Oracle has introduced Fusion Claw, a runtime designed to govern AI agents that automate business processes, running on Oracle Cloud Infrastructure and integrated with Fusion applications. It targets ERP, finance, and supply chain management, and uses frontier models from Google and OpenAI, with support for others planned. The company says it moves from AI assistance to execution with enterprise-grade governance.
Analysts welcomed the move. Gartner said Fusion Claw provisions isolated containers for heavy supply chain optimizations and subledger reconciliations within clear policy limits, and that Oracle is rebuilding its application foundation around native agentic workflows rather than bolting on chatbots. A key design choice separates reasoning from compute execution: an agent defines logic once, while deterministic engines handle calculations, avoiding the cost of passing millions of records through expensive LLM context windows.
However, Gartner warned that Fusion Claw is still in early piloting mode, and Forrester's Craig Le Clair said trust issues remain before agentic systems can be left to create tools and process patterns independently. Forrester also noted that Fusion Claw is proprietary and tightly coupled to Fusion and OCI, which may be a problem for enterprises with hybrid environments. Domain clouds like Oracle suit generic, non-differentiating functions, while more complex custom agents may need open platforms or hyperscaler frameworks. For organizations already deep in Oracle, the tighter governance and lower running costs may be worth it; for others, the lock-in is harder to justify.