In 2024, VTB chat bot on the platform from Nanosemantics processed more than 50 million requests
Customers: VTB Bank Moscow; Financial Services, Investments and Auditing Contractors: Nanosemantics Lab Product: Nanosemantic: DialogOS (Dialog Operating System)Project date: 2021/02 - 2024/12
|
2024: Processing more than 50 million requests
In 2024 chat boat VTB , more Nanosemantics than 50 million requests were processed on the platform from "." This was announced by "Nanosemantics" on February 10, 2025.
February 2025 marks the 4th anniversary of the project to introduce the DialogOS platform from Nanosemantics into the perimeter of VTB Bank. According to VTB, at the end of 2024, the load on the contact center increased to 96 million calls, the share of calls to the chat increased in a year and a half from 44.3% to 60.9%, and VTB's chat bot took a high position among such solutions for client service.
The DialogOS platform is designed to develop and support conversational systems of any complexity based on. AI It works in 40 languages and includes a huge knowledge base, which as of February 2025 has 3611 dialog scenarios, 5230 specialized dictionaries and more than 3 million adaptive questions. One of the main advantages of DialogOS is the combination neural network algorithms of manual rules.
The platform is used by several divisions of VTB Bank, creating on its basis text and voice chat bots that meet customer needs and specific tasks. On the bank's side, the platform is supported by a separate team that formulates new tasks and challenges for its developers.
VTB's chat bot for individuals advises on more than 1,800 topics, and among the most popular requests are details, complaints about fraudsters, connecting credit holidays, ordering certificates and changing the pin code on the card. In May 2024, VTB's chatbot began processing all customer requests using artificial intelligence, which significantly improved information recognition. At the same time, in 2024, the load on call centers increased to 96 million calls compared to 78 million a year earlier, and it is automation of service that helps to cope with it. VTB chat bot works in six digital channels, including the VTB Online application for Android, a web version on the VTB website, instant messengers and the VKontakte social network.
Since April 2021, the chatbot has also been installed on the website of the VTB Business Platform, a digital remote customer service system for small and medium-sized businesses. The chatbot is available in the authorized and unauthorized zone, as well as in the TG channel of the Internet bank SMB @ vtb_sme_bot. He advises on the entire functionality of the bank: sending ruble and currency payments, salary transfers, loans, deposits, guarantees, bank support, opening accounts, etc. Per month, the chatbot conducts, on average, 72,000 dialogues. Since 2023, it has reduced the burden on operators: today 75% of calls are processed by a bot, the rest falls on the operator.
VTB's chatbot is developing in leaps and bounds, automating consultations and expanding the list of topics you can ask about and get a qualified quick answer. Previously, we have already noted that over the past year and a half, the share of calls to the chat has increased from 44.3% to 60.9%, and thus in 2024 exceeded the share of calls to the contact center. This very accurately speaks of the demand for chat bots, their convenience both for banks and for customers, - said Alexander Raikhin, leader of the stream "Messengers and Chat Bots." |
Cooperation with VTB Bank is a great honor and at the same time a great challenge for our team. The scale of the project allows you to test and implement updates in DialogOS that are in demand among business and end users. So, for example, for 2023-2024, along with the constant development of dialogue scenarios, markup data for training neuronets and the expansion of the integration line, one of the most notable updates was the Guardsman service, which at the preprocessing level corrects grammatical errors, incorrect declinations, syntactic errors in user requests, transmitting a corrected version of the request to the chatbot, ‒ said Ilya Ivanov, commercial director of Nanosemantics. |