Mycelium

Mycelium is a novel semantic filtering engine that processes embedding vectors and associated text through a unique biological ecosystem simulation. It models knowledge units as nodes that compete, cooperate, merge, and die over a simulated 60 'ticks.' This dynamic process classifies knowledge into categories such as 'pure' (unique and resilient), 'merged' (cluster centers), 'loner' (isolated), 'redundant' (duplicates), or 'dead' (outcompeted), providing a sophisticated assessment of information quality and relationships. The engine operates entirely in-memory, integrates seamlessly with external vector databases like Qdrant, and offers diverse output formats for various applications, including AI agents and content management.

Key Features

Use Cases