Oracle launches Fusion Claw to power autonomous enterprise AI agents

Oracle has announced the launch of Oracle Fusion Claw, a controlled execution environment designed specifically for Oracle Fusion Agentic Applications. This technology aims to bridge the gap between AI reasoning and deterministic enterprise computing, allowing for the large-scale completion of complex professional tasks such as deep research, modeling, simulation, and continuous replanning.

The system introduces 25 new Claw-driven Agentic Applications, expanding Oracle’s existing portfolio to 75 applications capable of coordinating end-to-end business processes. These applications, which include tools for ledger management, workforce staffing, and shipping consolidation, run on Oracle Cloud Infrastructure and utilize advanced AI models including Gemini and OpenAI.

To address security concerns regarding autonomous AI, Fusion Claw operates within an ‘Enterprise Operating Envelope.’ This framework manages autonomous execution through predefined goals, standard operating procedures, policies, and risk thresholds. An ‘Outcome Trust Harness’ applies controls to identity, data, and actions, while an ‘Outcome Receipt’ provides an auditable record of all decisions and transactions made during a task. By utilizing an isolated runtime container environment, Oracle aims to prevent AI agents from operating outside of authorized business contexts, effectively mitigating the risk of uncontrolled autonomous behavior.