Vektor memory launch

VEKTOR provides persistent graph memory for agents, ensuring they retain context and knowledge across sessions. Self-hosted and built on Neo4j, it offers a robust memory layer that prevents agents from starting from scratch every time. Key features include: * Persistent memory across agent sessions * Storage of entities, relationships, and context in a true graph structure * Four distinct graph types: Episodic, Semantic, Procedural, and Temporal * Easy integration into existing Node.js agent setups * One-time payment model with no recurring subscriptions Unlike systems that rely on flat vectors, VEKTOR stores information as a relational graph, enabling agents to reason about their history rather than just retrieving disconnected chunks. This structured approach means agents can build upon past interactions, understanding sequential events, factual knowledge, learned procedures, and the chronological significance of memories. The architecture efficiently processes raw input, extracts structured scene representations, and persists them into the MAGMA graph. VEKTOR is designed for developers and organizations building sophisticated agents that require continuous learning and contextual awareness. It's ideal for maintaining long-term user preferences, project progress, and evolving knowledge bases for tasks ranging from customer support to complex data analysis. This system ensures agents develop a consistent and reliable understanding over time, significantly enhancing their utility and performance.