Oracle is pitching AI agents that can carry out guarded business processes, not just help people with individual tasks. At Oracle AI World, it introduced Fusion Claw for agent execution and announced an AI tool intended to help configure Fusion applications.
Oracle Watch analysis
What happened
Fusion Claw is described as a governed runtime for complex work within Oracle’s Fusion applications. Its six functions cover analysing situations, reconciling discrepancies, simulating options, optimising outcomes, predicting what may happen next and executing actions under controls. Oracle also reported 5.7 million agent runs, more than 100,000 certified builders, 7,000-plus customers using AI or agents, and 3,648 production agents. The figures were shared at the event, not independently assessed.
Oracle also announced Fusion Setup & Configuration Registry, or FSCR. The tool uses AI and Oracle Fusion product information to suggest implementation workflows and software configurations. The stated aims include reducing implementation costs and helping customers realise benefits sooner. Those are intended outcomes, not evidence that FSCR has already delivered savings.
The details come from Diginomica’s account of Oracle AI World. The article discloses that Oracle is a premier partner of the publication.
Why it matters
This is a bet on embedding agents inside business software, where they can work with the processes and data customers already use. If the controls and workflows hold up, that could mean more than another assistant drafting a summary: software could help move work through a process and take approved actions.
FSCR points to a second consequence. If AI can make setting up business applications quicker or easier, it could change the work customers and their implementation partners do. For now, Oracle has described the capabilities and their intended benefits; customers will need evidence on reliability, effort and results.
Our read
The interesting shift is from AI as a feature people consult to AI as a system that can act within business applications. That raises the stakes for good controls, but also makes the product proposition more concrete than a new chatbot with a name badge.
Oracle’s customer and usage figures suggest activity at scale, though they do not tell us how often agents are used successfully or what outcomes they produce. The useful test is whether these tools complete valuable work safely, and whether customers can show the gains.
What to watch
- When Fusion Claw and FSCR become available to customers, and what they can do in practice.
- What controls, human approvals and audit trails govern agents’ actions.
- Whether customers report measurable implementation savings or improved business outcomes.
Discussion spark: Would you let an AI agent execute work inside core business software if its actions were governed and auditable, or should it remain an adviser that people must operate themselves?
Sources and evidence
- Want to hear more about Agentic AI? (16 November 2026, 08:35 UTC)
Independent WittyWires tracker for public updates about Oracle. Not affiliated with or endorsed by Oracle; this is not an official account.