 From around the globe, it's theCUBE with digital coverage of OutSystems Next Step 2020, brought to you by OutSystems. I'm Stu Miniman and welcome back to theCUBE's coverage of OutSystems Next Step. Of course, one of the items that we've been talking a lot in the industry is about how artificial intelligence, machine learning are helping people as we go beyond what really human scale can do and we need to be able to do things more machine scale to help us really dig into this topic. Happy to welcome to the program first-time guest, Antonio Allegria. He is the head of artificial intelligence at OutSystems. Antonio, thanks so much for joining us. Thank you, Stu. I'm really happy to be here and really talk a little bit about what we're doing at OutSystems to help our customers and our leveraging AI to get to those goals. Wonderful. So I saw ahead of the event a short video that you did and talked about extreme agility with no limits. So, you know, before we dig into the product itself, maybe if you could just, you know, how should we be thinking about AI? You know, there's broad spectrum as I said, you know, machine learning, that there's various components in there. Listen to the big analyst firms, you know, the journey, its big steps and something that is pretty broad. So when we're talking about AI, you know, what does that mean to you? What does that mean to your customers? Yeah, so AI at OutSystems really speaks to the vision and the core strategy we have for our product, which is, you know, if you saw the keynote, you know, we talk about, you know, really enabling every company, even those that, you know, that have existed for decades, perhaps have a lot of legacy to become, you know, leading elite cloud software development companies and really can develop digital solutions at scale really easily. But one thing we see, and then this is a big statistic, one of the things that limits CIOs the most nowadays is really the lack of talent, lack of, you know, engineering and software engineering, you know, ability in people that can do that. And there's a statistic that was reported by Wall Street Journal, I saw it recently, perhaps last year, that said that according to federal jobs data in the US, by the end of 2020, there would be about a million unfilled IET and software development jobs available, right? So there's this big problem that all of these companies really need to scale, really need to invest in digital systems. And so our belief at OutSystems, we've already been abstracting and we've been focusing on automating as much as possible the software development tools and applications that we use. We've already seen amazing stories of people coming from different backgrounds, really starting to develop really leading edge applications. And we want to take this to the next level. And we believe that artificial intelligence with machine learning, but also with other AI technologies that we're also taking advantage of can really help us get to a next stage of productivity. So from 10x productivity to 100x productivity. And we believe AI plays a role in three ways. We believe AI by learning from all of this data that we now collect in terms of, that projects are being developed. We are essentially trying to embed a tech lead, so to speak, inside a product in a tech lead that can help developers by guiding them, guiding the most junior ones by automating some of the boring repetitive tasks or by validating their work, making sure that they are using the best practices, making sure that it helps them as they scale to refactor their code to automatically design their architectures, things like that. What wonderful, Antonio, I got Gonzalo stated it quite clearly in the interview that I had with him. It's really about enabling that next 10 million developers. We know that there is that skill gap, as you said. And everybody right now, how can I do more? How can I react faster? So that's where the machine learning, artificial intelligence should be able to help. So bring us inside, I know the platform itself has had guidance and the whole movement, what we used to call low code was about simplifying things and allowing people to build faster. So bring us inside the product, what are the enhancements, what are the new pieces, some of the key items? Yeah, so one interesting thing, and I think one thing that I think OutSystem is really proud of being able to achieve is if you look at how OutSystem has been using AI within the platform, we started with introducing AI assistance within our software development environment, Service Studio, right? And so this capability, we've been iterating it a lot, we've been evolving it, and now it's really able to accelerate significantly and guide novices but also help pros dealing through the software development process and coding by essentially trying to infer and understanding their context and trying to infer their intent and then automating the steps afterwards. And we do this by suggesting you the most likely, let's say function or code piece that you will need, but then at the next step, which we are introducing this year even better, which is we're trying to auto-fill most of the, let's say the variables and all of that and the data flow that you need to collect. And so you get a very delightful frictionless experience as you are coding. So you're closer to the business value even more than before. Now, this was just the first step. What you're seeing now and what we're announcing and we're showing at this next step that we showed at the keynote is that we are starting to fuse AI across the out systems products and across the software development lifecycle. So we took this core technology that we use to guide developers and assist and automate their work. And we use the same capability to help developers, tech leads and architects to analyze the code, learning from the bad patterns that exist, learning from and receiving runtime information about crashes and performance. And inside the product that we call architecture dashboard, we're really able to give recommendations to these architects and tech leads, where should they evolve and improve their code. And we're using AI and refusing AI in this product in two very specific ways now that we're releasing today, which is one is to automatically collect and design and define the architecture. So we call this automated architecture discovery. So if you have a very large factory, you can imagine have lots of different modules, lots of different applications. And if you need to go and manually have to label everything, so this is a front end, this is a back end, that would take a lot of time. So we use machine learning, learning from what architects have already done in the past of classifying their architecture and we can map out your architecture completely automatically, which is really powerful. Then we also use our AI engine to analyze your factory and we can detect the best opportunities for refactoring. So refactoring is one of the top problems in the top smells and technical debt problems that large factories have, right? So we can completely identify and pinpoint what are these opportunities for refactoring and we guide you through it, we tell you, okay, all of these hundreds of functions and logic patterns that we see in your code, you could refactor this into a single function and you can save a lots and lots of code because as you know, the best code, the fastest code, the easiest to maintain is the code you don't write, you don't have. So we're trying to really eliminate code from these factories, from these capabilities. Well, it's fascinating, you're absolutely right. I'm curious, I think back to some of the earliest interactions I had with things that give you guidance, spell checkers, grammar check, how much does the AI that you work on, does it learn what's specific for my organization and my preferences? Is there any community learning over time because there are industry best practices out there that are super valuable, but we saw in the SAS wave when I can customize things myself or learn over time. So how does that play into kind of today in the roadmap for AI that you're building? That's a great question. So our AI, let's say technology that we use, it actually uses two different big kinds of AI. So we use machine learning definitely to learn from the community, what are the best practices and what are the most common patterns that people use? So we use that to guide developers, but also to validate and analyze their code. But then we also use automated reasoning. So this is more logic-based, reasoning-based AI. And we pair these two technologies to really create a system that is able to learn from data, but also be able to reason at a higher order about what are good practices and kind of reach conclusions from there and learn new things from there. Now we started by applying these technologies to more of the community data and kind of standard best practices, but our vision is to more and more start learning specifically and allowing tech leads and architects, even in the future, to tailor these engines of AI, perhaps to suggest these are the best practices for my factory. These patterns perhaps are good best practices in general, but in my factory, I do not want to use them because I have some specificities for compliance or something like that. And our vision is that architects and tech leads can just provide just a few examples of what they like and what they don't like in the engine just automatically learns and gets tailored to their own environment. So important that you're able to have the customers move things forward to the direction that makes sense on their end. I'm also curious, you talk about what partnerships OutSystems has out there, being able to tie into things like what the public cloud is doing, lots of industry collaboration. So, how does OutSystems fit into the broader AI ecosystem? Yeah, so one thing I did not mention and to your point is so we have kind of two complimentary visions and strategies for AI. So one of them is we really want to improve our own product, improve the automation in the product and the abstraction by using AI together with great user experience and the best programming language for software automation. So that's one, that's what we generally call AI assisted development and infusing AI across the software development lifecycle. The other one is we also believe that true elite cloud software companies that create frictionless experiences, one of the things that they use to really be super competitive and create these frictionless experiences is that they can themselves use AI and machine learning to automate processes, create really, really delightful experiences. So we're also investing and we've shown and we're launching and announcing at Next Step, we just shown this at the keynote, one tool that we call the machine learning builder, ML builder. So this essentially speaks to the fact that a lot of companies do not have access to data science talent, they really struggle to adopt machine learning like just one out of 10 companies are able to go and put AI in production. So we're essentially abstracting also that, we're also increasing the productivity for customers to implement AI and machine learning. We use partners behind the scenes and cloud providers for the core technology with automated machine learning and all of that, but we abstract all of the experience. So developers can essentially just pick the data they have already inside the out systems platform and they want to just select, I want to train this machine learning model to predict this field, just click, click, click and it runs dozens of experiments, selects the best algorithms, transforms the data for you without you needing to have a lot of data science experience and then you can just drag and drop in the platform, integrate in your application and you're good to go. Well, it sounds phenomenal. You mentioned data scientists, we talked about the skill gap. Do you have any statistics? Is this helping people higher faster, lower the bar to entry for people to get on board, increase productivity? What kind of hero numbers do your customers typically, how do they measure success? Yeah, so we know that for machine learning adoption at companies, we know that, sorry, this is one of the top challenges that they have. Companies do not, it's not only that they do not have the expertise to implement machine learning in their products and their applications, they don't even have a good understanding of what are the use cases and what are the technology opportunities for them to apply. So this has been listed by lots of different surveys that this is the top problem. These are the two of the top problems that companies have to adopt the ISOs, access to skill, data science skill, understanding of the use case and that's exactly what we're trying to kind of package up in a very easy to use product where you can see the use cases you have available, you just select your data, you just click train, you do not need to know the new degree details. And for us, a measure of success is that we've seen customers that are starting to experiment with ML Builder is that in just a day or a few days they can iterate over several machine learning models and put them in production. We have customers that have, no machine learning models in production ever and they just now have two and they're starting to automate processes, they're starting to innovate with business and that for us is we've seen as kind of the measure of success for businesses. Initially, what they want to do is they want to do POCs and they want to experiment and they want to get to production, start getting the field for it and iterate. From a product standpoint, is the AI just infused in or are there additional licensing, how do customers take advantage of it? What's the impact on that from the relationship without systems? Yeah, so for AI and machine learning that is fused into our product and for automation, validation and guidance, there's no extra charge, it's just part of the product. It's what we believe is kind of a core building block in a core service for everything we do in our product. For machine learning services and components that customers can use to in their own applications, we allow you to integrate with cloud providers and the building is done separately. And that's something that we're working towards and building great technical partnerships and exploring other avenues for deeper integration so that developers and customers do not really have to worry about this thing as well. Yeah, well, it's such a great way to really democratize the use of this technology, platform that they're used to, they start doing it. What's general feedback from your customers? Did they just like, oh, it's there, I start playing with it, it's super easy, it makes it better. Are there any concerns or pushback have we gotten beyond that? What do you hear any good customer examples you can share as to general adoption? Yeah, so as I said, as we reduce the friction for adopting these technologies, we've seen one thing that's very interesting. So we have a few customers that are, for example, more in the logistics side of industry and vertical. And so they have a more conservative management, like they take time to adopt, they're more of a laggard in adopting these kinds of technologies, the business is more skeptical, doesn't want to spend a lot of time playing around, right? And once they saw what they could do with a platform, they quickly did a proof of concept, they showed to the business and the business had lots of ideas. So they just started interacting a lot more with IT, which is something we see without systems platform, not just for AI machine learning, but generally in the digital transformation is when the IT can start really being very agile and iterating and innovating and they start collaborating a lot with the business. And so what we see is customers asking us for even more. So customers want more use cases to be supported like this. Customers also, the ones that are more mature that already have their centers of excellence and they have their data scientists, for example, they want to understand how they can also bring in perhaps their use of very specialized tools, how can they integrate that into the platform so that for certain use cases, developers can very quickly train their own models, but so specialized data science teams can also bring in and developers can integrate their models easily and put them into production, which is one of the big barriers. We see in a lot of companies, people working on year long projects, they develop the models, but they struggle to get them to production. And so we really want to focus on the whole end-to-end journey. Either you're building everything within the OXSYS platform or you're bringing it from a specialized pro tool, we want to make that whole journey frictionless and smooth. Great, Antonia, final question I have for you. Of course, this space we're seeing maturing, rapid new technologies out there. It gives a little look forward. What should we be expecting to see from out systems or things even a little broader if you look at your partner ecosystem over kind of the next 6, 12, 18 months? Yeah, so you're going to continue to see a trend, I think from the cloud service providers of democratization of the AI services. So this is a trend that's just starting to advance and accelerate as these providers start packaging. It's like what our system is also doing, starting to packaging some specific well-defined use cases and then making the journey for training these models and deploying super simple. That's one thing that's continued to ramp up and we're going to move from AI services more focused on cognitive pre-trained models, which is kind of the status quo to custom AI models based on your data. That's kind of the trend we're going to start seeing and that out system is also pushing forward. Generally from the AI and machine learning applications and technology side of thing. I think one thing that we are leading on is that machine learning and deep learning is definitely one of the big drivers for the innovation that we're seeing in AI. But you start seeing more and more what is called hybrid AI, which is taking machine learning and database artificial intelligence with more logic-based automated reasoning techniques and pairing these two to really create systems that are able to operate at a really higher level, higher cognitive level, which is what out systems is investing internally in terms of research and development with partnerships with institutions like Carnegie Mellon University and institutions such as that. Wonderful, why Antonio, who doesn't want a tech expert sitting next to them helping get rid of some of the repetitive, boring things or challenges? Thank you so much for sharing the update. Congratulations. Thank you, Stu. Definitely look forward to hearing more in the future. Thank you, Stu. Have a good day. Stay tuned for more from OutSystems Next Step. I'm Stu Miniman and thank you for watching theCUBE.