Close

Contact Us

Need service or support? Start your request online and we’ll find you a solution.

Your request was succesfully accepted!

How Machine Learning, Neural Networks, and Chat Bots Make Banking Easier

It won’t be an exaggeration to say that nowadays AI in banking is ubiquitous. AI algorithms are used to perform a number of functions, such as credit scoring based on a client’s Facebook profile, debt collection, financial advising to investors, fighting fraud, and doing routine tasks. In this article, we’ll talk about how AI saves billions of rubles for major banks, such as Sberbank, VTB, and Tinkoff Bank, and other financial organizations.

 

How much money banks are saving with AI

According to the forecast by the research agency Autonomous Next, by the year 2030 banks worldwide will be able to cut their expenses by 22% thanks to AI technologies. The overall economy can reach a whopping 1 trillion dollars.

The leading Russian banks are actively using AI to reduce their expenses. For example, in 2017, Sberbank made an additional 2-3 bln dollars through using AI and analyzing data for risks and sales management. (The bank’s net revenue equaled to ca. 11.6 billion dollars in 2017.)

Below we’ve outlined the ways in which AI makes banking easier and more profitable.

The seven banking tasks AI can successfully perform

 

  1. Credit scoring

Credit scoring is probably the most promising field for implementing AI algorithms. Most of the banks that took part in the 2018 survey (including Tinkoff, Gazprombank, MTS-bank, Moscow Credit Bank, Russian Standard, etc.) by the rating agency “Expert RA”, said they were using the AI capacities to solve that task.

At Sberbank, AI is now responsible for 98% of lending decisions (but only when it comes to loans for individuals). Credit risks are assessed based on the user’s “digital footprint.” According to Herman Gref, Head of Sberbank, “the digital footprint has already reached 500 Mb per day. We can call it the person’s “digital personality” which is very close to their “human personality.”
When it comes to companies, AI is less effective, only accounting for 30% of lending decisions.

  1. Debt collection

Debt collection is the second-popular area of AI application. In 2016, Sberbank was the first to launch a pilot project by its subsidiary, Aktiv Bank. After a year, the robot’s performance proved to be almost 24% higher than that of human employees. Debtors paid back their debts within 2 weeks after receiving the robot’s call.

After the successful start, Aktiv Bank collaborated with another 27 banks (Otkritie, Binbank, etc.). As of 2017, AI developments accounted for 25% of the company’s total revenue. In the autumn of 2018, after a 3-month trial period, VTB adopted a debt collector robot developed by Aktiv Bank.

“Currently, the robot is only effective for insignificant loan repayment delays. The average call duration is 1-1.5 minutes, which is the same as a call made by a human operator. While an employee can make about 200 calls a day, for the robot this number is basically unlimited,” said Anatoliy Pechatnikov, Deputy Head of the Management Board at VTB.

 

  1. Fighting fraud

In 2015, Pochta Bank was among the first financial organizations to adopt biometric technologies in its outlets. Today, face recognition system is used at more than 4,000 bank outlets and 50,000 stores, which are the bank’s POS partners. To access CRM system and other business apps, bank employees are using two-factor identification (login/password + identification by photo).

In 2016-2017, the new technologies allowed Pochta Bank to save the total of 3 bln rubles. In 2016, the bank received 9,200 fraud loan requests. The requested loan amount was 1.5 bln rubles. In 2017, the number of fraud requests was at 10,000. The bank used the AI technologies to identify the individuals who applied the requests.

  1. Routine tasks

In 2018, Alfa-Bank set the goal to replace human employees with robots in 30 routine business processes. The automation of the first seven processes saved the bank 20 million rubles a year. As a result, Alfa-Bank is now planning to cut its annual expenses by up to 85 million rubles.

The operations delegated to robots included: processing payments by companies and individuals, processing unidentified payments, processing incoming emails, changing the client’s data at the client’s request, revising crediting agreements for individuals by request, and responding to common inquiries.

To manage robotic apps, Alfa-Bank used the Blueprism platform, paying less than 1 million rubles for a 3-year license.  Each robot was given its own workplace, with installed Blueprism agent and other software. As the next step, AI systems were taught by the employees with a good knowledge of the bank’s business processes and robot teaching methods. Before the project launch, Alfa-Bank planned to expand its operational staff by 3.3.%. Obviously, after the adoption of AI robots, the bank changed its plans.

  1. Investment advising

Robo advising is another technology major Russian banks are benefiting from. In July of 2018, Tinkoff Bank launched a robo advisor on its brokerage platform Tinkoff Investments.

According to the bank’s press release, “A robo advisor only needs a few hours to create an investment portfolio based on the available investment amounts and with an optimal risk-reward ratio.”

During the first month, the app welcomed by 42,000 users and generated 142,000 investment portfolios. The average investment amount was 60,0000 Russian rubles and 1,678 US dollars. For the most part, investors were buying ruble securities.

Earlier, in 2016, similar projects were launched by Sberbank together with FinEx, AK Bars, and VTB24. (VTB24 merged with VTB in 2018.) In the same year, Conomy company launched its robo advisor “Right.”

  1. Finding new locations for banking outlets

In 2018, Rosbank started to apply AI capacities for growing its retail network. According to Arno Deni, Deputy Head of Rosbank Management, the bank used the technology developed by Marketing Logic, a company specializing in geomarketing.

Based on machine learning, the system assesses the potential of a selected location using 250 factors. The factors are divided into three groups. The first group includes geo characteristics (distance to the city center, distance to the metro station, price for 1 sq. meter, etc.). The second group covers traffic issues, such as number of available bus/tram routes going to the location. Finally, the third group of factors focuses on the objects in the vicinity of the location (shopping malls, business centers, banks, houses, etc.).

Currently, Rosbank has 350 outlets all over the country. In the next couple of years, the bank plans to “significantly enhance” the performance of its branches.

  1. Employee consulting

Chat bots are successfully used to answer standard inquiries made by employees and clients 24/7. According to the 2017 survey carried out by R-Style Softlab, one in five banks in Russia and other CIS states was willing to use chat bots. Also, most banking organizations were planning to adopt chat bots in 2018.

As a successful case, we can mention a chat bot developed by Alpha-bank for its payroll program employees. Before that, the bank’s support team got daily more than 100 calls from bank employees asking about the conditions for opening card accounts. Most of the time, callers asked the same questions. After those calls had been delegated to the chat bot, the performance of the support team increased by 50 times.

Along with chat bots, banks can also benefit from voice assistants. However, here we’re talking about a more sophisticated technology. Currently, the only Russian company using this technology is Yandex. In December of 2018, Oleg Tinkov, Head of Tinkoff Bank, announced that the bank was planning to create its own voice assistant.

“We’ve decided to call it “Oleg.” But who knows? We may change our minds and call it “Ivan” or something,” said Tinkov.

According to Tinkov, the virtual assistant will be able to perform both financial and day-to-day tasks, e.g. make a money transfer, book a table at a restaurant, etc. Despite being named after the head of Tinkoff Bank, the assistant won’t have Tinkov’s voice. As for other banks, currently they are not planning to adopt voice assistants.