Get best of the week

Two tapas to money: how we developed a financial chat bot with instant card issuance in Azerbaijan

The modern world is obsessed with speed. Imagine that you live in a world when an application for a plastic card or a loan is submitted in the usual messenger, it is almost instantly approved and you immediately receive the requested product. Imagine that such a “contact” in the chat can become a replacement for the mobile banking application. Not complete, of course, but taking over 60 percent of frequently used functions such as a balance sheet, transferring money to friends and paying in the store. This is exactly what is happening now in Azerbaijan, where all this has become possible thanks to our development.


Under the cut, we will tell you how the idea of ​​creating such a bot came about, how it was developed, developed and promoted. 

Background


In 2019, our team in Azerbaijan was set a non-trivial task: to come up with how to quickly increase the volume of business without any serious investments from the parent company.

In fact, it was a financial startup, where any decisions that could satisfy 4 main requirements were taken as ideas: 

  1. Highest Availability. The new product in its basic version should not burden the client either financially or energy-intensively. Ideally, no more than a few seconds should elapse between “want” and “use”. When we started, there was no talk of coronavirus, but now “contactless” communication is in demand more than ever. 
  2. . . , , .
  3. . MVP ( ) .
  4. . , .

After a long brainstorming session, the team realized that the chatbot meets all 4 criteria. Not to say that this was a completely new idea in the banking sector, but our chat bot was to become not just a useless “talker”, often annoying the client with his stupidity, but a full-fledged and clear interface of interaction with the bank. In fact, a sort of banking client, only packaged in the format of "contact". Nobody has done this yet.

As a rule, advanced chat bots are developed in Telegram. However, market research showed that only 16% of instant messenger users use Telegram in Azerbaijan. Most Azerbaijanis predictably sat on WhatsApp (more than 95% of users) and, which turned out to be very curious, on Facebook Messenger (about 75% of respondents). It would seem that the logic suggests that you need to bet on WhatsApp, however, in the case of Facebook Messenger, we got a unique link between the display of ads within the user’s feed, instant transition to the messenger and fast delivery of the product. As a result, MVP was decided to do on Facebook. 

In parallel with research, we studied legal regulation. It turned out that there are also no obstacles, since the state in Azerbaijan is actively moving towards digitalization: there is a regulatory permitted issue of unidentified payment cards, there is a developed state system that accumulates the financial behavior of citizens. The Central Bank is also developing a platform for remote identification of customers by banks. Thus, the legal, financial and technical capabilities of our bot gradually emerged.

Development


Technically, the solution was divided into 3 components: 

  1. Chatbot development. 
  2. Development of an “easy” solution for a loan conveyor. 
  3. Integration 1 and 2 with ABS VTB Azerbaijan and local processing. 

The Chatbots.Studio company located in Kiev was chosen to develop the chat bot. The loan conveyor was created with the help of VTB Group's partner - PST Lab (Minsk). A project manager was sent to Baku to coordinate the implementation team, and the project headquarters was located in Moscow. Thus, a distributed team was formed immediately in 4 countries. 

It took us some time to set up and motivate the administration and implementation team working on the spot. Because people who go to work in a bank usually get used to strict regulations and rarely changed rules. But we had a pure start-up with constantly changing tasks and regular testing and rejection of hypotheses. But we quickly rebuilt.

Chatbot programming is a simple task at first glance. But here we needed not just a "talk", but a banking service that complied with all the requirements of information security. In addition, it required the ability to flexibly configure scripts using templates and a special microservice architecture. After all, MVP is a living organism that constantly requires improvements. I must say that, despite the conditions that are completely atypical for working on Agile, the development from A to Z itself took only a little more than 2 months.

Functionality


How our bot works: most often a new client learns about us from advertising received directly in the messenger, or from a message from friends who already use our services. Upon transition to contact, the client can get a digital non-personalized DIRECT card.



Since the client has not yet been identified by the bank, he can deposit no more than 100 AZN on the card (this is about 60 USD), and the total volume of his operations cannot exceed 1,000 AZN (600 USD). These limits allow the client to evaluate the level of service, its advantages, and then confirm their identity with a visit to the bank or via video link. In this case, all limits are removed. 

An interesting point: it looks a little funny when the name of the dog or fictional character that users write on the digital map is displayed. But after the user passes the identification, in his details, of course, the real data is already displayed.

Immediately after the establishment of the card through the messenger, you can replenish the account from other cards (a compact WebView pops up), make transfers, pay in the store, on the Internet or through a QR code, pay for utilities. You can borrow money by generating your own QR code and sending it to friends. You can even withdraw cash through a one-time password via SMS. In other words, use it as a full-fledged card through a full-fledged client bank. Only almost instantly, without a visit to the bank and in the usual messenger.



You can also apply for a loan through the messenger. To get a solution, you need only two things: 1) the desire of the client; 2) his passport number. All data for analysis is pulled by API from public services.

Implementation


The project was expected to encounter difficulties due to its unusual nature for classical banking. Many local services with which integration was required, for example, to replenish cards from cards of other banks, had outdated data exchange technologies that were not built into the “instant” paradigm of the chat bot. To adapt to them, the development of additional solutions was required. But the main difficulties were of human nature.
 
For the success of a startup, it is important that not only the product, but also its team meet the principles of innovation. There is an opinion that restarting a business on a digital platform certainly requires the creation of a completely new, digital team. In our case, for a number of reasons, this was not possible, so there was a hard way to adapt the old bank team to a new reality. Here are just some examples of the difficulties that had to be overcome:

  • The need to abandon the usual axioms, starting from the desire to definitely see the VTB logo instead of a new digital brand and ending with the reluctance to abandon classical banking procedures even at low risks. By the way, they decided simply with the logo: we use the DIRECT brand, but everywhere we denote the parent brand - DIRECT powered by VTB.
  • , , . .
  • , , ( « »).
  • , .
  • : vs. .
  • , , .

In fairness, it should be noted that not only the team sometimes showed unpreparedness for an innovative product, but also the customers themselves. For example, the undeniable advantage of opening a real card in just a few seconds through the chat bot played a cruel joke. It turned out that many customers, having activated the card, abandon the product, simply because they do not even suspect that the card can already be used. I had to make an additional emphasis on this in the client way.

What's next?


While DIRECT is being tested at Azerbaijani customers, other banks of VTB Group are already beginning to take a closer look at the technology. Now the product is used by several thousand customers, and judging by the feedback, people like the speed and convenience of working in the messenger. Technically, our solution allows you to deploy another digital bank in the messenger in just 2 months.

Additionally, for greater coverage, we are also developing a chat bot on WhatsApp. Of course, this messenger is a leader in the post-Soviet space, while the Facebook messenger in some countries is not as widespread as in Azerbaijan. But already the results that we see on Facebook allow us to call the start of the project successful.

More articles:


All Articles