How banks in Chile and Argentina are preparing data for AI

Banks in Chile and Argentina are leveraging cloud-based data lakes as their first move towards harnessing the benefits of artificial intelligence (AI).

“We have about two petabytes of data [in the data lake] and we’re taking advantage of it with everything related to customer segmentation, generation of offers, user behavior to determine channel preference and various other topics,” Rodrigo Montaner, data director at Chile’s Banco Estado, told BNamericas on the sidelines of an event hosted by AWS in Buenos Aires.

The bank previously had about 60 units that handled data, but they have now been centralized in the cloud data lake. 

“We’ve been working intensively on having a corporate-level registry of the existence of data and going beyond knowing that the data is there, we want to take it to the cloud and have it consumed from the data lake as the only source,” he added.

According to Montaner, the data lake already has between 70% and 80% of the bank’s data. Banco Estado has already completed 95% of the organization's existing data registry and is currently purchasing a service from AWS for training and support. 

“We’re going to accelerate this migration with formal training and support [for the internal users of the data],” Montaner said.

The Argentine subsidiary of Banco Santander has also worked with AWS in the data ecosystem and it believes that the cultural change is the biggest challenge it entails.

Fernando Turri, head of technology and operations for the Argentine subsidiary, told BNamericas that the bank is “culturally a long way from where we need to be” in terms of the data use by the organization.

"We're moving towards a more federated world and democratic use of data. We believe that our CDO [chief data officer] should oversee the best models and technology, but this culture should also permeate the business areas. There’s a significant effort to reskill people and departmental leaders," explained Turri

Santander Argentina is one of the Santander Group 's global hubs for technology issues and it is seen as an example for its data strategy.

"We’ve moved away from older technologies and have been able to migrate all of our data technology to AWS, which was the technological solution we needed to become a data-driven company," Turri said.

The bank has been a pioneer in cloud migration, departing from outdated technologies that are still used in other subsidiaries of the Spanish financial group. According to Turri, the Argentine operation is being used as a pilot for other global technological developments.

ARTIFICIAL INTELLIGENCE 

Montaner at Banco Estado explained that the bank has already developed machine learning models for fraud detection, with nine models now in operation. It also recently launched a model that predicts service degradation. 

"We’ve developed a model that allows production teams to have early degradation alerts. We conducted an exercise with past events and we’re able to predict service degradation within three minutes," he added.

It also has models for audit and claims management. "There’s a new law in Chile that obliges us to settle a claim within five days. We’re also developing a payment recommendation model based on a set of rules that were previously manual and are now almost 100% automated," he underlined.

The bank is also taking its first steps in generative AI. Montaner stated that it is working on a pilot project to improve annotations on expenses, or descriptions that appear on account summaries, using these systems.

"Digital businesses generate annotations on statements that nobody understands and it causes a problem for us because the bank receives lots of complaints. So, we want to use AI to see if we can improve the annotations," he explained. 

The bank has already created an algorithm in an effort to improve these annotations, although at this stage, it is a more of an "artisanal" initiative used to handle complaints when customers call the bank, the executive said.

"In generative AI, we’re in the investigative process for multiple services and even multiple providers. We have challenges for the first half of next year to try to have something up and running," Montaner said.

Banco Estado de Chile is working with the top three cloud providers for AI.

Santander also has several AI pilots, although, for confidentiality reasons, it has not provided details. However, Turri said that AI will assist the bank in three areas: productivity of engineers in development, models for personalized customer service, and internal employee productivity.

"We will have a vision of how to embrace global AI technology because we consider the security aspect to be very important," he added.

The company is preparing to hold a workshop on AI topics in Madrid in November.