According to DeepLearning.AI, SAP's Christoph Meyer and Lars Heling said enterprise AI agents often fail because they choose the wrong API and lack business process context (source: DeepLearning.AI). They emphasized that APIs execute in a discrete, ordered sequence over time, meaning agents must understand orchestration rather than isolated endpoints (source: DeepLearning.AI). They explained that knowledge graphs resolve this by defining semantics via ontologies, modeling resources, APIs, and business processes as connected nodes to guide correct execution (source: DeepLearning.AI). For traders tracking AI infrastructure, this positions ontology-driven knowledge graphs and API orchestration inside SAP-style stacks as key enablers of enterprise deployment readiness, and the session included no mention of cryptocurrencies or tokens (source: DeepLearning.AI).