AI in Enterprise ERP: What Actually Works (And What Doesn't)

The conversation around AI and enterprise resource planning has reached fever pitch. Vendors are racing to bolt "AI-powered" onto every feature, analysts are publishing breathless reports about transformation potential, and CIOs are under pressure to show AI initiatives before the next board meeting.

Having spent 25 years implementing, configuring, and rescuing Oracle ERP systems, I want to offer a more grounded view. Not pessimistic — I genuinely believe AI will reshape enterprise operations. But the gap between what's being promised and what's delivering results today is significant, and organizations deserve an honest map of the terrain.

Where AI is Delivering Real Value Today

• Invoice and document processing — Touchless AP automation is mature and ROI is measurable within months.
• Demand forecasting — ML-enhanced forecasting reduces safety stock and stockouts in supply chains with good historical data.
• Knowledge retrieval — AI agents that surface the right ERP documentation, SOPs, or historical resolution data are quietly saving hours per week.
• Anomaly detection in financial data — Catching outliers in GL entries, purchase orders, and inventory adjustments before close.

Where the Hype Outpaces Reality

The areas where I see organizations burning budget without results tend to share a common thread: they are trying to apply AI to processes that are fundamentally broken at the data or process level. AI does not fix bad master data. It amplifies it.