Context Graph
A context graph extends a knowledge graph with business rules, terminology definitions, and access metadata — such as what 'churn' or 'fiscal year' means for a given team — so AI agents can interpret enterprise data correctly, not just know where it lives.
A knowledge graph captures entities and the relationships between them — customer, account, opportunity — but says nothing about how those entities should be interpreted. A context graph builds on top of that structure by attaching business rules and synonyms: that ‘customer’ and ‘account’ mean the same thing in one system, that fiscal year starts in February for one business unit and July for another, or that ‘cohort’ has a different definition depending on which team is asking.
Without this layer, agents given access to raw data still make wrong decisions because they lack the tacit knowledge humans carry around in their heads or buried in internal docs. Feeding that context to an agent alongside the data itself is what lets it answer questions like ‘why is churn up’ consistently, instead of producing a different answer per team the way human analysts historically have.
