The Future of Enterprise Low-Code

• Low-Code Development

• Transcript

  [00:00:00.299] (upbeat music) [00:00:13.550] <v ->Hi, I'm Sheryl Koenigsberg,</v> [00:00:15.700] Director of Global Product Marketing here at Mendix. [00:00:18.620] And I am thrilled to be talking with Gordon Van Huizen today [00:00:23.050] about the future of low-code. [00:00:25.160] Gordon wish we were both in Rotterdam, [00:00:27.540] but we'll have to settle for Boston. [00:00:30.175] <v ->(laughs) Settle for Boston.</v> [00:00:31.290] Yeah, I wish we were there too. [00:00:32.940] I haven't been to Rotterdam in gosh! [00:00:35.270] Since February. [00:00:36.810] So I'm Gordon Van Huizen, VP of Platform Strategy, [00:00:39.360] which is why I spend a lot of time in Rotterdam. [00:00:42.170] But great to be hanging out virtually with you [00:00:45.030] this morning Sheryl. [00:00:46.510] <v ->So let's get started.</v> [00:00:47.540] We have a lot to talk about. [00:00:49.720] You have been in application development [00:00:52.060] for nearly your entire career. [00:00:53.950] And yet we've agreed it seems as if low-code [00:00:56.720] is having a moment right now. [00:00:58.540] Why do you think that is? [00:01:01.140] <v ->Well, that's a great way to put it,</v> [00:01:02.370] having a moment for sure. [00:01:03.810] Well, to begin with, [00:01:06.280] I guess it's fair to say that low-code has been [00:01:09.730] in the limelight for the last several years. [00:01:11.990] It's become a very identifiable component of digitalization. [00:01:18.390] And it's reached the point where low-code [00:01:20.190] is perceived by many to be the future [00:01:22.320] of application development overall. [00:01:24.711] So it was ready for primetime. [00:01:27.170] I think what's making the focus [00:01:29.720] on low-code even more dramatic right now is digitalization, [00:01:34.100] on the one hand, organizations are pushing forward [00:01:37.350] with digital strategies right and left [00:01:39.770] to ensure that they can compete within the markets [00:01:42.290] that they're in. [00:01:43.290] But the world that we find ourselves in across the globe, [00:01:47.490] with the pandemic, [00:01:48.860] has only accelerated certain aspects of that, [00:01:51.600] organizations have to find new ways [00:01:54.460] to work with their customers, work with their partners [00:01:57.880] and to collaborate internally. [00:01:59.590] And that requires new forms of digital solutions, [00:02:03.030] which in turn, requires experimentation [00:02:06.150] and getting solutions out [00:02:07.550] into people's hands really quickly, [00:02:08.890] which are of course the things that low-code [00:02:10.460] is particularly good at. [00:02:12.540] So we're seeing a really interesting surge [00:02:16.250] in new developers on the platform, [00:02:19.730] and the new apps that are being created. [00:02:21.640] And I think it really stems from that. [00:02:24.150] So it took an existing trend, really, [00:02:26.510] that was happening within business more broadly, [00:02:29.140] and has accelerated some really key aspects of it. [00:02:33.358] <v ->And so you mentioned new solutions,</v> [00:02:35.300] what are some of the changes we're seeing [00:02:37.700] in how low-code is being used these days? [00:02:40.830] <v ->Well, so if we think about the last few years,</v> [00:02:44.520] there's been a very interesting progression [00:02:47.100] where low-code and no-code, [00:02:50.510] the simpler form of development platform [00:02:53.640] or a developer tool, [00:02:55.930] were initially perceived as being [00:02:57.140] only for departmental-level use, [00:02:59.690] I build an app for myself, [00:03:00.950] I build an app for my immediate team. [00:03:03.790] And it's effectively a disposable app in some way, [00:03:07.170] if you wanna think about it that way. [00:03:08.820] And then we saw a shift towards low-code for innovation. [00:03:14.210] If you're building a new innovative solution [00:03:16.210] that requires some degree of experimentation, [00:03:19.481] and low-code is a great way to go about doing that. [00:03:22.550] But starting two to three years ago, [00:03:24.650] I would say we saw a real evolution [00:03:26.910] of low-code in the enterprise. [00:03:29.120] And by that, I mean, applying a low-code approach, [00:03:32.450] developing a wide variety of enterprise solutions, [00:03:35.830] not just at the departmental level, [00:03:37.610] not just for new innovative solutions, [00:03:40.900] but for rebuilding core systems, [00:03:43.330] legacy modernization or legacy replacement, [00:03:46.850] which was something that I think very few people [00:03:48.940] would have anticipated prior to that point in time. [00:03:52.010] So I think that's one of the big things that we've seen. [00:03:56.680] The other thing that... [00:03:58.410] Yeah, go ahead. [00:03:59.243] <v ->Yeah, I'm wondering,</v> [00:04:00.690] are we seeing people use low-code for things [00:04:02.580] like, I don't know, the Amazon Alexa, [00:04:05.510] or any other sort of-- <v ->Well yeah, that's yeah.</v> [00:04:09.270] That's actually where I was thinking about going next, [00:04:11.560] which is at the same time, [00:04:14.090] we're seeing a change in the way [00:04:16.400] that people interact with businesses. [00:04:19.180] And by virtue of that, how they interact with with systems. [00:04:23.530] And where there's a shift, gosh! [00:04:26.650] Close to 10 years ago now, [00:04:27.800] I began from web-based user interfaces, desktop apps, [00:04:32.790] apps that run on computers to leveraging mobile devices. [00:04:36.760] And initially that kind of parroted what happened [00:04:40.060] on larger displays just on the smaller displays. [00:04:42.990] We've moved well beyond that now [00:04:44.780] to highly tailored mobile solutions [00:04:47.030] that really support what people want to do [00:04:49.710] and need to do on mobile devices. [00:04:51.670] And it's quickly expanding beyond that [00:04:53.900] into new forms of interaction. [00:04:56.310] Conversational user experiences that could happen [00:04:58.970] on a mobile device. [00:05:00.180] It's just another mode on a mobile device, [00:05:02.010] if you wanna think about it that way, [00:05:04.080] to voice-based interactions. [00:05:06.290] And then we're also seeing things like, [00:05:09.690] in the concept of digitalizing things, augmented reality, [00:05:14.360] where we overlay user interfaces onto the world, [00:05:18.340] when people wear wearable devices, and that kind of thing. [00:05:22.290] So the way many people think of that now [00:05:25.440] is multi-experience. [00:05:26.970] And how do we build software solutions [00:05:31.130] that support multi-experience, [00:05:32.700] not just those additional channels, [00:05:35.270] if you wanna think of them as channels, [00:05:37.070] but in a way that really satisfies what the user needs [00:05:39.950] and wants at each touchpoint because really, [00:05:43.120] the system they work with is always there. [00:05:45.050] It's running in the cloud. [00:05:45.940] It's doing things on their behalf, [00:05:47.500] hopefully on their behalf and there's a set of times [00:05:51.211] that individuals interact with that system, [00:05:53.690] we call those touch points. [00:05:55.180] So that in and of itself is a really significant endeavor [00:05:59.790] for virtually every organization that needs to do it. [00:06:02.400] And that's happening at the same time as this other shift [00:06:05.650] around how software is developed and needs [00:06:07.640] to be developed to satisfy enterprise requirements. [00:06:11.840] <v ->So do you think there'll be a difference</v> [00:06:13.960] between low-code platforms and multi-experience platforms? [00:06:17.990] Or do you think that'll become the same thing? [00:06:20.400] <v ->Well, I think there are platforms that can satisfy both.</v> [00:06:25.190] And I think that comes with a lot of different requirements, [00:06:29.110] there's having the right architectural foundation, [00:06:31.920] so that you can create very tailored experiences [00:06:34.940] for each touchpoint. [00:06:36.950] There's what should that experience be? [00:06:39.300] And that's where low-code can really, [00:06:41.510] really add a lot of value as well. [00:06:43.740] And then there's thinking about it really [00:06:45.590] as the user journey and how to map out [00:06:47.840] that user journey and how do you support it? [00:06:50.390] So I think what we'll see is [00:06:52.210] that some low-code platforms can address all that. [00:06:56.540] And there are other tools [00:06:59.850] that have been used outside of app development [00:07:03.270] that may start to make their way into low-code platforms [00:07:07.020] and multi-experience platforms. [00:07:08.660] <v ->So you said something a couple minutes ago</v> [00:07:10.460] that caught my attention. [00:07:11.380] You talked about taking a low-code approach [00:07:14.410] to different types of technology. [00:07:16.730] And when I think low-code approach, [00:07:19.420] what I immediately see is, [00:07:22.270] either drag-and-drop UI development, [00:07:25.110] or I see drag-and-drop flow modeling of logic. [00:07:30.250] But it sounds like what you're talking about [00:07:31.660] is more than that. [00:07:33.420] <v ->Yeah. Yeah.</v> [00:07:34.970] I'm glad you picked up on that. [00:07:37.090] Because low-code approach to me [00:07:40.720] and I think to everybody at Mendix means a lot more [00:07:44.510] than just simplifying things and creating visual tools [00:07:48.260] versus text-based tools. [00:07:50.810] The core principles of it that have been part [00:07:53.260] of our ethos and our product strategy since the beginning [00:07:56.700] are abstraction and automation. [00:07:59.430] So the abstraction is [00:08:00.810] what is the best way to surface something so [00:08:03.570] that people can understand it and work with it? [00:08:07.220] And certainly visual modeling is an approach to abstraction. [00:08:11.520] But the automation part of it is incredibly important. [00:08:14.230] Because there's a lot that needs to be done [00:08:17.370] in the process of building an app, [00:08:19.070] getting an app deployed and managing that app. [00:08:22.260] That is, quite honestly a set of road steps that could [00:08:25.990] and should be automated. [00:08:28.410] So for example, deploying an app to the cloud [00:08:31.860] is a non-trivial thing under the cover, [00:08:33.730] there's a lot of steps that occur, [00:08:36.480] provisioning and managing cloud infrastructure. [00:08:38.830] If you had to do all that manually, [00:08:40.580] it would take a very, very long time. [00:08:42.440] So and also it would be a lot [00:08:44.760] of not exactly high value work at the end of the day. [00:08:49.210] So it's that combination of having the right abstractions, [00:08:53.030] with automation that really comes together in what I think [00:08:57.060] of as the low-code approach. [00:09:00.150] Another aspect of it that is starting to be applied. [00:09:03.420] And I think will have a lot of value moving forward [00:09:05.840] is applying artificial intelligence, [00:09:09.330] augmenting what the developer can do, [00:09:11.220] not just automating what they could do, [00:09:13.690] so that they can do things more quickly [00:09:15.480] and move more quickly. [00:09:17.420] Because that burden is taken off, [00:09:18.860] but help them do things [00:09:20.330] that they might not have been able to do before, [00:09:22.970] or to do things that are not that rote, and repeatable. [00:09:26.700] And that's where AI comes in [00:09:28.700] as a capability for augmenting the developer. [00:09:33.590] So if we take the abstraction, automation, [00:09:36.530] oddly enough, the third one happens to be an A, [00:09:38.910] but artificial intelligence [00:09:40.240] and we put those three things together. [00:09:42.470] That to me is what low-code really needs [00:09:44.430] to do moving forward. [00:09:47.230] But, a key part of what we do though, [00:09:51.220] is try to apply that to virtually everything we can [00:09:54.720] on the behalf of the developer, [00:09:56.520] so that they can move more quickly [00:09:59.430] and be much more expressive about what they do. [00:10:01.820] So they can just declare, I want it to do this. [00:10:04.270] And it will. [00:10:05.960] What that leaves you to [00:10:07.300] is applications become more sophisticated [00:10:09.420] and multi-experience was one example of that [00:10:12.550] is that it takes low-code app development [00:10:14.830] into new domains. [00:10:17.690] <v ->So going back just a minute to AI</v> [00:10:20.580] because I know a lot of people are concerned, [00:10:23.142] AI is gonna take our jobs away, [00:10:25.680] AI is gonna replace people, [00:10:27.700] but you talk about it as augmentation. [00:10:31.930] Can we get into that a little bit? [00:10:33.930] Cause I know that that's kind of a big topic [00:10:36.970] that people think a lot about. [00:10:38.940] <v ->Yeah, yeah.</v> [00:10:40.400] I think it goes back to the nature of tools [00:10:43.320] in the first place. [00:10:45.090] Why do we make tools? [00:10:46.730] We make tools so that we can do bigger and better things, [00:10:49.860] that we can do things... [00:10:52.010] We can move more quickly. [00:10:53.460] While we're doing things, [00:10:54.630] we can accomplish them more quickly. [00:10:57.610] And that's where AI really adds benefit. [00:11:02.490] So I think of AI as being almost entirely [00:11:05.610] about augmentation. [00:11:07.760] Now that augmentation can become very smart [00:11:10.200] and start doing some things for us. [00:11:13.100] But I think that it isn't just a matter [00:11:16.580] of having a philosophy around it [00:11:18.310] to avoid the takeover by the machines. [00:11:20.490] I think that's what the machines do. [00:11:22.440] And when we think about augmentation, [00:11:26.360] we can think about applying it to the developer experience, [00:11:30.790] to allow the developer to do more. [00:11:33.920] And then in the applications that get built, [00:11:36.860] AI clearly will play a very significant role moving forward [00:11:40.560] in augmenting what the users of those applications can do. [00:11:44.250] Because at the end of the day, [00:11:45.360] the applications exist to help the user do something [00:11:48.250] in the first place. [00:11:49.560] And the more that we can bring intelligence to bear [00:11:52.710] within those solutions, [00:11:54.140] those solutions become more competent and more helpful. [00:11:57.830] <v ->The way you describe that, Gordon,</v> [00:11:59.020] it's like there's a direct line [00:12:00.470] from the creation of the wheel. [00:12:03.483] <v ->Yeah, to me, I see it that way, very much see it that way.</v> [00:12:06.020] Yeah, absolutely, chisels. <v ->Creating tool</v> [00:12:08.310] that helps people do more and makes their jobs easier. [00:12:10.930] And there's just this progression from [00:12:12.610] that cave person who figured that out all the way through [00:12:16.510] to augmented reality. [00:12:18.864] <v ->Yeah, yeah.</v> [00:12:19.940] <v ->So one of the other things really interesting,</v> [00:12:23.050] you mentioned a couple minutes ago was about taking [00:12:25.920] this low-code approach or this abstraction [00:12:28.920] and automation approach and applying it to new domains. [00:12:32.560] So which domains should we go tackle first with this? [00:12:36.502] (laughs) [00:12:37.610] <v ->Well, the reason one tackles new domains isn't just</v> [00:12:41.910] to expand an offering. [00:12:45.660] When we think of the offering as an app platform, [00:12:48.370] it's really what does the individual need to do? [00:12:51.220] We refer to them as makers. [00:12:52.470] What does a maker need to do moving forward to make things [00:12:56.640] and there are some aspects of enterprise IT [00:13:02.260] that have been their own domains effectively historically [00:13:06.240] that become part and parcel of making things, [00:13:11.230] building digital solutions. [00:13:13.940] But the first one on the list, I think is data. [00:13:17.800] Data has been a part of applications forever. [00:13:20.320] But in a very limited way. [00:13:22.020] Typically the data with an application belongs to [00:13:24.640] that application, that's less and less true moving forward [00:13:28.610] of the solutions that we create. [00:13:30.670] So how can we make it easier for developers [00:13:34.120] to discover and leverage data within the solutions [00:13:38.610] that they create? [00:13:40.090] Historically, that's been thought of as integration, [00:13:42.930] data integration, data management. [00:13:46.360] I believe very, very strongly, [00:13:48.180] that key aspects of that need to be brought directly [00:13:51.730] into the development realm [00:13:53.570] and into the developer experience. [00:13:56.050] And that will only become more important [00:13:59.130] as the types of data and the shapes that it comes in, [00:14:03.320] and the velocity that it comes in at. [00:14:05.460] And the other characteristics it has, [00:14:07.860] data is time-based, for example, [00:14:09.560] or it isn't in many scenarios, [00:14:12.580] how do we make all of that approachable in a low-code way, [00:14:17.180] how do we apply the notions of abstraction [00:14:19.177] and automation to working with data? [00:14:21.790] So that's one domain. [00:14:26.440] The next one that comes to mind is automation. [00:14:32.115] When we think about it, virtually every application. [00:14:36.800] <v ->So we're not gonna start small with either of these,</v> [00:14:38.580] Gordon (laughs) [00:14:41.700] <v ->Yeah, these are super, super easy problems to solve.</v> [00:14:45.510] Yeah, well, automation really is at the core [00:14:47.400] of what we've been doing with apps forever. [00:14:48.830] I mean, that's the purpose of an app is [00:14:50.470] to do something that an individual did before, [00:14:53.000] but or couldn't get done before. [00:14:55.140] And increasingly, that means more and more digital solutions [00:14:59.800] are inherently process and automation-based in some way. [00:15:06.080] When we think about insurance, [00:15:08.310] what's one of the big things that's happening [00:15:10.030] in retail insurance, [00:15:11.300] you try to radically cut down the time [00:15:13.860] that it takes to quote and bind a policy, right? [00:15:18.370] What is that? [00:15:19.203] That's process improvement, [00:15:20.430] that's optimizing a process, that's automating a process, [00:15:24.030] and perhaps even applying some intelligence to it [00:15:26.870] so that individuals don't have [00:15:28.690] to get involved in very much, [00:15:30.320] that's what allows you to get a quote done instantly. [00:15:33.570] There's a ton of integration behind that. [00:15:35.890] And there's a ton of automation behind that. [00:15:38.881] So therefore, to create digital solutions, [00:15:42.370] process and process automation has to become part [00:15:45.600] of the palette of capabilities [00:15:47.830] that a maker has available to them. [00:15:50.420] <v ->Okay, and now I'm starting to see that connection</v> [00:15:52.640] with multi-experience that we were talking about. [00:15:55.020] Because I may wanna get a quote on my desktop, [00:15:59.360] but I might just wanna get a quote on my mobile device [00:16:03.430] or I might want to speak to a voice device to get a sense [00:16:07.600] of what it might cost. <v ->Absolutely.</v> [00:16:09.890] Absolutely. Absolutely. [00:16:12.010] Or if I'm ordering a vehicle or even deciding what vehicle [00:16:16.580] to buy, right, that process begins and then it takes [00:16:20.430] some length of time for that to complete. [00:16:22.360] So systems need to keep track of who the person is, [00:16:24.790] what they're doing, [00:16:25.910] and increasingly a variety of processes that come together [00:16:30.570] to support the their journey. [00:16:33.330] And I love the way you put that. [00:16:36.380] They check in on those processes and they interact [00:16:39.030] with them on their terms whenever they want to [00:16:41.170] on whatever device they want to. [00:16:43.240] So yeah, we need to support that. [00:16:46.780] <v ->Okay, so we talked about data,</v> [00:16:48.210] we talked about process automation. [00:16:50.570] Is there anywhere else you think we could get a real bang [00:16:54.070] for our buck by applying a low-code approach to things? [00:16:58.550] <v ->Well, the other component that I think needs to come in</v> [00:17:01.920] and it's beginning to come in [00:17:03.250] and we kind of alluded to it earlier in the discussion [00:17:06.190] is applying intelligence within digital solutions. [00:17:13.050] AI is an interesting thing. [00:17:15.470] Because at one end, [00:17:18.930] there's a lot that we still can't do with it, right? [00:17:21.130] And you have to have a very, [00:17:22.640] very clear idea of what you need to create. [00:17:24.870] And then you need data scientists to go off [00:17:27.070] and build it and all that stuff. [00:17:28.380] But on the other end, which is available today, [00:17:31.040] we have cognitive services [00:17:32.930] and intelligent processing services. [00:17:34.680] That's all very clear, [00:17:36.440] very repeatable problems using natural language processing, [00:17:40.630] that gets applied in digital solutions in a lot [00:17:43.710] of different ways. [00:17:46.220] Even Intelligent OCR, is a component of that. [00:17:49.850] So, being able to bring cognitive services directly [00:17:54.750] into the solutions that are being created, [00:17:58.080] again as a natural extension of the pallet [00:18:00.280] that the developer has available to them to work from, [00:18:03.630] or the toolkit that they have. [00:18:06.030] I think that's the third leg of the stool really, [00:18:09.830] that if we look at data and how to bring in data [00:18:13.420] and work with data and manage data, [00:18:15.840] we bring in the process and automation component. [00:18:18.610] And then we bring in AI to use a broad term [00:18:21.850] for how to bring intelligence into the equation. [00:18:24.000] I think that's the triangle of frames, [00:18:26.710] need to have moving forward. [00:18:28.700] <v ->Okay, so you mentioned the developer,</v> [00:18:30.940] and I think we should talk about him or her a little bit [00:18:34.740] because this feels like a lot of new technology for them [00:18:38.310] to absorb and a lot of additional types of output [00:18:44.890] and expectation around what a developer skill set is. [00:18:51.120] Is that gonna be just impossibly hard for them to absorb? [00:18:56.390] <v ->Well, clearly the intention of abstraction</v> [00:19:00.050] and automation is to remove [00:19:03.620] from the developers experience the details they don't need [00:19:05.730] to worry about. [00:19:06.770] But as you point out really, really nicely there, [00:19:09.950] there are new dimensions [00:19:11.070] that the developer needs to consider. [00:19:13.700] We can make it easier for them, [00:19:14.950] but they need to consider those dimensions. [00:19:17.160] And it's also the intersection of these things [00:19:20.510] that creates digital solutions. [00:19:22.850] So what we're beginning to see is a scenario [00:19:27.820] where developers don't build things from scratch [00:19:31.230] from very kind of low-level technical capabilities anymore, [00:19:35.680] where they begin to build more [00:19:37.780] and more sophisticated solutions out [00:19:40.070] of bigger building blocks [00:19:42.370] and more sophisticated building blocks. [00:19:45.310] And at Mendix, [00:19:47.430] the way we think about that is a concept [00:19:50.620] that we think of as an app service. [00:19:53.080] And an app service isn't just a widget or a component. [00:19:59.480] It may certainly contain those. [00:20:01.880] But it contains a set of widgets, components, [00:20:06.020] connections to cloud hosted services [00:20:08.790] that when brought together solve a specific problem. [00:20:12.940] It could be a fairly fine-grained problem [00:20:16.480] like natural language processing. [00:20:20.010] But what makes such a service more valuable [00:20:23.720] is when it is contextualized with respect to say, [00:20:28.100] a domain or a use case, [00:20:30.350] what makes it really valuable is when you put a number [00:20:32.860] of these things together into what you could think of [00:20:35.580] as a subassembly, [00:20:37.400] or even a completely templatized solution, [00:20:41.860] which would allow the developer to then [00:20:45.210] basically snap together solutions [00:20:47.830] that are very sophisticated and then customize [00:20:51.330] and tailor the aspects of it that they need [00:20:53.610] to customize or tailor for their personal situation. [00:20:57.800] So that I think is fundamentally required moving forward [00:21:01.650] to give people larger and larger things [00:21:03.860] that they can leverage that are more sophisticated, [00:21:06.260] that bring various pieces together to solve a problem, [00:21:09.590] but do it in a componentized way, [00:21:11.960] so that they can then do whatever they want with that. [00:21:15.420] So for example, [00:21:16.253] if they want to add a new form of interaction channel, [00:21:20.130] or alter the process, [00:21:21.990] or nest it in the larger process, make it their own, [00:21:25.620] they can do all those things. [00:21:26.880] So app services, I think were our real key [00:21:31.410] to creating a low-code platform that enables people [00:21:35.430] to build these much more sophisticated solutions [00:21:39.210] that have these new dimensions that we're talking about. [00:21:42.500] <v ->Okay, so I think a more concrete example might help here.</v> [00:21:45.430] I think maybe we've gone too far, I mean abstraction side, [00:21:48.230] if you will? [00:21:49.990] So what would be an example of an app service [00:21:53.560] that we might offer developers the ability to use? [00:21:57.770] <v ->Well, if we look at the more kinda fine-grained ones</v> [00:22:01.140] that you can start with things like 3D visualization. [00:22:05.930] That's a technology that's used a lot in engineering [00:22:09.000] and manufacturing scenarios. [00:22:10.330] So you can imagine that Siemens has a lot [00:22:12.330] of great technology in that area, of course. [00:22:15.210] But it's also really important in a lot of other scenarios. [00:22:18.920] AR and VR, obviously, are fundamentally 3D-based. [00:22:22.690] So providing app services that allow people [00:22:26.410] to use 3D visualization, [00:22:29.070] and build 3D user interfaces, [00:22:32.070] would be a great example of a sophisticated component [00:22:37.340] that would be available as an app service. [00:22:40.730] Same thing would be true of say, [00:22:43.300] I brought up natural language processing, right, [00:22:45.300] so those will be there. [00:22:46.420] But if we just stopped at that level, [00:22:49.130] that wouldn't be that different than if we were [00:22:51.250] to put abstraction and automation on top of services [00:22:53.780] that are available just in the world. [00:22:56.490] What's interesting is when you put them together, [00:22:58.230] as you pointed out to solve a particular problem. [00:23:00.780] So if we think about document processing, [00:23:03.720] every organization has hundreds, [00:23:06.590] if not thousands of processes [00:23:09.360] that effectively document processing processes. [00:23:12.790] And they invariably involve exactly the same thing. [00:23:16.880] So you need to be able to analyze a document [00:23:20.240] in some way create, figure out what its structure is, [00:23:22.900] apply some natural language processing [00:23:25.370] to actually extract the content, [00:23:27.150] based upon that content, [00:23:28.330] make an evaluation of what to do next, [00:23:30.220] where should I go in the process? [00:23:31.470] Does it need to be reviewed by a human, things like that. [00:23:35.750] So there's an intelligent decision-making aspect of this, [00:23:38.840] which again, gets back into process. [00:23:41.950] One could imagine app services that solve that problem [00:23:45.630] in a generalized way, [00:23:47.360] and then allow for customization as necessary at each [00:23:50.380] of those stages or in each of those dimensions. [00:23:53.080] So for example, natural language processing [00:23:56.250] has much lower error rates when it is very domain-specific. [00:23:59.660] I'm handling POs, in this industry, for example, [00:24:03.090] allows POs to be processed more effectively. [00:24:06.100] So that could happen. [00:24:07.900] But I would also anticipate that moving forward, [00:24:11.260] there'll be highly specialized versions of something [00:24:14.450] that's as common as document processing into something [00:24:18.080] that's very appropriate for specific industries. [00:24:22.330] <v ->So that brings a kind of obvious question to bear,</v> [00:24:26.870] which is, that sounds like a very big engineering investment [00:24:32.800] for Mendix to make, to build all of this stuff, [00:24:35.780] and have all of this domain expertise, [00:24:38.520] so how are we thinking about that? [00:24:41.500] <v ->Yeah, well, that's a really good point.</v> [00:24:43.950] And certainly, [00:24:46.950] no platform vendor [00:24:49.780] is ever going to have all [00:24:51.550] of the domain expertise that's necessary. [00:24:55.070] On the technology front, [00:24:56.450] there are specialist providers of horizontally [00:25:00.400] applicable technology. [00:25:02.750] And sometimes those are used more in some industries [00:25:06.220] than others, for example, geolocation, and geo-tagging [00:25:11.940] and those types of things tend to be used [00:25:13.624] in certain industries, right. [00:25:15.370] But it wouldn't make any sense for Mendix, [00:25:18.270] or any platform vendor to invest in creating all of those. [00:25:23.470] And then, there are people and organizations [00:25:26.850] that have a lot of domain expertise. [00:25:28.740] They understand this particular form of manufacturing. [00:25:32.150] They understand specialty insurance, [00:25:34.100] they understand what they understand. [00:25:35.880] And they might have some really intriguing ideas [00:25:38.130] for how they could more broadly solve problems [00:25:40.930] within those domains. [00:25:42.570] So what we envision [00:25:45.400] is an ecosystem [00:25:47.720] where there's the platform, [00:25:51.090] and the capabilities and app services that we provide. [00:25:55.550] There are contributors to that marketplace [00:25:59.630] that provide specialized technology in areas. [00:26:04.680] I gave you the example of geolocation and geo-tagging. [00:26:07.580] It's also content management systems and things like that, [00:26:10.550] that would be part of it. [00:26:11.860] And then there are the individuals that do bring in [00:26:14.570] that domain expertise to build higher level solutions [00:26:17.730] that have value and how these capabilities [00:26:19.930] are brought together [00:26:20.763] to solve a specific problem in a domain. [00:26:23.510] And so what we are creating is a marketplace [00:26:27.350] to facilitate that, [00:26:29.110] to create that ecosystem with a variety of participants [00:26:34.090] that are necessary for creating these building blocks. [00:26:37.280] So that enterprises and organizations regardless [00:26:40.320] of what industry they're in, [00:26:42.140] can find the capabilities that they need [00:26:45.010] to build their digital solutions. [00:26:48.200] <v ->And do we see sort</v> [00:26:50.470] of an inception thing going on here where, [00:26:54.140] a solution provider might find a piece of a solution [00:26:57.480] in the marketplace, put it into their solution [00:26:59.920] and then reoffer it in the marketplace? [00:27:02.040] So as-- <v ->Oh, absolutely, absolutely.</v> [00:27:05.532] I believe that will naturally happen. [00:27:07.870] And we've witnessed similar things happening [00:27:10.900] in other industries, for sure, right. [00:27:12.790] I mean, that's exactly how the auto industry works. [00:27:15.870] So absolutely the same thing could happen [00:27:18.860] in the low-code space. [00:27:20.930] And the great thing is that everybody [00:27:22.940] that participates gets value out of that. [00:27:25.270] They get compensated for what they contribute. [00:27:27.390] And they also obtain value by creating something [00:27:30.600] that has more value on top of that. [00:27:32.960] So yeah, I like that infectious notion. [00:27:36.910] Yeah, I believe so. [00:27:38.600] <v ->Okay, so we've covered a lot in a short amount of time.</v> [00:27:43.800] And I'm excited about all of this technology. [00:27:47.490] I'm sure our viewers are also, [00:27:51.140] how soon can we have all of this? [00:27:53.050] What's the time horizon on the sorts [00:27:55.820] of things we've been talking about today? [00:27:58.050] <v ->Well, the core of what we're talking about today</v> [00:28:00.980] is already there of course. [00:28:02.910] The foundational elements have been part of our platform [00:28:06.560] and our platform strategy for quite some time, [00:28:09.580] and we've been delivering against that. [00:28:12.600] So much of what we're talking about in some way, [00:28:15.060] shape or form is available now or very, very shortly. [00:28:20.440] Some of the things that we're talking about, [00:28:22.890] the technologies involved [00:28:24.600] are at different levels of maturity. [00:28:27.656] As we talked about, at least at a high level a bit ago, [00:28:30.620] you can build some very comprehensive solutions today, [00:28:33.550] leveraging existing AI capabilities. [00:28:37.350] You may or may not call them AI, [00:28:39.920] they're definitely AI though at their core, [00:28:42.000] so cognitive services, for example, [00:28:44.800] but it's gonna be quite some length of time before anybody [00:28:49.140] that wants to, [00:28:49.973] somebody that doesn't have a data science background, [00:28:52.580] can create their own machine intelligence, right. [00:28:56.600] So, you leverage what's available versus creating your own. [00:29:02.260] One could imagine a future in which many more people [00:29:06.370] can basically create their own machine learning systems [00:29:09.950] and train them. [00:29:11.860] How close are we to that? [00:29:13.180] Three years, five years, maybe longer? [00:29:16.980] Will AI play a role in even creating those solutions? [00:29:20.130] Will AI help the maker make them? [00:29:22.880] Absolutely, there's no question about that. [00:29:25.260] And I would anticipate that the low-code approach [00:29:29.720] that combination abstraction and automation [00:29:32.400] is at the core of it as well. [00:29:35.670] So some of this plays out over many, many years. [00:29:40.120] But I think the beginnings of [00:29:41.750] it are absolutely there today. [00:29:44.530] <v ->That's exciting.</v> [00:29:45.690] Well, Gordon, as usual, [00:29:47.160] great catching up with you on what is coming in the future [00:29:50.550] and how we're thinking about it. [00:29:52.030] Always a pleasure. [00:29:53.710] Next year, we'll do it in Rotterdam. [00:29:56.033] <v ->(laughs) I certainly hope so.</v> [00:29:57.900] Great chatting Sheryl. [00:29:58.840] As always. [00:30:00.324] (upbeat music)