AI in Russia Is to Collect and Analyze Information at “Deep” Scale
Alfa-Bank has expanded the capabilities of its generative AI platform AlfaGen by launching a new Deep Research feature.
With the new feature, the model acts as a digital analyst, helping businesses produce structured and reliable analytical reports.
At a client’s request, Deep Research can gather data from multiple sources, verify its reliability, and compile a final document with a clear structure and verified data. Unlike standard AI assistants, the system can process requests much faster — within 15–20 minutes. Its higher performance allows clients to analyze significantly larger datasets and draw more well-grounded conclusions.
Alfa-Bank says internal testing has shown clear advantages over foreign analogues. The system applies stricter data verification (a known weak spot for AI), uses a transparent methodology, and delivers analytics in a user-friendly report format.
A Digital Colleague
Svyatoslav Solovyov, Director of AI at Alfa-Bank’s IT division, says the launch of Deep Research changes the role of artificial intelligence within the company.
Logic, Analysis, and Security
The system is built on a modern IT stack. At its core is the GPT-OSS-120B language model, responsible for reasoning, and the LangGraph framework, which manages the multi-stage analysis process. Yandex Search API is used to find up-to-date Russian-language information, while Jina.ai technology ensures accurate data collection. Operational security is provided via the Kaspersky Private Security Network.
Getting to Work Immediately
The AlfaGen platform is already being used as the foundation for several of the bank’s AI solutions, including internal assistants and customer support tools.
At the initial stage, Deep Research will be available to the bank’s internal teams for analyzing markets, competitors, and trends. Later, the technology is expected to be offered to business clients that require fast, in-depth analytics for decision-making.
