What is Knowledge graph?

A knowledge graph represents information as a network of entities and the typed relationships that link them: nodes for things (people, services, documents, decisions) and edges for how they relate (owns, depends-on, supersedes). This structure captures meaning that flat text or a pile of embeddings cannot, because the connections are first-class and queryable. Where semantic search answers what is similar to this, a knowledge graph answers what is connected to this and how, letting you traverse multi-hop questions like which deployments depend on a service this team owns. In AI agent systems, knowledge graphs are one way to structure persistent memory: facts become entities and relations rather than loose snippets, which makes recall more precise and reasoning over relationships possible. Some memory MCP servers, including Anthropic's reference memory server, model their store explicitly as a knowledge graph the model can read and extend through tool calls. Graphs and vector search are complementary, many systems combine a graph for structure and relationships with embeddings for fuzzy, meaning-based lookup over the same underlying facts.