Trading Tomorrow - Navigating Trends in Capital Markets
The Impact of Low-Code Technology on the Financial Industry
This episode of Trading Tomorrow - Navigating Trends in Capital Markets explores low-code's potentially transformative impact on the financial industry. Low-code has been presented as an innovative drag-and-drop coding method that seamlessly integrates with traditional programming languages like Python, dramatically accelerating application development and maintenance. In this episode, host Jim Jockle of Numerix and Brian Sathianathan of Iterate.ai discuss if low-code technology is well-suited for the financial industry. The discussion also covers the challenges associated with legacy infrastructure and regulatory compliance, and if low-code's modular design is a solution that enhances scalability and security.
This episode of Trading Tomorrow - Navigating Trends in Capital Markets explores low-code's potentially transformative impact on the financial industry. Low-code has been presented as an innovative drag-and-drop coding method that seamlessly integrates with traditional programming languages like Python, dramatically accelerating application development and maintenance. In this episode, host Jim Jockle of Numerix and Brian Sathianathan of Iterate.ai discuss if low-code technology is well-suited for the financial industry. The discussion also covers the challenges associated with legacy infrastructure and regulatory compliance, and if low-code's modular design is a solution that enhances scalability and security.
Welcome to Trading Tomorrow navigating trends in capital markets the podcast where we deep dive into the technologies reshaping the world of capital markets. I'm your host, jim Jockle, a veteran of the finance industry with a passion for the complexities of financial technologies and market trends. In each episode, we'll explore the cutting-edge trends, tools and strategies driving today's financial landscapes and paving the way for the future. With the finance industry at a pivotal point, influenced by groundbreaking innovations, it's more crucial than ever to understand how. Thank you ahead of the curve. Join us as we engage with industry experts, thought leaders and technology pioneers, offering you a front row seat to the discussions shaping the future of finance, because this is Trading Tomorrow navigating trends in capital markets, where the future of capital markets unfolds.
Speaker 1:For many in the finance industry, there has been a real push lately from legacy platforms to SaaS options, but building these platforms and processes can be time-consuming and expensive, creating a need for rapid, scalable and efficient development. That's where the idea of no-code load-code comes into play. Both are development approaches designed to simplify the process of building applications. As the name suggests, low-code might include some coding, while no-code includes none. And joining us to discuss is Brian Sathianathan, the co-founder, chief digital officer and chief technology officer at Iterate AI, an enterprise innovation ecosystem launched in 2013. Major companies across industries like Ulta, beauty and Circle K, leverage the intelligent low-code capabilities invented and patented by Brian and his team. Interplay dramatically speeds up and simplifies digital and AI-based innovation. The largely bootstrapped company is highly capital efficient, growing their company by more than 400% in the past four years. Brian has also spent time working for Apple, avid Media and Turner Media. Hi, brian, and thank you so much for joining the podcast today and welcome.
Speaker 2:Thanks, Jim, for having me Excited to be here.
Speaker 1:Well, just to start off, perhaps you can explain to the audience what is low-code.
Speaker 2:So think of low-code as a method of basically building applications or programs for your computer or for your cloud right, but in the traditional method way of building code or writing code is you have to write textual code.
Speaker 2:Low code? You're sort of dragging and dropping these Lego blocks, sort of looking components right. It's more of what I would say you are essentially drawing code opposed to writing code, right? This has a number of bad business benefits. One is you already tend to use existing components, so it's a faster development. And also, from the business leadership side, you get to something where you can bring things to market much, much faster. In fact, for our platform we benchmarked it it's almost 17x faster, right. And then overall ongoing maintenance. It makes maintenance also easier because all these things are like broken into these Lego blocks, so you can always go do diagnosis and fix things when it's all nicely componentized and there's a lot of reusability too. Yeah, so that's what low-code is.
Speaker 1:So you're saying it's better than Python?
Speaker 2:No, actually what happens is you can actually write Python capability inside low-code, but low-code is where you know. Imagine a lot of the Python stuff that you are doing physically today, like in textual code, are already componentized and you can go to an editor and start dragging and dropping them. So the beauty is you can combine traditional languages like Python on top of it, so it sort of supercharges you right, which is where it becomes very cool.
Speaker 1:Wow, okay, so I did not know that. And the only reason I bring up Python because I think in finance Python has taken over has been the revolutionary code of choice, if you will, over the past, let's call it five to seven years, but so, just focusing on the financial industry, you know why is low-code an attractive technology for this market segment.
Speaker 2:That's because I think what's really interesting in the financial industry is, just like every other industry, the things that's very important or very unique in the financial industry is that there is a lot of data involved, right, especially financial data transactions, or even like things like banks and insurance companies.
Speaker 2:You are looking at, you know, like documents that are converted or OCRed into unstructured text, right, and then most of them is either financial data, health data, whatever data that's critical, right? So now the idea about you know, taking the data and building applications, and every customer, every player in the financial industry their problem to some degree is unique. It is basically like very generic to some case, but also unique to their industry and their problem domain. So the beautiful thing about low-code is you can take the. If you look at the problem at whole, you can look at things that are generic they can be already drag-and-droppable blocks but then the problems that are unique, you can stitch these blocks together and write custom applications. So now that gives you the best flexibility in terms of bringing something that's already generic and customizing that to become unique for an industry. That's where low-code is really powerful in the finance industry, right, I mean, I can give you tons of use cases around that.
Speaker 1:Yeah, oh, I'm going to get to that because I want to make this as practical and actionable for our listeners as possible. But you bring up the concept of speed and agility and even in your own work you gave a little bit of a benchmark there in terms of gains and productivity there, in terms of gains and productivity, but specific to financial markets, what has the benefit been in terms of developing new financial products and services, and perhaps you could share a couple of use cases that you have seen so I'll give you a couple of use cases, jim.
Speaker 2:So one of the things that we are doing in the banking, specifically in the banking finance industry, is that typically there is a lot of processes that are not going through STT, the straight through processing. So all the way from the time you deposit a check to the time that the check actually goes through deposit, or the time you actually apply for a loan till the time all the loan gets applied, or if you're in an automotive industry, like you go and apply for a loan and banks are competing for it, and the internal processes to get that loan processed within five days, right, or seven days, all those processes have a lot of manual steps, right, but if you look at it, they're like basically glorified workflows. Most of the time what happens is these workflows will have sections parts of it human-done, human-verified, parts of it you know some business rules, and parts of it is some legacy systems involved in. The data is sitting in all these silos. What's really fascinating with low-code is if you build a building block that talks to the legacy systems and if you have the building blocks that can use machine learning and AI and can do automated decision making, you can chain all these blocks together to create workflows that are straight through, and that's something that banking and finance industries really like working with us because we are able to do that.
Speaker 2:But yet, you know, this is all running in a platform under container, secure and compliant, and also running on-premise, because not all banking want to use cloud. They want to run it on a private cloud which is super secure and gone through all the audit processes and so on. That's why this whole low-code is interesting. Now you can take various disparate elements together and you can string them along and teach them together in a very graceful way. That was one that straight through processing was one of the use cases. Another use case that we are doing is basically, like you know, when certain type of financial products are recommended for folks, it actually looks at their prior analysis and it goes through a personalization process right, especially in the private banking management area. So we are actually building software using low-code that helps these advisors recommend solutions much gracefully and manage them, but that's all done through low-code processes.
Speaker 1:So arguably and I've heard this said before and this is not me saying it is that we tend to think of financial institutions as transfers of money, but they're just as much a tech company as Apple or Microsoft. At this point, I mean, what is the thrust? Is it improved customer experience? Is it personalization? Customer experience, Is it personalization? Is it ways of automating processes? What would you say the key driver that this technology kind of fits into within a financial institution, or is it all of the above the simple answer.
Speaker 2:It's a great question, jim. First of all, I think the simple answer is it's the all of the above, but it depends on. It differs from company to company. Like especially traditional banking institutions, the backend automation is where cost savings in backend automation is a big factor for them, because that's significant part of the cost and cost reduction through automation is critical. And then ability to serve customer requests in timely manner. So so there's a lot of trust in, there's a lot of trust in the automation back end, especially with big banks and you know larger companies, insurance companies.
Speaker 2:But now on the on the customer side, banks are also trying experiences with better experience, consumer experience, right, but at the same time, some of those things will have to go through a lot more scrutiny because you're in the interface between the bank and the customer and you want to maintain the same quality levels at the same and, especially when it comes to AI, avoid things like hallucinations and all that stuff. But the biggest trust, I think, is in the back, especially where processes were involved as part of the processes were involved as part of. The processes were automated, part of the processes were manual. This allows you to create a unified platform to stitch things together in a very graceful, timely manner, that is maintainable, that is compliant, that you can easily extend into the future, right and providing a fairly decent total cost of ownership, because now all these things, all these price costs, will come down.
Speaker 1:You bring up an interesting point in terms of total cost of ownership and, arguably, while the financial institutions have been very innovative in some ways, in other ways they haven't been. And I think one of the biggest challenges is probably legacy systems, as you alluded to before. Are there barriers in utilizing a low-code strategy? With legacy systems, where perhaps they were architected with a poor API strategy, it's very hard to get data in and out within them. You know what is the impact of legacy systems in adopting a low-code strategy.
Speaker 2:The legacy system impact is very big. Right, that's one of the biggest, but not only adapting low-code, but adapting any new technology within banks. Right, because banks have a lot of legacy systems. Even if you look at like 10 systems in a bank, probably, based on what we've seen, five or six of them are completely legacy and probably don't even have APIs.
Speaker 2:Right, Probably, and even probably on-prem, on-prem, on-prem, mostly on-prem, right. But the biggest challenge, though, is like over the last few years, right? I mean, the good news, of course, is that the banks have spent quite a bit of money and investment in somehow connecting to these systems that are not proper API-oriented, because they needed that type of connection for whatever connecting two legacy systems together or having some business flow work right. So I think what's interesting with low-code coming in is you can sort of structure some of those things and you can put those capabilities into these low-code adapters. In fact, some of the banking processes that we are doing right now involves actually connecting into a lot of legacy systems which don't have any API connectivity. So you have to look at alternatives like either file-based or event-based or some kind of a technique to talk to these systems. But that, again, is interesting because in a low code once you build, because if you look at all these so-called legacy older systems within banks, some of them are homegrown.
Speaker 2:It depends on where they come from right If they are homegrown. But some of them are like traditional software, older vendors who've been there for the industry for a long time. They've never upgraded their software right, but they are third-party vendors. But the good thing about local is, once you develop one connected to one of the vendors whether it's API or like whatever file base you get the ability to sort of repeat a lot of that stuff, because it's the same component can be repeated. So you get a lot more repeatability and reusability in cases like this. So that's why I think low-code is really interesting, because now you can componentize what you do and then also expose some of these experiences into business users more gracefully.
Speaker 1:How accessible is low-code? I've played with certain things like Microsoft has Power, automate and simple connectors within the Microsoft ecosystem in which you can automate your own processes on a personal level and whatnot. But as someone who has a degree in politics on a personal level and whatnot but you know, as someone who has a degree in politics, I would say my ability to use low-code was pretty limited. So you know, obviously that was a couple of years ago. I played with it. You know how is low-code advancing, what is the skill set of the individual and how widely accessible to someone like me would you think low-code will become eventually?
Speaker 2:I think all code is eventually coming to a point where, I mean, depends on the time horizon If you look at a long time horizon about 8 to 10 years, I think, because of the nature of large language models and where generative ai is going, most code is eventually going to be generated, right? And then you will be interacting with these systems in in plain english, just like how you interact with your chatbot, right? So that's where everything is going, and code is going to be generated, and then maybe code can be generated as low code. In fact, we have some capabilities in our system where you can go and type in things, and it could generate codes in blocks for you, right? And in fact, we're talking about legacy. One of the capabilities now we are bringing in we've already bought in our platform is, you could say I want to do write an sap connector or I want to write a salesforce connector for this job, it it'll automatically go. The generative AI will build a piece of block for you to make it much faster, right?
Speaker 2:So now the question about low code and no code right? No code has always been designed for the business user, right? But what's happening with no code, though, is no code is so high level, it it's it becomes impossible and it's non-pervious to customize it because it's a. It's a fine trade-off between providing deep customization that which most of these enterprises need, right, because there is security, there is compliant, there's authorization, access and every, every, every business process is super, super complicated within these large companies. So using no code is not a good option, right, because you need to provide customization. It's hard to provide customization when code is not involved. So that's why low-code platforms play a really powerful way, because they give you existing customizable components. Some of them you need code to work with. Some of them you don't need code to work with. So it gives you that flexibility to customize things.
Speaker 2:But what's interesting, though, is even in our own solution, we provide a low-code platform mostly targeting application engineers right, not the business users but then on top of it, we have applications that are built on top of the platform.
Speaker 2:For example, our local platform is called Interplay, and then on top of that platform, we have an application, especially for the banking, called OCR Manager Workflow Managers. So those applications actually have business user customizable blocks, so you can go into an app and upload your document and say it's a loan document, whatever the OCR and AI deep learning will kick in, you can actually use large language models to ask questions to the document and then, if you want to build a business workflow like I want to send this document to route it to somebody you can build all these things visually and that is Pretty much pure no code, no programming at all right? So that's something. That's how we are currently. So you have this fine balance you build, you implement a low-code platform for your bank across your bank, and then you build an application on top of it that provides no-code interfaces for folks to the business leaders and folks to customize things.
Speaker 1:So you know, I guess one of the questions that comes up in my mind is certain languages have suitability. So if I'm going to take advantage of GPUs, I'm writing in CUDA. If I'm writing financial models, I'm writing in C++. If I'm really old, I'm writing in Fortran 77. But what about the arguments around suitability? Or is no code or low code an opportunity to be a wrapper on top of some of these types of languages, making them more accessible and improving development time? What's the right way to think about that?
Speaker 2:I think what you said later you sort of hit the nail on the head, right? So low-code becomes an opportunity to provide a very powerful wrapper for application engineers not system engineers, right Application engineers to take advantage of these underlying technologies. Right, I'll give you some really practical examples. So today, our Interplay platform, in our platform we support the ability to write like drag and drop components. But underneath, if you need to customize something, you can actually write code in Python or Nodejs, two popular languages. So you don't need to learn a special scripting language to use our low code. Right, it's all standard programming languages that you already, folks already know, right?
Speaker 2:The other thing that's interesting is you mentioned CUDA. So one of the things that we've done is say you are doing a data analysis application, or let's say you are training an LLM. You can actually drag and drop and train an LLM. In our platform you can do the RAG. You know the RAG operation. The entire RAG operation can be done as drag and drop. So underlying, underneath it it'll actually talk to CUDA libraries, but if you're using a CPU, it'll actually talk to the Intel libraries underneath the code and it creates a wrapper, right? For example, we've wrapped NVIDIA's Rapids data analysis engine to do drag and drop right. We've done similar things with Intel's OpenVINO and others right. So this becomes a really nice conduit for you at the very top to do simple operations, and underneath the complexity is taken care by the local platform. In fact, I do have a lot of examples I can actually show. I mean not in this podcast, but I can share with you later.
Speaker 1:Oh cool, yeah, yeah, yeah. So we, you know we talked about the benefits, so what are some of the challenges? You know, especially financial institutions that are perhaps prohibitive to adoption at this point.
Speaker 2:See, I think that there are real challenges. And then there are perception challenges, right? The real challenges, like you said is said, is, you know, legacy, one right, um, and the other, the other, of course, is the ability to run and integrate across a lot of the platforms because, you know, some of these financial institutions have a very closed environment. So, you know, talking to systems and I mean not only no systems can talk to each other, so that connectivity is definitely an issue. And then, of course, the industry moves slowly and it's a very large industry with a lot of compliance. So those are some of the issues.
Speaker 2:But there are also perceived challenges too, because low-code has been around for a while, right? So if you look at traditional senior leaders like it or like cios or any of them, if you talk to them about low code, the first thing they say is or low code is slow, or low code, you know, because there is so much abstraction, it's not running at the system level, or whatever. It's not fast enough, or um, or or it's too much abstraction, right? Or it's it's just a toy, right? My platform is a serious system and and I need heavy scalability. But those are older concepts, those have actually changed over time. But if you look at our low-code platform, the traditional Nodejs on a four-core CPU could probably do 5,000 to 6,000 simultaneous HTTP sessions. Our low-code platform can do 56,000 sessions on the same machine. That's because what we've done is we've optimized it. So there's a lot of these perceived challenges that have been coming from years and years of prior implementation from low-code. So I think those are some of the things over time will get cleared out.
Speaker 1:Clearly they haven't read their programming books from the 70s and they forgot about Fork yeah.
Speaker 2:I mean not the 70s, but even like what happened was, I think, in the 2000,. In 1999, 2000, 2005 timeframe there was a lot of attempts to do low-code right. But just like the evolution of technology, the first attempt always is not a great attempt. So people get a view of a taste of those attempts and it sticks in people's memories for a long time. But that's changed over the last 20 years pretty heavily.
Speaker 1:So I'd be remiss in talking about financial services without mentioning the two dreaded words security and compliance. They're always the top concern when it comes to financial institutions. Are low-code platforms safe and secure?
Speaker 2:They are. They are In fact I would even argue they are more safe and secure. I mean, of course, the benefit of a low-code platform is because it's all highly componentized, right? So if you do analysis and structure and set up a compliance program like all these components in the low-code platform is already certified, then you can continue to trust and build on top of them, because now it's at the component level opposed to at the individual code level. So overall maintenance and compliance is much easier with a low-code platform compared to a traditional coding platform, because in a million sublines of code you're going to run an automated tool and you don't know that could be something hiding somewhere, right? So I would argue that low-code platform is, because of its abstraction and assuming each component is prior certified previously, as long as people honor that structure, it's much more easier to maintain and control and apply various compliance capabilities to it. And it's secure because all the methods are containerized and so on. So definitely it's more secure.
Speaker 1:Well, unfortunately we made it to the final question of the podcast. We call it the trend drop. It's like a desert island question. So if you could track only one trend in low code and finance, what would it be?
Speaker 2:The integration of generative AI. Just like everybody automated agents, automated generative AI powered automated capabilities. Right Because generative AI-powered automated capabilities. Right, because generative AI is becoming more and more powerful every day. I'll give you a practical example.
Speaker 2:Like three months ago, in our entire OCR you know, object character recognition technology we rewrote our entire. We've been providing OCR technology for banks for quite a while, but we rewrote all of our OCR technologies using large language models and it's amazing, it could do every language. So things are changing and now large language models are getting better and better doing and building doing something called agent work. Agent means you take a problem and the agent will look at the problem and one part of this model agent will solve the problem and it'll give it to another agent, just like how different people with different skills handling different tasks. So I think the trend is going to be all these agent capabilities large language model generative AI capabilities are going to be integrated into low-code and making it super drag-and-drop. In fact, we already have some solutions already and we are integrating more and more.
Speaker 1:I was going to say I'm not surprised they're called agents, because that's what they were called in the matrix. Well, brian, I want to thank you so much for joining us. This conversation was enlightening and I know our listeners are thoroughly going to have enjoyed this. So thank you so much.
Speaker 2:And thank you, jim, as well. I had a great time talking to you as well, great questions, and hope your listeners like it. Yeah, absolutely.
Speaker 1:Coming up next week on Trading Tomorrow, navigating trends in capital markets. It feels like you can't go a week anymore without hearing about blockchain. It's being utilized more and more, especially in the finance industry, but the question remains will this technology play a pivotal role in the future of financial markets or will it fail to be adopted? For more insight into this topic, we hear from blockchain and fintech trailblazer, Igor Telyatnikov, the CEO and co-founder of AlphaPoint. It's an episode you can't miss.