Neo Cortex empowers AI agents with a robust long-term memory system, eliminating the challenge of amnesia across sessions. It automatically captures conversations, distills knowledge into atomic memories using advanced LLMs and embeddings, and stores them locally for efficient retrieval. This ensures that every interaction with your AI agent builds upon past knowledge, allowing it to remember critical decisions, project context, and preferences, effectively starting each new session where the last one left off. It features an intelligent pipeline for processing, storing, and retrieving information, complete with a web UI for visualization.
Key Features
01Consolidates and prunes similar memories through periodic 'dream cycles'
020 GitHub stars
03Interactive web dashboard with a 3D concept graph for visualizing knowledge connections
04Dynamic memory energy model where memories gain energy when recalled and decay when forgotten
05Automated conversation capture and knowledge distillation using LLMs
06Dual retrieval system with semantic search and full-text search
Use Cases
01Providing AI agents with continuous context to avoid repeatedly re-explaining project architectures or preferences
02Allowing AI agents to track, consolidate, and evolve project knowledge over extended development periods
03Enabling AI agents to recall specific project decisions and technical specifications made in previous sessions
