VIMNORA™ / Personal AI
A working prototype of persistent personal context: one memory across models, a browser, and a voice endpoint in the room.
Engineering / Selected work
Context worth keeping.
Our product work and public engineering share one principle: useful context should live outside the model that uses it.
Active prototype
Mnemosyne / Public context engine
Local code and document retrieval, context compression, and model integrations. The public engineering foundation behind the personal-context work.
The product initiative
One persistent system.
More than one way in.
VIMNORA is Cast Net Technology's personal AI initiative. The current prototype combines memory, identity, permissions, source information, and connected tools. Models provide the answers; the accumulated record stays with the user.
The browser, camera, and device in the room make that principle tangible. We use the prototype daily, and are developing the product around what that use teaches us.
See the working prototypeThe public foundation
From software context
to personal context.
The public Mnemosyne repository is a local code and document retrieval engine. It indexes useful material, compresses context, and makes it available to models through its CLI, MCP server, and Ollama integration.
Mnemosyne is the engineering groundwork for VIMNORA, the personal AI prototype shown here. The shared principle is to preserve useful context outside the model, then supply what is relevant when it is needed.
Mnemosyne and VIMNORA are different scopes of work. The repository documents the retrieval engine and its integrations; it is not a packaged release of the complete assistant shown in the films.
Interested in integrations, model portability, or the product direction?
Build with us