At the Finopolis forum, MTS AI presented an AI assistant based on generative artificial intelligence for the fintech industry. The solution can automatically process large data arrays, allowing up to 50% savings in working time and accelerating customer service by up to 10 times.
The AI assistant from MTS AI analyzes internal databases, financial reports, and legal documents in 3 seconds, providing a brief, clearly formulated response to employee queries. Thus, a loan manager won’t need to recall small business loan terms or search for them in internal systems each time, and a deposit specialist will always have access to current rates and programs — the AI assistant will provide all necessary information via text query. The solution also allows users to receive brief document overviews and automatically compare different versions. Additionally, the AI assistant can identify trends and risks, create structured reports, and generate prompts for employees.
The solution is developed based on the Cotype large language model, created by MTS AI specifically for business, and the RAG (Retrieval-Augmented Generation) method. This method involves adding information from corporate sources to the language model’s existing knowledge base. As a result, the LLM can provide users with more complete and accurate answers compared to publicly available large language models. RAG is particularly useful for tasks requiring access to current or specific information, such as in customer support chatbots or question-answering systems.
“Finopolis has gathered Russian fintech leaders. That’s why we chose this platform to present our AI assistant. We’ve created not just a tool for automating routine processes, but a solution that has proven its effectiveness, accuracy, and security in handling banking information,” noted Dmitry Markov, Executive Director of MTS AI.
