Unlock the secrets of digital transformation as we sit down with Tiago, the CIO of OutSystems, who reveals how the convergence of low-code platforms and generative AI is reshaping enterprise strategies. This episode uncovers the profound impact of AI on customer support, showcasing significant advancements in automation and query resolution. Tiago shares real-world examples and candidly discusses the hurdles of scaling AI initiatives, underscoring the necessity of robust data and strategic frameworks.
Venture into the future of application development with insights into Data Fabric and GenAI. Discover how seamless data integration boosts business efficiency and learn about the key performance indicators that truly matter in digital transformation projects. We also tackle the ethical and governance challenges posed by AI, providing a balanced view of its potential to drive positive change. Finally, get a glimpse into the future of automated decision-making and its market implications, all through the visionary lens of Tiago and the groundbreaking work at OutSystems.
Unlock the secrets of digital transformation as we sit down with Tiago, the CIO of OutSystems, who reveals how the convergence of low-code platforms and generative AI is reshaping enterprise strategies. This episode uncovers the profound impact of AI on customer support, showcasing significant advancements in automation and query resolution. Tiago shares real-world examples and candidly discusses the hurdles of scaling AI initiatives, underscoring the necessity of robust data and strategic frameworks.
Venture into the future of application development with insights into Data Fabric and GenAI. Discover how seamless data integration boosts business efficiency and learn about the key performance indicators that truly matter in digital transformation projects. We also tackle the ethical and governance challenges posed by AI, providing a balanced view of its potential to drive positive change. Finally, get a glimpse into the future of automated decision-making and its market implications, all through the visionary lens of Tiago and the groundbreaking work at OutSystems.
Hey everybody, fascinating chat today, without systems, on what happens when low-code meets Gen AI in the enterprise with Tiago. How are you, tiago?
Speaker 2:
I'm good, evan. Thank you for having me Well. Thanks for being here. I'm very grateful for our conversation.
Speaker 1:
Yeah, you guys do amazing work and I'm so intrigued by your mission and roadmap and vision. Before that, maybe let's dive into some introductions your role as CIO at OutSystems and, for those who aren't familiar, your really unique position in the enterprise space.
Speaker 2:
Oh, thank you. Yeah, so, as you said, we're a software development platform operating in the low-code space and AI-powered, as it could only be. Right now, I am the CIO, so I'm in charge for all the systems and architecture that supports our business. We like to call it customer number one of our own platform.
Speaker 1:
Well, brilliant poll position to help us understand what's really happening on the front lines of digital transformation, business transformation. Maybe talk about what you're seeing on those front lines in terms of addressing the massive uptake of digital applications and AI within your own practice. What are some of the opportunities you see and some of the challenges you're facing firsthand?
Speaker 2:
Yeah. So we've been seeing a huge uptake in digital transformation over the last years and companies taking advantage of multiple devices mobile web, the cloud, etc. To be able to address their challenges, become more digital, improve and generate business impact by either improving operations or finding opportunities to generate more revenue. What we are seeing over the last year with the introduction of GenAI, is that that mega trend seems to be more powerful than any of the ones that I just mentioned before. So what we are seeing is that the business impact generated by infusing generative AI in solving some of the problems we have, the business impact is kind of literally off the charts and it's almost exceeding our expectations in everywhere we put it. I'll be happy to talk to you about some of those examples as we go along.
Speaker 1:
Yeah, that's a great point and I love the fact you're a practitioner. No disrespect to your marketing colleagues, but we like to talk big ideas and themes when you're really hands-on driving transformation agility. So what are some of the maybe specific ways it's helping drive growth, productivity, new opportunities, et cetera? When it comes to Gen AI, yeah.
Speaker 2:
So as a high-tech company, we kind of are faced with all the challenges of hyper-growth and being able to create operations that are able to scale as we grow, and always looking towards being able to create operations that allows us to scale without having to always be adding more people as the operation grows. And with the introduction of generative AI, we started looking at different opportunities across the board to be able to apply this technology, infused in our own platform, to be able to change some of the digital experience of our customers and employees, and that's exactly what we've been doing and we are seeing really amazing results, to the point that I'm always having to catch up with all the teams every time I have to mention some of the numbers, because they change so fast that I face the risk of being out of date most of the time.
Speaker 1:
Yeah, I bet, and staying ahead of the game is probably what this is all about. Are there any use cases or examples of projects?
Speaker 2:
So we started by picking up one use case first, which was our customer support or customer service operation, and the idea that came to mind was to try to alleviate that operation by being able to answer some of the inquiries that our customers have. And our target along the way, with many other techniques before Gen AI came along, was to be able to sort of like answer about 10% of those inquiries, and I can tell you that we've never reached that number. We were always below that number, even though we served Google-powered searches and many other different techniques. Our documentation is very powerful and very good and we would serve most of those articles, but a lot of the customers ended up always opening a ticket to the other end. And since Gen AI came along, we're now able to automate 30%, which is three times higher, what we've never been able to achieve that operation, to achieve that operation, and if we focus particularly on specific cases of the types of incidents or questions, these numbers go up to 60% and 70%. So basically, we are kind of raising up the bar and the target as we go along and we infuse more. You know, generative AI also has seen some of the progress in the last year, so as we infuse more and more of those things, it becomes even better.
Speaker 2:
But when we started and we had multiple ideas for other use cases and I used to say I had people lining up my office with kind of folders in their hands saying like I have an idea, how can we move forward? And we started identifying that they all followed a certain pattern. Either you had to sort of like query data and structure data in the format of several documents and articles and so on, or you had to analyze huge volumes of data and kind of building it to a specific structure, or you had to which we call it content deconstruction, and there were a few other examples of patterns that we saw that every use case would follow. So we started, we created a number of foundations and a data strategy to be able to answer those so that every time a new use case would come along, we would see where it would fit in the space of the different patterns and then we would apply the same solution so that, basically, I could let go of the people that were lining up and queuing up in front of my office with all those requests and allow them to scale One of the things that I mean amongst the CIO group we talk about is that it was easy for most everyone to start experimenting with a first use case and a lot of people had a few proof of concepts along the way.
Speaker 2:
But the difficulty that we all have faced is the ability to scale that throughout the enterprise and really capture business value. Experiments are great for you to kind of learn how the technology works and start facing some of the challenges. We learned the hard way that good data provides amazing business impact. Bad data has disastrous results. So that allowed us to also focus on the quality of the data that we use to feed these LLMs and these models in every use case and basically, with the setting up of all these foundations, we were able to accelerate and kind of move these use cases to a kind of a corporate environment and enterprise scale.
Speaker 2:
Another example of a great use case is with translation. So we have a community website. We have a very active community of developers. It's actually amazing. People help each other and, uh, they're very active.
Speaker 2:
But uh, in in countries where english um is not the main language and people have a hard time on on speaking the language we we would face a smaller um engagement with those communities.
Speaker 2:
So we started with the japanese community, where we have some of our biggest customers, like Toyota, and we invested on translating that content so that that community was able to engage. I can tell you that we spent about five to six years before and we never made it, because every time we would approach the community with content, they would say, no, these translations do not work, please don't do this live, this is terrible, et cetera. And when Gen AI came along, just in the first three months, we had a 35% uptick in the community engagement. It's now over 75% and I risk that I'm already telling you data that's not good. Next week we're going to launch new languages like Korean, french and new others will be coming along along, and on our support portal you actually can also ask the questions in your own language, if you, and then the system would answer in your own language, irrespectively of the fact that the documentation is all in English.
Speaker 1:
Fantastic. I love the Japan example Because I actually read an article today that 40% of Japanese companies have no plans to implement AI or Gen AI zero plans initiatives underway, and I thought that was shocking. Though, despite a global concern like Toyota being on the leading edge, it's amazing how many companies are still resistant to change or change adverse. You know change is hard in so many ways. How do you think companies can overcome that resistance to change and digital transformation and new technology adoption? You know what role does leadership, like you're doing, having an open door, talking to change makers internally how do you overcome that resistance or reluctance that's out there?
Speaker 2:
So we're always learning. One of the things that happened to me and to us the most in the group that we are implementing is to be very honest with you at a certain stage we, we, we are hard to be understood. It's hard to be understood because we talk a language that the majority of the people don't understand, and so we're actually now moving into um, an approach where we show more than we tell um, so literally demonstrate and prototype examples so that people understand what is the power behind the technology. Because in this field, you'll hear a lot of people saying, according to what I read or someone told me that, and literally what you hear is that they've never experimented. They're assessing and evaluating based on what they hear and what they read and the media that they consume. But you have to actually experiment. There's a reason why prompt engineering became a profession you have to understand how to play with prompts and provide context to be able to take the best out of these models, and what we have been able with our own technology is to be able to do that really fast to pivot into first kind of small use cases and prototypes and then iterate basically pivot, then measure the results, learn with that and continuously iterate. I can tell you I don't even know the number, because it's like in the space of the hundreds or even thousands we have made so many iterations on our support model as we went along that it doesn't look anything like it was at the very beginning, but we weren't afraid.
Speaker 2:
I think one of the things we spent quite a lot of time with our legal department on making sure that we did this right created an AI policy that I believe and we believe that is responsible because we don't.
Speaker 2:
We don't have automated decision-making, which means that in the majority of the cases when a decision is required, there's human oversight, and with that we've been able to move along. Of course, we protect every client data. There's no PII being exchanged with any commercial available model, so we protect all of those things available model, so we protect all of those things, and with that we've been able to come to a place where we're able to make progress, we're able to see the results of our own investments and they pay off almost on the first day. So it's really incredible. I tend to say that I'm ancient enough to have lived through mobile web, cloud and even others, and I've never seen a trend or a mega trend that has the potential of the impact that generative AI has. It's not the solution for every problem but where it can be applied, it's really massive the impact you can generate.
Speaker 1:
Yeah, very exciting times. And as CIO, I imagine you are customer number one for the new platforms and applications and tools that your colleagues develop. Talk about, maybe, some of the tools in the toolkit. You know your new data fabric, your relatively new architecture. You know event-driven architecture and what that means as a practitioner. How are you using your own products?
Speaker 2:
So the way this literally works is that Data Fabric to start with Data Fabric it literally creates and allows you to create a mesh of all the data that you have in the company and literally you can build an application that combines and reads and writes from most of your systems of record. What this means is that you know a lot of the times there's circumstances where you know you would hear many companies in different industries would tell you oh, every time a customer calls us, we have to go through four or five different systems to be able to answer their question or solve their problem. What Data Fabric is able to do here is you can actually build one application only that merges the data of all these different systems and allows them to interact with that data. It's actually funny. So I started in September 2019 at OutSystems and this was back then one of my first conversations with our CEO, who is a big visionary and kind of projects a lot of the vision that we have in our product and we were discussing and kind of testing how impactful data fabric would be. And all these years have passed and it's always a little bit ahead of the market. Data Fabric comes along and literally the beauty of it is that you include your credentials from every one of these systems of record and all of a sudden you see in front of you access to all the different data structures that you have behind these systems and then you can say I want to pick this one and that one, and this one and the other one, and you start combining the logic of how you're going to present them on the screen and when you have a transaction, so you write back onto them. It speeds the development of the application massively, because it's not just about putting all the logic, but it's manipulating all this data.
Speaker 2:
With GenAI now coming into place, it also allows us to query a combination of unstructured data with all this knowledge of different files that we have, together with structured data that we have recorded in systems, and you can combine the two to produce a certain result. So, for example, I am able to look into all my database of different use cases and customer success stories and map that according to the customer, the account, industry and be able to pitch a specific use case to a certain prospect in a certain industry. And that's how all these pieces come together when you're building and developing an application. I'm a big fan of how you can actually build something meaningful like this without having to be accessing different APIs and doing selects here and there. It's literally very visual and very graphical and you're able to combine all of this data together.
Speaker 2:
That is a massive improvement from a digital transformation perspective, because you know like when you have a problem to solve in a specific area, us CIOs could sometimes make the decision of going with a specific package that you implement in that section, but then you put another one here and another one there and all of a sudden you have multiple different systems with different sets of data that is integrated, but you have to have one place where you interact with that data for a specific business process, and that's where all these different new products come along. And the beauty of the evolution and the times that we're in right now is that being customer zero customer one means that we are constantly passing on real challenges, real data, to our product team and, in great collaboration, they also come and work with us on how to solve some of those problems and we keep on providing feedback that goes into the product. So it's very, very exciting times, amazing work and jobs that we are able to give to our people.
Speaker 1:
Well, fantastic opportunity, and you often hear about stories around. You know low-code and Gen AI implementations that have paybacks on day one, month one, so the ROI can be extraordinary. But what other metrics or KPIs do you recommend for measuring success, not just on specific low-code projects, but transformation in general?
Speaker 2:
formation in general. Yeah, so, like I tend to define the kind of, the core item is business impact. There's our business value. Business value I mean if you start with a literally a kind of dependent tree is usually generated by, by two branches. One is either and let's keep the innovation on the side for a moment, but you can either increase revenue or decrease cost, and so that is how it starts. And then, as you kind of open in more and more branches, you get to the specific kind of input metric that is going to generate that output.
Speaker 2:
For example, when we started with the customer support, what we wanted to do is to improve the agent's productivity. So we started by doing one thing which was alleviating them from something that we could solve on their behalf and so they wouldn't have to spend time with that. That's very critical because as we grow in number of customers, if we don't scale well, we'll have to always be hiring more and more people to answer new customer inquiries. And then the second component is to be able to. And then the second component is to be able to provide the agents with the first proposal of the resolution of the customer problem, so that they spend less time in investigation and research, and they just have to validate the proposal that we made.
Speaker 2:
This is an amazing increase in productivity. For example, when we are able to summarize and structure all the information that comes out of a discovery call with a prospect, we're saving hours of that rep or that SDR going through their notes summarizing everything that they've been through, listening to the call again because they do so many they might not remember all the details of that one anymore and literally we're able to do this 15 minutes after the call and it takes just a few minutes to go through all the fields and click save. And we're even able to provide a kind of a proposal of a follow-up email to that meeting. I imagine, evan, you have tons of meetings a day and how would you like to summarize and follow up everything and so on. So this saves huge amounts of time and great improvements in productivity. Those are just a few examples.
Speaker 1:
Yeah, just a few. We only have a short amount of time today, but I'm sure there are many, many more. And the other challenge are all the landmines associated with AI ethics and governance and data protection, which is really stopping a lot of companies from embarking on this journey. What's your philosophy around those areas or others that need to be really overcome?
Speaker 2:
So I am an optimist and I believe that this is out there for the greater good. I think that companies like OpenAI or Anthropic, who are even kind of training their models on constitution or things like that, there's literally an objective of improving the world with this technology. The idea behind some of these technology was that not only the big companies were able to use them, but, like as I saw in one example, like the kind of the pizza shop in the corner, I would be able to take advantage of it as well, and now it's becoming, like they, a daily tool that most everyone uses. One uses and I believe that the signs that I'm seeing are telling us that there's only good as an objective of applying this technology. I think it's allowing also any professional in their field to be even like superhero in what they do, because with the power of these, they become like almost superhumans.
Speaker 2:
So I know that there are questions out there. I think that if you start small which is kind of the approach that we took and you grow and you keep growing and as you grow, you pay attention to the evolution that's out there in the market and you make sure you continue to always to make sure that you're not kind of breaking any of the rules and the principles that you started with. Like I said, no automated decision making, for example. I think that we're all going to be safe and the world will be in a much better place as we take advantage of this. It's going to put pressure in different places out there in the market, but I think that we're all going to live in a better place by just taking advantage of it.
Speaker 1:
On that very optimistic note we'll end our discussion, but thanks so much for taking time away. I know you're super busy, have so much going on, but appreciate the insights and spending time. Look forward to keeping in touch onwards and upwards at OutSystems. Thanks so much.
Speaker 2:
Thanks, Simon. Thank you very much. It was a pleasure. Thanks so much.
