AI inside a bank's closed contour: how ArtCloud experts brought ZuBrIQ experience to MTBank

ArtCloud has delivered an AI service for MTBank with functionality similar to ZuBrIQ, the company's own platform running inside Belarus's closed contour. Behind the project are the same engineers who invented and built ZuBrIQ: Alexander Zubik and Evgeny Bratkovsky.

ZuBrIQ began as a product platform at ArtCloud: chat with large language models, document workflows, assistants, and team tooling — all within the country's perimeter, billed in Belarusian rubles, with an [OI] compatible API. That same principle — full-featured AI in a closed contour — became the foundation of the bank's solution. Part of the groundwork created for ZuBrIQ was adapted and implemented by Alexander Zubik in the banking project.

What was built for the bank

Banking is one of the most conservative industries when it comes to data: regulatory requirements, internal security policies, strict resilience standards. That is why the MTBank service follows the same architectural logic as ZuBrIQ: model and document interactions stay inside the bank's infrastructure, with no data leaving for external services.

Inside are elements familiar from ZuBrIQ: chat with language models, assistants for internal documents and regulations, and orchestration tooling. What differs is the requirements: the banking contour imposes its own audit and resilience standards, and the solution was assembled with them in mind.

"The idea behind ZuBrIQ was always that corporate AI does not have to live in someone else's cloud. We built a platform you can hand a working process to: models, documents, assistants — all inside a contour the client controls. In the MTBank project we applied the same approach: take proven groundwork and assemble a solution that meets banking requirements for security and resilience," said Alexander Zubik, co-author and developer of ZuBrIQ.

"When we built ZuBrIQ, we never saw it as a product for a single industry. It is an engineering framework: chat, models, document workflows, AI-agent orchestration. A bank was a good test of whether this framework could withstand fintech requirements. As the project shows, it can," added Evgeny Bratkovsky, co-author and developer of ZuBrIQ.

Why it matters

The project confirms ArtCloud's strategy: its own products and corporate deployments grow from a single engineering base. ZuBrIQ continues as a product line — chat with models, ZuBrIQ Academy with its 3D mentor, ZuBrIQ Bot Creator for assembling digital consultants — while the groundwork it accumulates is carried into closed corporate contours: from banks to payment providers.

For banks, this means a practical alternative to choosing "either foreign cloud AI or nothing": full functionality inside one's own contour, with local delivery and predictable economics in BYN.

ArtCloud will continue to develop ZuBrIQ as a platform and share its engineering base with corporate customers in Belarus.