Customer Ignite Talk: Ravneet Shah (CTO, Allica Bank) & OpenAI

[music] >> Hi everyone. Very nice to meet you. Um, my name is Clem. I'm part of the go-to-market um, team here at Open AI. And I work with um, amazing customers like Allica Bank here. And thank you Ravneet for joining us. So, if you want to introduce yourself. >> Thanks. Thanks uh, for inviting me here. I'm Ravneet, CTO at Allica Bank. Um, Allica Bank, if you don't know, so we are an established SME bank. Uh, so uh, we uh, have been uh, operating in UK since 2019 when we got a banking license. And uh, we are building specific lending and business current account and deposit products for SMEs in UK. And our unique selling uh, point is like we do it through technology and relationship banking. >> All right, that's great to hear. And it's been one of the fastest challenger bank uh, growing in the in the UK. And I think like the organization you lead is very special because you're applying AI across every steps of your business from uh, product development to relationship management like you were saying. And also um, lending operations, right? So, um, today's conversation I think is mainly to try to showcase how modern regulated banks operate with AI and how they augment um, their team to simply go faster, better, and uh, better serve their customers. So, let me ask you that first question. Allica scaled quite quickly as a digital bank. Um, how are you thinking about AI in this current phase of growth, but also in what's coming next? >> So, uh, we've been on this journey like I think we started back in 2023. So, uh, made a lot of mistakes, tried few things. I think that was the time when we didn't know what to do, what are the use cases, trying to establish where we can fit it. But over the years and I think building from where we started, so I think we have a clear idea of how we can scale and we want to scale it as well. I think the first thing what we learned is like adoption across the organization. A lot of and myself, even though I come from tech background, there's a lot that we can we need to unlearn before we start learning the new technology. I think the first motto that we had was like to scale this technology, we need to make sure that everybody within the organization is adopting, they're changing the way they operate, be it in operations, distribution, technology, product, finance, everybody. I think that was the drive that we went through. So we increased the adoption across the organization. So last year I think it was like 25% so we the today we have a median workday of 77%. So that's what what we are approaching. The second thing that was quite key and I think it's quite key for any technology organization is how we are actually building our product. So that is more about our product engineering organization, our operating model there. So we've been working through what what needs to change to be actually ensure that we are leveraging AI in the best way. So obviously what works for the rest of the organization doesn't work for the product engineering squad. So we had a very specific operating model there, so which we looked into. And the third part was like our products itself. So I think the way we have been using product historically, traditionally through software engineering applications is significantly changing. So what we could actually do can not could not do with the traditional machine learning or even software applications. I think we can do it with AI. I think that added another layer and I think it's just like a motor within the the business. We keep on repeating within the organization. Um as in we need to think differently how we are building our applications using agentic application. That's the third third way that we used. >> And that that's actually very interesting. I think one of the things you mentioned there is a specific operating model, especially the one that has allowed you to I think um do more than like 3,700 deployment last year, which is quite impressive given the size of the team and how you've been working. Um can you just tell us a little bit more about like how you've organized the team and build the team to foster that innovation, but also allow them to build fast? >> Sure. So, uh we are not a huge team. So, we have like 100 engineers uh or less than around 200 uh colleagues within the product engineering team, including product, data, everybody. Um so, we historically operated in using Spotify model, uh where we had cross-functional squads um product, data, uh software engineers, back end and front end and as that's and designers working together. So, um and that that model worked perfectly well for us uh previously and we've been shipping uh uh our products using that model. With AI, we s- saw that that needs to change and we've been observing over the last year as well. So, what we changed is like we had like a blueprint of our squads previously, which we have changed significantly. We've reduced the size of the squads. We call them squadlets now. And then the squad structure differs from one team to another based on on complexity and the nature of the product that they're building. So, we also uh found out that the hand-offs that were required previously is no longer needed. So, we believed in the concept of T-shaped model. So, having a deep specialization specialization skill, but then you're building your adjacent skills. Obviously, it's skill that's coming the new term that's coming. And I think that is something that we encourage within the organization. So, rather than having back end, front end, and separate tested, so we combine the role into one. And then we had the product role as like we used some in some of the squads we used to have like product owners and product analyst. We combine the two role the two roles into one again. And even we evolved that role where we have in some squads we have product representative could be a product owner or a product analyst, but then we had some squadlets where we just have product engineer. So, someone who can do product as well as engineer at the same time. So, I think that's that's how we are evolving. I would still say we are in a journey. We've experimented that in some of like very few squadlets, but that is the kind of methodology. But towards the end of the year we are going with an objective that probably our product, our design, our engineers all engineers can ship the code today, but our product and designers will be able to ship the code to production by the end of this year where it it is happening in some of the squad not at the same not to the extent that we would like to, but I think we'll see an improvement and and over the year. And what works well is like everybody is excited about that change. >> And thank thank you for that. And I think when we're catching up before um, to talk a little bit about like this content and what we're going to talk about like, what one thing that I have loved and I love to hear is, um, I think Arica's a specific approach that allows them to scale fast, be and avoid the governance drug that usually happen with AI because of that, um, pod system that you have when you have those groups where everyone that needs needed for the decision or pushing the product is in the squad. So, the product can go from creation to publishing without needing to leave the squad unless absolutely necessary. And so, it allows you to go faster and to iterate faster. And I think that's something we need to remember as well. We need to unlearn, but we also need to unstructure the way some of the teams have been structured to be able to rebuild them to move faster with that new technology. Um, when you think about another area of use talked about about the landing and underwriting aspect of the business, um, I'm curious how is AI changing that within your organization right now? >> So, that's quite an interesting one. So, um, lending is our core business, um, and the customers that we deal with and the whole lending process is quite complex. It's not automated. So, we deal with introducers, brokers, um, and within asset finance segment, we receive a lot of applications through emails. We have portals, but our customers or brokers or introducers, they don't like to use that. And even with the broke with the portal itself, you sometime get like half-baked information or there's missing information. And I think that's probably where, uh, rather than just forcing our end customers to, uh, change their approach, we changed our approach slightly. Uh, so, we've introduced a kind of agent where, um customers can send us emails, the brokers can send us emails, but then the agent actually look into the information that comes through the email, and they can um find the missing information, so they can ask more information from the brokers before the information gets fed into our portal. So, I think that journey is kind of automated, and by using the combination of deterministic and non-deterministic agents, so we've got to a position where some of those applications, we've reduced the time to decision to like less than 7 minutes or 12 minutes. So, that's what we are heading towards. I think we use the technology to actually improve the process rather than just changing the experience of the customers. There are more use cases that we've been using. I think the the the idea that where we are going is like we introduced we've been introducing technology over the last few years already, but now where the processes were manual could not be solved through software applications, we are just rethinking them and see like, "Okay, where can agents help?" Obviously, there has been harnesses in in place, but I think there's there's a drive, and we see a huge opportunity there. >> And I think it links as well to what was said right before we joined the stage around like the the example of voice and banking, and the idea of like some customers are willing to change, some are not, and so it's important to meet the customers where they are and to make sure that we improve the overall experience using the technology that is accessible to us, and that we build the technology that is not accessible to us. I think that's one of the things that has been talked about by everyone is using tools like Codex allow you to build technology that was not available before and easily as well. That leads me to just another question that's because you've mentioned relationship banking, and I know that it it has a core of Addiko's business model. Um where did you see AI having an impact in that area as well? >> So, we had number of debates internally on this as in like how can we use the technology there? Obviously, there's a lot of conversational chatbots that we could have introduced, but considering relationship banking was a unique selling point, we didn't want to change it. We didn't want to replace relationship banking by a chatbot. I mean, we have messaging, but that's a kind of core functionality that we need to that we wanted to bring. So, we came up with the idea, and we're still iterating and developing on that. So, rather than replacing relationship bank relation relationship managers by AI bots, what we wanted to do was like actually support them with the insights and the information that they need to have about their customers. I think that's what how we are thinking about it. >> Yeah, and I think it's augmenting the employees so they can do more and focus on the work that matters. And I think it's spending more times not necessarily more time, but better more quality time with their customers and where they're trying to build relationships. >> And I think it's also about like better context as in so they so that they don't spend a lot of time understanding and building information about their customers, but rather we provide them context using AI enough so that they can drive that conversation. >> All right. Um one last question before we wrap up this session, and I think it's something that is maybe a bit idealistic, but if you're looking ahead the next 12 months, next 24 months, like we all know at the crazy pace at which the technology is evolving. Um where do you see Allica Bank taking their AI journey and where do you see AI taking the banking or the lending journey that you're part of? >> So we are evolving. We are on a journey, I guess, but I think now, if you had asked me that question 6 months ago, I think we were still experimenting, but I think we have a clear direction like from product engineering world what way we want to get to. So you mentioned like we had 3,700 deployments last year and we are we want to double it, but then doubling it just could be playing with the numbers or gaming the system, but what we actually want to get to is like increase the product increments for our customers both just customer-facing product increments and internal-facing increments as well. So that could be related to risk and compliance and security. So I think that's what we are targeting at the moment. So obviously we want to keep the speed of our service. We want to make sure that the quality of our service stays and improves while we're doing it. >> That's great. So just better serve the customer and go faster and being more efficient while doing so. Like it reminds me a little bit of what we were saying about OpenAI at the beginning. It's like better models, more cost-efficient, and just building for you. So first, thank you very much for coming here today, but also thank you very much for building on our technology. Thank you for the feedback because I think we've talked twice and you've already gave me like two three piece of feedback of things that we need to get better at or we can help you with. So thank you for that. I appreciate it. >> [applause]