Relate is an open-source, embedded graph database designed to give AI agents structured, connected memory. Rather than keeping information as isolated text, agents can represent people, decisions, tasks, documents, and other entities as nodes linked by relationships, then use Cypher queries to trace how facts connect. The site demonstrates an agent retrieving details about a launch by querying a small graph of remembered facts. Relate runs within an application, so there is no database server to deploy or account to create. It is available through a command-line tool, agent skills, an MCP endpoint, and libraries for languages including Rust, Python, JavaScript, Go, Java, Swift, and C.
Users can store data in memory, a directory, or a single database file, create it with Cypher, import CSV data, or mirror data from Neo4j Aura. Query results can be returned as tables or JSON, and a browser playground illustrates how successive statements build a graph. Relate supports openCypher and is tested against the openCypher TCK, with documented differences from Neo4j. The site also provides an installation workflow that verifies downloaded releases, setup guidance for coding agents, documentation, and community recipes. Its central pitch is that agents can combine graph relationships with similarity searches in one query, giving them a practical way to retrieve not just stored facts but the connections among them.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.