AI/ML
Graph database supplier Neo4j’s Relate is an in-memory database built in Rust for AI Agents and Agentic Systems.
Relate brings multihop reasoning capabilities and GraphRAG to AI agents for more accurate results and fewer hallucinations. It’s engineered specifically for the execution loops of autonomous agents, natively auto-detecting agent harnesses and returning structured errors equipped with explicit retryable flags. These flags enable agents to handle exceptions and self-correct without custom parsing logic. Relate enforces a row cap combined with an in-band truncation marker to prevent context window saturation during large retrievals. This constraint ensures agents receive bounded and predictable payloads.
Sudhir Hasbe, president of technology and CPO at Neo4j, said: "Relate is the ultimate agent-native database. 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.”
Cypher refers to the openCypher query language specification and implements the large majority of the newer ISO GQL (Graph Query Language) standard. Neo4j has been evolving Cypher toward GQL
Neo4j cites Gartner projections in Forecast Analysis: Agentic AI in DBMS, Worldwide, 2026 (June 2026), saying that: “machine users will surpass human and AI assistant usage by 2029, reaching 75 percent of DBMS workloads, up from 10 percent in 2025.” It claims early enterprise adoption focused on simple conversational AI assistants for information retrieval and synthesis, but now budgets and workloads are shifting toward autonomous systems that can independently plan, execute and coordinate multistep operations.
It says the impact of AI agents/machine users will be far reaching, creating a new class of resource consumers -AI agents - that are becoming the primary users of databases and generating 10 to 100 times more interactions than humans and skyrocketing concurrent user loads
Relate can run anywhere and is built from the ground up as a lightning-fast, in-memory database engineered specifically for the agent paradigm. Neo4j says that, with Relate, AI coding agents can create a database and start building applications in milliseconds.
It delivers sub-millisecond reads in memory to handle the high-frequency read/write cycles of agentic workloads. A single Cypher statement can search by vector similarity and execute graph traversals simultaneously. The database maintains strict transactional integrity using an in-memory engine with an optional fsync'd write-ahead log for durable stores. It provides native options to flush state to disk for long-term persistence. Relate is 100 percent openCypher and more than 97 percent GQL compliant out of the box.
Relate deploys seamlessly across Jupyter notebooks, web browsers, software containers, and edge devices, and runs natively wherever agents are at work. The database ships with built-in agent connectivity via a Model Context Protocol (MCP) Server and a dedicated CLI for direct execution. Relate also exposes Bolt and HTTP Query API v2 endpoints to provide drop-in support for existing graph drivers, orchestration frameworks, and standard developer tooling.
Relate is open source software under the Apache 2.0 license, and is available now from here.
Bootnote
There are other in-memory databases built for AI agents:
- ArcadeDB (embedded mode) is a multi-model database (property graph + document + key-value + more) that can run fully embedded / in-process.Memgraph is shipping a high-performance open-source in-memory graph database written in C/C++ and designed for GraphRAG, AI memory, agentic AI, and real-time graph analytics.
- FalkorDB is an in-memory graph database (sparse-matrix / GraphBLAS-based engine) focused on low-latency multi-hop queries for AI agents, knowledge graphs, and GraphRAG.
- Redis/Valkey (including Amazon ElastiCache for Valkey) is an ultra-low-latency in-memory key-value store that’s widely used as the short-term/working-memory layer for agents (session state, conversation buffers, and semantic caching.
Suggested positioning:
- Relate - lightweight embedded graph store that agents can start themselves.
- Memgraph - high-performance real-time graph analytics and AI context
- FalkorDB - low-latency GraphRAG/multi-tenant agent memory.
- Redis/Valkey - simple ultra-fast short-term state or vector cache.