SAP says AI governance required to prevent agentic sprawl
sap asserts that enterprise ai governance is essential to secure profit margins and protect against operational risk. manos raptopoulos warns that agent sprawl mirrors past shadow it crises but with higher stakes, noting that the distance between 90% and 100% accuracy is existential. he emphasizes that integrating vector databases with legacy systems increases compute costs and requires deterministic control.
SAP Global President Manos Raptopoulos argues that enterprise AI governance is mandatory to secure profit margins, noting that the gap between 90% and 100% accuracy is existential for business operations. He warns that deploying agentic AI systems without strict lifecycle management mirrors shadow IT crises but carries higher stakes due to autonomous access to sensitive data.
Raptopoulos highlights that grounding agentic models in proprietary data increases computational latency and token costs, making governance a hard engineering constraint rather than just compliance. He identifies three baseline issues for boards: establishing accountability for agent errors, creating audit trails for machine decisions, and defining thresholds for human intervention.