VIDEO TRANSCRIPT
This comes from Gartner's use case prism that was released last year for generative AI for banking. Now Gartner as well as a lot of other industry publications around the March twenty twenty three to August twenty twenty three time frame, they all released a list of kind of what generative AI is for in terms of different use cases that that have been considered by, thought leaders and think tanks and industry leaders as this is where this particular industry can benefit from generative AI. As you can see here, there's a lot on there, and there is a a broad categorization and, balance between value in terms of the value that it can create and feasibility in terms of difficulty of implementation. At the top there, you see creation of synthetic credit data, and at the bottom, you see payments exception processing. I would say that, you know, while this is interesting to look at in terms of the analysis that Gartner typically does when it comes to a particular industry and technology.
I would take this with a grain of salt because what we're seeing is is a couple different things in this regard. One is that there are some use cases that are specific to financial services that can be very valuable, such as synthetic credit data, anti money laundering models, and things of that nature, helping to evaluate, you know, credit profiles or making credit decisions based on based on AI, and models that the that the data team creates. All of those things are specific to banking as an example. Or in the case of insurance, you could talk about claims processing using AI to evaluate, the documentation that gets submitted as part of a plan.
However, that's not the whole story. There are lots of other use cases that aren't necessarily specific to financial services that can still add value to financial services organizations.
Two examples that I can think of off the top of my head would be anything that involves documentation. So customer onboarding. Anytime a customer is opening a new account or opening a policy in the case of insurance, they have to submit lots of documentation, sometimes in duplicate or triplicate. And, oftentimes, teams have to go through that documentation manually in some cases and extract important information that needs to go into systems.
With the advent of large language models, that can all be done automatically nowadays. Another use case that is particularly wide ranging and is near and dear to my heart because I spent a long portion of my career managing quality engineering teams and being neck deep in the challenges of delivering software into production. It is things like test automation, test case creation, test data creation, requirements analysis, production incident analysis. Basically, the entire software delivery life cycle has elements that can be sped up or it sometimes removed altogether by the application of AI.
Now that's not specific to financial services, but every bank I've ever worked at has had challenges related to making sure that requirements are, robust enough and fit for purpose, making sure that there's enough test data in order to be able to test all the different scenarios and ensure you have high quality tracking problems in production and making sure that you can tie those back to improvements that need to happen across the SDLC so that they don't happen again. All of those things are things that AI can can, can drive benefits in.
And on the right hand side here, I've listed out four case studies, and there's lots. Right? You know, there's the, JP or sorry. Chase Bank, Goldman Sachs, several, Citibank, several other, banks, particularly in the US have all published case studies of how they've been using AI and generative AI to do everything from create synthetic data for the purposes of testing or credit risk modeling to, creation of internal knowledge, internal knowledge repositories and chatbots to help team members find knowledge more easily.
But these four, I thought, were very interesting and worthy of sharing with you guys. So the first one there is, CBA. They have implemented Document AI, which is a product from a company called h two o to basically extract customer information from large volumes of unstructured documents. And what that helps them do is depending on the type of customer or the type of document, they're able to onboard, and service customers between fifty and eighty five percent faster, depending on the use case.
So anytime you're using, you're having to deal with large volumes of documents, considering an LLM powered solution to help you sit through that information and organize it and make it more readily accessible to the teams that need it is worth looking at.
Valley Bank in the US uses DataRobot to, improve the accuracy of their AML predictions. So they both use it to train AML models and also, reduce the false positives that some of those models generate by up to twenty two percent in some cases.
In India, federal bank, and I thought this one was really cool because this is a use case that a lot of banks have have at least considered doing, but have been reticent to fully implement because it's customer facing and it's inherently risky. They use a virtual assistant built on Dialogflow, which is a Google on the Google Stack as a way for the bank to interact conversationally with customers. So, basically, customers on the website can interact with Fetti, to handle a wide range of banking tasks and transactions. And that could be everything from changing the address on your account to opening a new account to asking questions about interest rates and the things of that nature. And then finally, Banca Mediolanum in in Italy uses SAS Via to develop credit scoring models, and that's helped them add or offer more lending products to more customers.