Privacy-first AI memory system with MCP server for managing user-authored memories, enabling AI agents to read/write memories with fine-grained vendor access control.
A privacy-first memory layer that pseudonymizes sensitive data locally before sharing a 'Working-Fiction' version with external AI agents. It enables secure agentic workflows by ensuring personally identifiable information never leaves the user's sovereign hardware.
Gives AI agents a private, on-device persistent memory using semantic search over stored facts, preferences, and decisions. It enables agents to remember information, recall it by meaning even without keyword overlap, delete memories, and list them by tag.