Vektor memory launch – Persistent graph memory for self-hosted AI agents

Summary: Vektor is a self-hosted graph memory layer built on Neo4j that provides AI agents with persistent, relational memory across sessions. It stores entities and relationships as a knowledge graph, enabling agents to reason about history rather than retrieving isolated data chunks.

What it does

Vektor maintains persistent memory for stateless AI agents by storing context as a graph database. It supports any LLM via API or MCP and preserves data on the user’s server.

Who it's for

Developers and teams needing persistent, relational memory for AI agents in a self-hosted environment.

Why it matters

It solves the problem of AI agents losing context between sessions by providing durable, structured memory that enables reasoning over past interactions.