Knowledge Graph

As of 2026-05-29

The SAP Knowledge Graph turns SAP business data from tables into a connected, machine-reasonable web of relationships — the grounding layer that makes AI/agents genuinely smart about a business, not just fluent. Tables describe (relationships are keys to join); a graph relates (entities = nodes, relationships = typed semantic edges, traversable). This is what lets an AI reason "which customers are hit by this delayed supplier?" without a hand-written join path. It's central to BDC + Joule grounding (M031/M046) and agentic reasoning (M049) — reasoning about the business, not generic text. The graph carries an ontology connected to the catalog's business glossary (M080) — the machine-readable form of the same governed definitions; the deep version of "data products with contracts" (M033). Consultant value: design the client's ontology correctly — a wrong graph grounds AI in a wrong business model. Honest caveat: only as correct as what's modelled; building the ontology is real design work, not auto-generated (§0i).