Artificial intelligence is one of the spearheads of digital transformation. Its potential to solve, efficiently and quickly, repetitive tasks of low value offers companies the possibility of reserving staff for more "creative" and critical tasks.
Banking, both traditional and digital, also found in this technology not only an ally to respond to user attention through chatbots, but also to resolve a key stage of their business.
It is about onboarding, as it is called in the jargon at the customer's registration, in which the system can check whether the data provided is real.
But it is also implemented in the scoring. That is, the credit rating, a phase in which "traditionally" an agent was responsible for reviewing documentation (salary receipts, financial history, etc.) to determine if a person could access a loan or credit card and at what amounts.
While this technology is used by fintech because its services are offered (almost) without human intervention, at least 80% of the "traditional" banking market is already taking advantage of some of the benefits of artificial intelligence.More than a "Veraz"
The credit rating process has a single input: information. Obviously, the more data, the credit profile of a client is defined more precisely, which is vital to determine your risk and, consequently, the amounts for a loan or the limit of your card.
At present, banks not only use the services of a financial information bureau (where the "Veraz" brand became a generic one), but they appeal to everything that the Internet "knows" about that person."We developed an engine on which, based on a controlled amount of variables, a model of Behavioral Scoring was built, which allows us to apply artificial intelligence for credit assessment," says Adrián Mastronardi, CTO of Finance Wenance.
"We also add behavioral variables according to the user's income channel, such as social networks, search engines, email marketing, etc. This achieves a level of appropriate discrimination, predicts risk in relation to default and adjusts the product offer for each individual ", adds Mariela Sandroni, Marketing Director of the firm.
However, these techniques are not limited to fintech: traditional banks are already implementing them.
BeSmart, a company that offers software solutions to 80% of the market, states that in addition to traditional data (such as willingness and ability to pay, amount of income and job stability), entities use millions of data to build an "X-ray" more exhaustive of its customers, namely:
- How much you buy: number of times, average ticket, maximum expense.
- When ?: schedules, days of the week, months.
- What ?: items, products, high-end or low-end.
- Where ?: zone, channel (face-to-face, web or mobile).
- Attitudinal data: fidelity to brands, claims, comments on networks, preferences, affinities, etc.
"Unlike conventional methods, where typically 20 or 30 variables intervene, the new machine learning techniques allow us to build models with thousands of variables," explains Adolfo Kvitca, director of BeSmart, at iProUP.
All this information matrix is not only used by fintech, but the main banks in Argentina are slowly implementing innovations to improve the credit profile of their customers."All entities are turning to these new technologies and applying them in different areas, such as complaint resolution, risk analysis, consumer trends and fraud," remarks iProUP Milagro Medrano, manager of Macro Institutional Relations.
In the City Bank, they also work on this technology and are adding new data to provide "customized" services for each person.
"As we evolve in Artificial Intelligence, we will be able to identify the client to consider variables that allow us to scoring and personalize the offer, and contemplate their activity in social networks and other sites, to enrich the profile and improve our proposal," the executive remarks.Speed and precision
The amount of data analyzed through big data and machine learning techniques not only allows banks to know the "risk" of their potential clients. It is also key so that the registration process is much faster and can offer more services.
"The ease and speed to develop models makes it possible to build scores tailored to each company, which even learn and adapt continuously to changing market conditions," explains Kvitca, of BeSmart.
According to the executive, "the impact on accuracy is dramatic, since it allows approving previously discarded customers, improving the allocation of credit limits and optimizing collection actions."This speed is highly appreciated by fintech, who promise to register customers in just minutes through an app.
"Times are reduced, costs are optimized and this allows us to invest in more innovation." The application of big data and machine learning allowed us to be more efficient in origination: we managed to optimize the process 61%, "says Adrian Mastronardi, CTO of Wenance.
According to the manager, these innovations reduce loan approval times from 72 hours to less than 30 minutes.
With regard to this point, Artificial Intelligence is favoring the contracting of banking services by people with little financial history.
"AI algorithms improve risk management, so entities can increase profitability by correctly rating each of their prospects, granting larger amounts to those with higher probability of payment," explains the BeSmart director.
In this same sense, Wenance stresses that the precision of new technologies enabled them to reach a greater number of people.
"We can include clients from the THIN segment, which are excluded by banks and credit institutions, improving the approval rate significantly," Mastronardi explains.
In addition, from the fintech ensure that these new modalities also resulted in benefits for users: the amounts of approval multiplied up to twice.Other uses
Artificial intelligence in banking landed strongly in digital assistants or chatbots, who interpret the demands of users and "filter" the simplest questions to derive the most complex to human agents.
"Our first stage of Artificial Intelligence contemplates that the Cognitive Seeker and the Virtual Assistant are being trained to answer questions about the products and services we offer, under the concept of 'consultative advice'".
"The ChatBot 'Alicia' simulates real conversations, answers the most frequent doubts of users, such as those associated with loans, accounts, procedures, etc. The idea is that if you do not understand the communication, call an operator", explains Patricia Parente, Manager of Commercial Intelligence of the City.
From the Macro, in addition, point to that directly chatbot is the one that approves the requests of products of the clients and assigns limits of credit.
"That our Virtual Assistant evolves to 'transactional' towards clients and not clients, contemplating the 'scoring', is one of the next objectives that we already consider within our Future Vision Plan," says Milagro Medrano, of the Macro.
On the other hand, from Wenance they affirm that, being a 100% digital entity, they implement automatic learning (or machine learning, one of the branches of AI), in different stages.
"We use it for the verification of identity and electronic signature, to compare the selfie that the client takes out with the photo of his ID," says Mastronardi.The transformation does not end there
In addition to artificial intelligence, banks are using other technologies to facilitate the use of their services.
One is that of biometrics, increasingly used to enter mobile banking without completing a username and password: simply by scanning the fingerprint, recognizing the face or reading the iris.
"Our goal is to streamline customer operations and improve the experience in the service model, mitigating fraud based on its univocal identification," points out iProUP Marcelo Abella, manager of Commercial Planning and Electronic Channels of Supervielle.
Macro and Galicia also incorporated the recognition of faces through a selfie for the registration of clients in their mobile applications, where the Artificial Intelligence compares the photo of the DNI with the one that the user uploads to detect if it is who they say they are.This approach of traditional banks to digital users does not end there: according to iProUP, they offer incentives for digital channels as a way to stop the advanced fintech.
In this way, the entities propelled that more than 50% of the fixed deadlines today are carried out by these means, with peaks of up to 80% in some entities.
An example of this is the sale of dollars: some banks choose to offer a better quote for the customer in the electronic channel, while others extend the hours of operation beyond banking hours.
In the first group, are the Credicoop, the Supervielle and the French, Nation and City, which sell the green ticket about 5 cents cheaper than in the branches. Thus, in the Buenosairean bank, 63% of the operations are already carried out by home banking.
The HSBC, for example, allows the sale of currencies outside bank hours by electronic means, in which it can be operated between 8 and 18.
Regarding fixed deadlines, Guillermo Jejcic, Marketing Director of Itaú, affirms to iProUP: "We have a preferential rate, which for high income is 40% above what we offer on blackboard through traditional channels". Thus, 70% is already done digitally in that entity.The City, the Province, the BIND, the French and the Macro, among others, also provide better rates for making fixed deadlines through the Internet or the cell phone.
According to the Central Bank, the efficiency level of the 10 largest entities is 58.7%: for every 100 pesos that banks earn, there is an operating expense of $ 58.70. On the other hand, between its digital counterparts it is between 25% and 35%.