Oracle rolls out 25 autonomous AI agents across its business cloud suite

Software provider Oracle has introduced 25 automated business applications designed to execute complex enterprise workflows autonomously rather than merely answering questions or generating text. Announced at its AI World conference in Las Vegas, the tools span the company's finance, human resources, supply chain and customer experience platforms.

The applications run on a new execution runtime called Fusion Claw. Oracle designed the runtime to let software agents carry out multi-step work, including continuous replanning, operational modelling and calculations, while adhering to existing business permissions and approval rules. Alongside the suite, Oracle launched Fusion Companion, which allows staff to trigger and run enterprise tasks through general tools such as ChatGPT while maintaining corporate data boundaries.

Moving from generation to task execution

Traditional enterprise AI tools have largely focused on summarising records, answering support queries or drafting messages. Oracle's new setup takes a different direction by handing tasks directly to automated agents that interact with accounting ledgers, inventory counts and staffing rosters.

Within finance, an application called Ledger reconciles accounting entries and investigates irregular transactions. In human resources, a tool called Payroll Activity Workspace monitors changes across pay periods to spot discrepancies before processing payroll runs. For logistics, the company demonstrated how an allocation tool balances competing demands for scarce materials, and how a spend tool examines purchasing patterns to detect unapproved buying or breaches of supplier agreements.

Executive vice president Chris Leone stated that business software is shifting toward systems that complete tasks on behalf of workers, leaving staff to set goals and review decisions. Instead of replacing back-office systems, the autonomous tools sit inside the core software database, keeping an audit trail of every automated step.

What this means for Australian businesses

Many Australian small and mid-sized companies already use AI for basic writing, email drafting and customer queries, but practical automation of core administration has remained difficult without custom software development. This release points to several practical adjustments for business owners:

• Set up explicit permission boundaries. Autonomous agents can now access accounting and inventory data directly. Before enabling any automated tasks in your business systems, ensure each tool has strict least-privilege permissions so it cannot approve payments or alter records beyond its remit.
• Audit routine back-office friction. Tasks like invoice cross-checking, inventory allocation and roster adjustments are becoming standardised features inside mainstream business platforms. Review where your administrative staff spend time on repetitive manual checks to see if native platform updates can handle them.
• Demand full activity logs. As AI takes actions inside financial and payroll ledgers, human review remains a legal and operational necessity. Any tool your organisation uses to update records must provide clear audit trails showing who authorised the workflow and what changes occurred.
• Look for built-in capabilities before buying standalone add-ons. Major enterprise software suites are integrating task-based agents directly into their monthly subscriptions. Small firms should check what their existing enterprise resource and payroll providers are shipping before paying for separate niche software.