Neo4j, the graph intelligence platform, has introduced Relate, an AI-native, open-source, in-memory and in-process embedded database built in Rust for AI agents and agentic systems. The new database enables AI coding agents to create a database and begin building applications in milliseconds.
Designed specifically for autonomous AI workloads, Relate provides high-speed data access, graph reasoning, and native agent connectivity. It can run across different environments, including software containers, web browsers, Jupyter notebooks, and edge devices.
The launch addresses the growing demand for database infrastructure designed to support AI agents that independently plan, execute, and coordinate complex tasks.
Neo4j has released Relate, an open-source embedded database built in Rust specifically for AI agents and agentic systems.
Relate enables AI coding agents to create databases and start building applications in milliseconds.
The database supports multihop reasoning and GraphRAG to improve AI agent accuracy and reduce hallucinations.
Built-in connectivity includes a Model Context Protocol (MCP) Server, CLI, Bolt, and HTTP Query API v2.
Relate delivers sub-millisecond in-memory reads and combines vector similarity search with graph traversal in a single Cypher statement.
The database is available under the Apache 2.0 license at relatedb.ai.
As AI agents become more capable of executing complex workflows independently, database infrastructure must support higher interaction volumes and faster data access.
Gartner projects in Forecast Analysis: Agentic AI in DBMS, Worldwide, 2026 that machine users will surpass human and AI assistant usage by 2029, accounting for 75% of DBMS workloads, compared with 10% in 2025.
The forecast highlights how autonomous AI systems are changing database workloads through increased interactions, concurrent user activity, and growing demand for computing, storage, and streaming resources.
Neo4j developed Relate to address these requirements with an embedded, in-memory database designed specifically for agent-driven applications.
"Relate is the ultimate agent-native database,” said Sudhir Hasbe, president of technology and CPO at Neo4j. “Built in Rust for maximum speed and efficiency, it acts as a lightweight, in-memory store with highly durable disk storage, when needed, that can run anywhere. From seamless agent connectivity via MCP to multi-hop reasoning through full Cypher support, Relate provides a unique out-of-the-box solution designed to run in-process for agent sandboxes and agent memory as well as on mobile or edge devices.”
Relate is released as open source under the Apache 2.0 license. It brings multihop reasoning and GraphRAG capabilities to AI agents, helping them retrieve relevant information and produce more accurate results with fewer hallucinations.
The database is designed around the execution loops of autonomous agents and includes several built-in capabilities:
Automatic agent detection: Relate can identify agent harnesses and adapt its responses accordingly.
Structured error handling: It returns structured errors with explicit retryable flags, allowing agents to handle exceptions and self-correct without custom parsing logic.
Controlled data retrieval: A row cap and in-band truncation marker help prevent large retrievals from overwhelming an agent's context window.
Predictable payloads: Bounded retrieval results help agents work with more consistent data volumes.
These capabilities are designed to simplify database interactions for autonomous systems and reduce the additional logic developers need to build around agent workflows.
Relate is designed to operate wherever AI agents are deployed. It supports Jupyter notebooks, web browsers, software containers, and edge devices.
The database includes built-in connectivity through a Model Context Protocol (MCP) Server and a dedicated command-line interface (CLI) for direct execution.
It also exposes Bolt and HTTP Query API v2 endpoints. These provide compatibility with existing graph drivers, orchestration frameworks, and standard developer tools.
By supporting these interfaces, Relate aims to simplify integration into modern AI development stacks without requiring developers to build custom connectivity solutions.
Relate is engineered to handle the frequent read and write operations associated with autonomous AI agents. It delivers sub-millisecond in-memory reads to support high-frequency agent workloads.
One of its capabilities allows developers to combine vector similarity searches and graph traversals within a single Cypher statement. This enables AI applications to retrieve semantically relevant information while exploring relationships between connected data.
The database also supports transactional integrity through an in-memory engine and an optional fsync'd write-ahead log for durable storage. Developers can flush database state to disk when long-term persistence is required.
Relate is 100% openCypher and more than 97% GQL compliant out of the box, providing developers with graph query capabilities for building agent-native applications.
Neo4j Relate is available as an open-source database under the Apache 2.0 license.
Developers can install the database and run queries in milliseconds through the official website:
About Neo4j
Neo4j is the graph intelligence platform that transforms data into knowledge to power the next generation of intelligent applications and AI systems. It includes enterprise-ready knowledge graphs for accurate, explainable, and governed AI; the most comprehensive, trusted, and easy-to-deploy graph database and capabilities across any environment and data source; and an unmatched ecosystem trusted by 84 of the Fortune 100 and supported by the world’s largest graph community.
Learn more at neo4j.com.
