This new system is already being used in more than 600 branches. CaixaBank was the first financial institution in Spain to invest in using artificial intelligence in the banking business.
CaixaBank has developed an innovation project using artificial intelligence to manage returned direct debit payments. As a leading Spanish retail bank, CaixaBank manages 450 million direct debit payments every year and is the primary direct debit service provider in Spain.
When a bill is sent to an account with an insufficient balance, the software notifies the customer’s banking adviser, who then analyses the case and decides how to resolve the situation: to permit the bill payment, to reject it, or to postpone it for a few days until the customer has sufficient funds.
The banking adviser receives incident information every day at 8 a.m. This task is difficult to automate since each case is different and involves multiple variables that influence the final decision: the type of bill, whether it is a regular payment, the customer’s payment history, the customer’s connection with the company in question, etc. However, the task of managing bills takes up a considerable amount of the banking adviser’s time.
The same decision as a human in 99% of cases
Artificial intelligence can be used to create IT systems that can imitate human reasoning and understand the specific features of each situation to make the best possible decisions.
„In terms of managing bill payments, we needed to create an in-house computer system capable of processing each case exactly as an employee would. By doing so, the system understands individual issues and decides whether to permit the payment, to return the bill to the issuer, or to postpone the payment for a few days before analysing it again to see if the account can be debited. The system can also choose to send the bill to an adviser for a more detaled analysis.”, according to the press release.
„This technology project started with the design of the model by our highly trained team of data scientists, who specialise in advanced analytics and machine learning. Next, the system required four months of training to “learn” how to manage different situations. Eventually, CaixaBank obtained a model with 99% accuracy, meaning that the system virtually always thinks and acts like a human.”, the bank said.
In May, CaixaBank started to deploy the new system as a pilot project in 600 branches all over Spain (including the entire Barcelona network plus selected branches nationwide). In the first four months of operation, the artificial intelligence technology has already resolved a high percentage of cases by itself. If the model were to be expanded to the entire CaixaBank network, it would save approximately 82,000 hours of work.
By simplifying routine tasks, such as managing direct debit bills, the aim is for employees to have more time to focus on providing customers with a personal service.
In addition, this system helps to reinforce the direct debit management service. For some time, CaixaBank has provided its customers with a „smart” bill payment management service based on machine learning and big data. This tool, called “My Finances”, notifies customers instantly if they have insufficient funds to make a payment. It also provides information about the estimated date of future bills and notifies customers if the amount of a specific bill is higher than usual.
CaixaBank, a leader in developing artificial intelligence for banking services
CaixaBank is the most innovative bank in Spain in using artificial intelligence for financial services. The company started to invest in artificial intelligence in 2014 with the introduction of IBM Watson, a technology which CaixaBank pioneered by helping the system to learn Spanish.
The first commercial application was the launch of a smart assistant for foreign trade. This initiative was followed by a virtual assistant for all business areas, which is capable of accessing huge amounts of data instantly, interacting with users using natural language, and learning from experience.
A system with this degree of sophistication needs to be trained by CaixaBank advisers to know how to solve the technical issues put forward by customers in accordance with an extensive corpus of trade laws, country-specific regulations, internal regulations, etc. Similarly, the system requires ongoing training by technology specialists to improve its understanding of the natural language of people and conversation skills. Currently, the CaixaBank virtual assistant resolves an average of 20,000 enquiries every day.
CaixaBank also offers its clients artificial intelligence-based services through chatbots capable of holding conversations using natural language and responding to questions and enquiries. In 2017, CaixaBank launched the first chatbot in the Spanish financial sector for its imaginBank customers. Currently, the bank has two chatbots: Gina, for imaginBank, and Neo, for CaixaBankNow. Both chatbots are able to offer information on any matters related to CaixaBank products and services.
CaixaBank has recently launched its latest customer service innovation using artificial intelligence, whereby the chatbot service can be used to carry out real transactions. Specifically, since May, Gina, the imaginBank chatbot, has been able to help customers set up payment instalments. CaixaBank has designed two options for use. On the one hand, the service can be activated by the customer accessing Gina and requesting (by text or by voice message) that payment of a recent purchase be split into instalments. The chatbot immediately detects which of the user’s most recent transactions can be split into instalments and displays them on the screen. The customer chooses the purchase for which they need the credit and the operation is processed automatically. Gina notifies the customer of the operation’s status throughout the process.
In the second option, the chatbot activates the service independently. When a customer may need to split a payment, the bot connects to their imaginBank app, and Gina sends the customer a message offering the option.
The transactions that the bot assesses to manage the split-payment service are purchases with imaginBank cards made in the last week of at least 40 euros. The customer can choose to pay them over a term of three, six or nine months, under the same conditions as requesting as such conventionally via the app.
By using this new chatbot function, it is estimated that the use of the CaixaBank instalment payment service, with approximately 1,200 transactions per month, will grow by some 15%. CaixaBank’s mobile-only bank currently has 1.2 million customers
CaixaBank is the leading financial group in retail banking in Spain and one of the most important in Portugal. It has 15.6 million customers in the Iberian market and boasts the peninsula’s leading business network.
Technology and digitisation are key in the company’s business model, which has the largest base of digital customers in Spain (6.3 million). In addition, CaixaBank has developed projects that have marked technological turning points in the sector, such as the creation of the first cash machines that allow customers to perform operations through facial recognition and without having to enter their PIN.
Thanks to its digital transformation strategy, CaixaBank has become one of the highest-rated banks in the world based on the quality of its digital products and services. In 2019, the company was recognised as the “Most Innovative Bank in Western Europe” by American magazine Global Finance.
In addition, CaixaBank was recognised as the world’s best Private Bank for its digital customer communication services in the Wealth Tech Awards of the PWM magazine of the Financial Times group.
In 2018 the bank was chosen as the “Best Digital Bank in Western Europe” by the publication Euromoney; and its mobile banking application, CaixaBankNow, was recognised by the Bank Administration Institute (BAI) and the British magazine The Banker as the best mobile technology project in its 2018 Tech Project Awards.
„For more than a week now, ScoreRise enrolls daily hundreds of users through an innovative facial recognition interface. Enrollment takes less than a minute and it does not require presence of a human operator or video recording. And, of course, it stays fully GDPR compliant with help from Reff & Associates and Deloitte Romania.”