us a little bit about the products that are in that, that spot of the umbrella basically and where the impact is there. Speaker 3 00:10:53 Yeah, thanks. So, um, you know, one of the, um, groups that we're looking at to, to kind of enable with this AI and machine learning are even our more traditional kind of, uh, audience, that it could be developers that know SQL know the data, but they're really trying to just use the insights from a predictive model within their applications or workflows. And for that they can use integrated ML. It might be a model that was developed with a lot of input from the data scientists and the other groups that were, um, you know, talking about here, but then when they actually want to use it integrated ML with SQL makes it very simple to use that model. And we support PMML as well in that workflow. So even if that model was developed in a completely different data science environment, you can use it within these, um, you know, mission critical applications that are your everyday data flow that's that this kind of, that developer audience really needs to, to rely on. So, um, we're going after, um, you know, AI and ML from a very practical viewpoint to try to overcome some of the challenges that people typically have in actually developing an application that uses a machine learning model. Speaker 0 00:12:14 Right, right. Nice. And, you know, I think that speaks to the, you know, evolution of this entire kind of umbrella of products and features, you know, capping it off with that really all the way through the machine learning and AI element of it, making it accessible to people that weren't previously experts there. Um, Tom kind of, uh, following up on that a little bit, can you talk about the benefit? So, you know, we've, we've run through these features, right? Like what is really the benefit for somebody to use InterSystems stack of analytics and AI features and products, um, versus using their own things like we talked about earlier with third-party tools and everything, like what's the benefit of, of really basing your analytics and AI approach around the inner system stack? Speaker 3 00:12:52 Well, as Carmen mentioned, you know, there's that real fundamental aspect of us being open and also providing some functionality built in, we're also doing, uh, you know, getting back to integrated ML we're, uh, pre configuring a lot of frameworks that, that are conflicts to set up and, and, and install and, and work together with all of your other, uh, components of your larger system. So that know-how, and then supporting that with like a single phone number and a consistent, uh, view of our customer can, can really make the difference in, in making it easy to work, work with us. Uh, we hear that in a lot of different metrics, uh, that, uh, from our customers. And so that's a really nice, uh, thing about working with us and also working for inner systems as an employee. Uh, it's unique. And another thing I wanted to mention, uh, for that developer audience, for people that are building systems, uh, we're, we're, we're coming out with embedded Python support within our platform so that people will be able to write server side and, and, and very complex applications just using Python and being able to go back and forth between our native, uh, object, script and Python code, uh, as we roll this out in the future. Speaker 3 00:14:12 So that's an exciting thing for people that are, that are really trying to, uh, have one central, uh, kind of application development environment. Python is a critical, uh, part of, of many ecosystems and provides a lot of functionality, a lot of different libraries. So it's important that we support that. Speaker 0 00:14:31 Yeah. And we've actually had some discussions, I think embedded Python could be a future episode, um, as we look at, uh, the development of these products and features. So we've had some discussions about that, for sure. Um, kind of building off of one of the things you said about being, you know, one of the advantages is being open and being able to connect with other technologies. And so I want to revisit Carmen, your area, the product and feature umbrella, which, um, a lot of times, as I understand it, with some of these newer capabilities that are higher up on the stack, they can sometimes be partnerships versus actually homegrown technologies. Can you talk a little bit about that difference in kind of where the benefit is for a user to be able to understand what's available between each of those options? Speaker 2 00:15:10 Yeah, sure. And, and that's, uh, yeah, that's an, I think an important question because, you know, I think as we extend into other parts of the organizations that we work with, um, you know, this whole idea of, of open and having embedded capabilities, um, is important. And I think as we extend, we, we broaden the amount of work that our team needs to do. Right. So in terms of building products, um, the other side of that is that, or the other aspect of that is that we know that our customers have tools that they're already using, that they want to keep using. So, so that's, that's important too. Um, but I think that, you know, the, the decisions that we make around when to build and when to, to partner, um, are largely based in trying to get a deep understanding of what the needs are of these end users and, um, also kind of building on our strengths. Speaker 2 00:16:05 So InterSystems has a lot of key strengths things that we've been working on for decades. And we want to continue to emphasize those while we add capabilities potentially more quickly with, with partner products in areas that are not key strengths for us, like reporting, right? Like I think that's a, that's a good example or API management, right? We have the, that's been a very successful partnership, um, with, uh, you know, the, that we've now, you know, has been around for a couple of years. And I think, you know, the reality is we've always done partnerships. They just probably haven't been quite as visible as they, as they are now, but it does really allow us to extend, extend the platform, get things to market more quickly, but still maintain, um, you know, the ability of our customers to count on in their systems for support, for making sure these things work together. Um, and for, you know, looking at, you know, looking at things in the long haul. So it just, it, it simplifies for a lot of our customers and partners, the decisions that they have to make about which solutions to include. Right. So, yeah, definitely not decisions that we take lightly. This is something that we, we really try to think about where our strengths are and, um, you know, what the end users really need. Speaker 0 00:17:23 So definitely. Yeah. Speaker 1 00:17:24 And also, where does it fit with the older, uh, with the other party? Because we look at these in isolation, um, each time, not just that they're technology, but also if there is a business fit, if, if these are people that are in it for the long run, just, uh, just like us, uh, this isn't, uh, this is really focused on do our, we do being the target users a favor by offering as a combined package, right. That's really what we're aiming for with these partnerships. And I think, uh, API management, uh, reports have been great examples of how we can, uh, take this combine the strengths of our data platform with a particular complimentary, uh, capability that our customers, our customers that we know well, uh, we're looking for. So those are really good examples where it's been working excellently. Speaker 0 00:18:16 Yeah. And just to add one more, and I know that that is in your ballpark, Tom is like our partnership with data robot, for example, where, where, you know, another partner has, has good experience and, um, expertise in machine learning. And we can roll that in with the service and products that we offer, right? Speaker 3 00:18:32 Yeah, absolutely. Uh, data robot is a category leader, a creator of a whole category of auto ML for enterprise. So they're, they're really, uh, uh, they know their stuff in machine learning and we have a seamless integration, so, and many different ways that we can support getting data that people aggregate normalized in our platform and then utilize, uh, get insights, drive predictions, using data robot machine learning models that they can create. So it really can work well. Speaker 0 00:19:07 Right. For sure. So, um, looking at kind of all of those partnerships and technologies that we've made ourselves, um, it can be a lot to, to look at everything that's available. Right. And I know Carmen, you had mentioned in your last answer there kind of identifying what solutions are right to use for a given person. So, um, Benjamin kind of looking at what a typical customer might be using, are they using everything that we have, or are they using a subset kind of one thing, like what's, what's the typical usage for a customer, you know, within this entire umbrella of analytics and AI products and features Speaker 1 00:19:40 It wouldn't be great if there would be a customer that uses everything. Uh, but it's, it's like walking into a home improvement store. You don't need anything. Uh, you don't need to think from each and every athlete, you may think that you need something. Or if I think that I need something from every alley, um, but usually it really comes down to the scenario that they're off. So if you are a supply chain management application developer, and you want the real time view on where your goods are, then InterSystems, Iris BI is the perfect solution for that because of the, of interaction with, uh, with our persistent data. Um, it will, it will give you exactly that if you are a big investment bank who needs to cater to a very broad audience that uses different, uh, different tools already, and you really have this huge data data volume that you can't possibly have indices for each and every dimension then InterSystems Iris adaptive analytics would be a terrific, uh, solution for that kind of use case. So it really depends on what you're trying to achieve. Then of course, the broader our footprint at a customer, the more likely it is that they'll be using multiple tools and we've made that they work well together, but it's, uh, it, it really depends on what you're trying to achieve. Speaker 0 00:20:52 Right, right. So, yeah. So I think certainly, you know, customers leveraging as much as they can and being able to maximize the benefit that they're getting out of, you know, their situation and the tools that are available to them. Um, Tom, I know you've worked with a particular customer who does this really well, if you want to share briefly about that. Speaker 3 00:21:08 Uh, yeah. We work very closely with, uh, Bay state, uh, health, which is a large healthcare provider in the Western, uh, half of Massachusetts. And, you know, they are leveraging a whole, uh, host of our technologies with aggregating and collecting healthcare data from millions of patients. And then being able to take that data along with information that they might have about their patients, not only their history of all that, their medications and, and encounters in the health, um, in, in the health system, but, uh, uh, whatever data they can bring together and try to predict things about care, about cost of care there. And also in this time of COVID being able to, uh, optimize their operations wherever they find some have built some, some needs and having a platform that can put the data together and then also, uh, develop machine learning models for, for these, uh, these kinds of applications is really critical for them. Speaker 0 00:22:10 Right, right. Nice. That's awesome. Um, so looking at kind of everything we've talked through in all these great products and features, um, I wanted to end with a quick kind of, you know, around the table here, where do you see things going next? And, and what's the future within this area of the product and the technology. Um, so Carmen, we can start with you. Speaker 2 00:22:26 Sure. Yeah. And I'll give you a very biased to answer because I'm really excited about what's coming out with adaptive analytics. And so I really think that, uh, this idea of being able to use the same platform for, um, you know, transactional and analytic data work workloads is going to be something that people that are customers will really be able to leverage. And, um, you know, and it will, I think show them good, good cost of ownership as well as additional capabilities. Speaker 0 00:22:56 Right. Nice Benjamin, your thoughts, Speaker 1 00:22:59 We're working on making analytical querying foster for you. So we're really excited about the preliminary results received. It's really an order of magnitude kind of, uh, kind of improvements for, uh, for these queries. So, um, promise a specific version number just yet, but stay tuned, uh, probably around virtual summit. We'll have a lot more to say about this. Speaker 0 00:23:20 Nice. That's exciting. And Tom, what about you? Speaker 3 00:23:22 Yeah, I think for integrated ML and for machine learning within our platform, we want to support more of the variety of data, different data types that you can store and, and be able to learn about as well as the volume. We want to be able to support ever larger collections of data. So those are two very important Speaker 4 00:23:46 To aspects that we're going to be improving and increasing our capabilities going forward. Speaker 0 00:23:52 Great. So Carmen Benjamin and Tom, thank you so much. It's been a great overview. I think of the entire umbrella here that we're looking at. So, uh, looking forward to deeper dives into these in the future. So thanks for joining us. Speaker 4 00:24:02 Thank you. Thanks for having me stark. Speaker 0 00:24:09 So thanks again to Carmen Benjamin and Tom for joining us for that discussion. Hopefully that provided a little more context around the products and features within the analytics and AI domain here at InterSystems. For more information, you can always refer to the main InterSystems website, intersystems.com for information about product releases for deeper dives into these technologies. Keep an eye on learning.intersystems.com for new materials on these topics. As we develop them, that'll do it for episode 17. We'll see you next time on data points.