Graph database company Neo4j has announced the Neo4j Context Engine, a product it says lets companies build a reusable “knowledge layer” that AI agents can draw on to reason consistently about business data. Neo4j says it is part of its graph intelligence platform and is meant to give agents “greater meaning, reach, and memory.”

The company’s argument is that data now has a third consumer. People understand how data relates intuitively, and applications did so through hard-coded logic. AI agents do not inherently know what “customer,” “account,” “supplier” or “risk” mean inside a particular organisation, Neo4j says, because that meaning is scattered across databases, applications, documents and individual AI projects.

Neo4j cites Gartner’s August 2026 Market Overview for AI Context Platforms, which defines such platforms as software that lets teams “construct, govern and deliver semantically organized knowledge to AI agents and other AI applications and services.” By treating context as an enterprise asset rather than a per-agent setting, the report says, these platforms separate the life cycle of organisational knowledge from that of individual agents.

Analyst Sanjeev Mohan of SanjMo, quoted in the release, said “context is not just more data, it’s really about relationships: not just that two things are connected, but why and how. That is exactly what graphs model natively.”

Neo4j’s own executive, Sudhir Hasbe, President of Technology and CPO, said modelling relationships and showing exactly why they exist is “absolutely essential to solving the problem of context for AI.” He called other approaches “a band-aid on the problem that breaks as AI agents grow in complexity and system access requirements,” adding: “Only Neo4j has decades of expertise using graphs and ontologies to solve complex business challenges.”

The announcement does not give pricing, availability dates, technical specifications, supported AI platforms or customer names, and the claims about performance and reliability are the company’s own.