The virtuous cycle: Unlocking AI outcomes through better data and integration (www.businessinsider.com)

SAP executive Manos Raptopoulos argues that enterprise AI’s usefulness hinges less on flashy models and more on tightly governed, integrated data — a shift from probabilistic generative outputs to business-deterministic results. He warns that even small inaccuracies (e.g., a 5–10% word-count error) translate into intolerable volatility when applied to finance metrics like EBITDA. Citing Databricks/MIT research that 72% of CIOs view their data-quality protocols as reactive, Raptopoulos positions the SAP Business Suite and SAP Business Data Cloud — which harmonizes SAP and non‑SAP sources into a centralized, semantically enriched layer — as the infrastructure to deliver reliable, auditable AI outcomes across regulated, global enterprises. Technically, SAP balances deterministic machine-learning systems with context-aware generative models (its Joule copilot) so NLP answers are routed by domain (finance vs. supply chain), improving relevance and traceability. Customer evidence includes Cirque du Soleil, which used the suite to cut finance cost objects per tour by nearly 80% and halve workload through unified data models and governance. The result, Raptopoulos says, is a “virtuous cycle”: high-quality, governed data attracts more usage, which generates more labeled operational insight to train better models and spawn practical AI apps — a blueprint for enterprises seeking predictable, auditable AI-driven outcomes.