Russian Firm Develops Secure Way to Connect Enterprise Data to Language Models
A new platform aims to speed up AI adoption by addressing data security and integration challenges.
Russian company Bercut, part of Rostelecom’s IT cluster, has introduced a product called AI Data Fabric that enables businesses to connect corporate data to large language models without compromising security.
Analysts estimate that in 55% of companies, projects involving large language models take more than six months to implement. The longer deployment takes, the lower the chances of recouping investments. Many firms hesitate to start due to concerns over data leaks and misinformation generated by AI systems.
Bercut’s experience suggests the core issue lies not in the models themselves but in fragmented data environments. Information is often scattered across databases, files and applications, making it unsuitable for direct use by language models. At the same time, companies must ensure strict protection of personal and sensitive data.
Overcoming the Barriers
AI Data Fabric provides tools to securely connect internal company data to AI systems. The platform supports databases, documents and enterprise services. If a company already uses Bercut’s integration bus, prebuilt connectors can link internal systems with external platforms, shortening the path from concept to real-world AI deployment.
AI Assistants for Business
The platform also allows companies to create assistants for different employees within a unified interface. These assistants can be embedded into daily workflows to search documentation, respond to customers or support engineers. Bercut’s HIP 2.0 platform, of which AI Data Fabric is a part, creates a unified layer across legacy systems, cloud services and partner APIs. It covers the entire data lifecycle and lays the groundwork for developing in-house AI models.
