Developer Mikhail Singatullin has submitted his MMW (Managed Memory Workspace) platform to the Russian software registry. The system acts as a middleware layer providing long-term memory for AI agents, preventing data loss between sessions in closed corporate environments.
The software utilizes the Model Context Protocol (MCP) and graph-based memory to ensure data remains within a company's infrastructure, adhering to Russian personal data laws. It is designed for on-premise deployment or use within secure Russian cloud environments like Cloud.ru.
Singatullin invested nearly 3 ₽ million of his own funds into the project's research and development. The platform is currently in the pre-seed stage, and the developer is seeking external funding to build a team and launch pilot projects for corporate clients.
The solution is compatible with both Russian models, such as GigaChat, and international agent frameworks. It claims to reduce LLM token consumption by up to 70% by filtering out obsolete information and providing microsecond context retrieval.
While the project currently lacks commercial contracts, it has gained recognition in industry competitions and maintains a separate international version, mmwhub.tech, which has received high ratings in global MCP server catalogs.
