Oracle has introduced the Fusion Claw governed agentic execution runtime to help Fusion Cloud Applications complete complex, specialist-level work with greater autonomy while keeping enterprises in control.
Fusion Claw combines AI reasoning with deterministic enterprise computation, an approach that allows AI-powered applications to handle tasks such as research, modelling, simulation, financial analysis, and continuous replanning at scale.
Oracle has also launched 25 Claw-powered Fusion Agentic Applications to bring the broader portfolio to 75 applications designed to coordinate and execute end-to-end business processes.
“Fusion Claw helps answer that question by moving from AI assistance to execution, enabling customers to automate increasingly complex work across their most important business processes, with enterprise-grade governance,” said Mike Sicilia (top), CEO of Oracle.
Fusion Agentic Applications is a new class of outcome-based software that works towards defined business outcomes with people setting the objectives, authority and accountability.
Fusion Claw provides the runtime for more demanding work. A frontier AI model reasons, plans, learns, and adapts while deterministic enterprise computation carries out precise, high-volume execution.
The separation helps enterprises control costs by applying AI reasoning where judgement is needed and relying on more efficient computation for repeatable execution.
The system can support different levels of automation, from quick assistance to full-auto execution within explicitly delegated authority.
Built-in governance
To manage autonomous execution, Fusion Claw uses an Enterprise Operating Envelope that defines an organisation’s objectives, procedures, policies, permissions, risk thresholds, decision rights, approval requirements, and escalation boundaries.
An Outcome Trust Harness applies controls to each outcome run, covering identity, capabilities, data, and actions. Once work is complete, an Outcome Receipt creates an auditable record of the authority used, evidence considered, decisions made, transactions executed, and final result.
The model reflects a growing shift in enterprise AI as organisations are looking for systems that can do more than generate recommendations but still need clear boundaries around what AI can access and execute.
Applications targeting bottlenecks
The first 25 Claw-powered applications include tools for finance, human resources, supply chain, and sales teams.
- Ledger helps finance teams reconcile accounting information, investigate exceptions and anomalies, and perform complex financial analysis.
- Workforce Staffing creates staffing plans based on skills, accreditations, availability, schedules, labour rules, costs, and service requirements.
- Shipping Consolidation models shipping alternatives across timing, capacity, customer commitments, and cost before carrying an authorised plan into execution.
- Account Territory Growth Plan compares territory configurations and opportunity allocations to improve sales coverage while accounting for business constraints.
Running on Oracle Cloud Infrastructure, the applications are powered by frontier models including Google’s Gemini and OpenAI models. Oracle plans to support additional frontier models over time.
Fusion Claw expands Oracle’s wider Fusion AI strategy, which is anchored by Oracle AI Agent Studio for Fusion Applications. The platform provides no-code and pro-code tools for creating and managing agentic applications using Oracle, partner and external agents.
Built-in observability, return-on-investment measurement and safety controls are intended to help enterprises monitor performance and scale AI responsibly.
