Oracle has unveiled a new generation of role-based artificial intelligence agents within Oracle Fusion Cloud Applications. Designed to work directly where the data and processes reside, these assistants act as co-pilots that understand the context of each role and help complete tasks without leaving the workflow.
The proposal focuses on an approach of specific responsibilities by function, so that a CFO, HR manager, or purchasing agent receives suggestions and actions aligned with their daily work. The goal is to reduce downtime, minimize repetitive errors, and facilitate more informed decisions with updated information from the system itself.
What Oracle has announced
The company introduces AI agents that operate within the modules of ERP, HCM, SCM and CX of Oracle Fusion Cloud. It is not an external tool, but embedded capabilities that include process status, company policies, and user permissions to provide recommendations and execute specific steps.
These agents can suggest next steps, summarize key information from multiple records, flag anomalies, and, when the user allows, trigger actions such as creating requests, updating files, or opening support cases. All this with the ability to maintain a natural conversation to refine the context and achieve the desired result.
The role-based approach means that each agent is “specialized” in the metrics and controls relevant to their area: a financial profile will see alerts of budget deviations; a personnel manager will receive indications about absences or selectionA purchasing team will have suggestions for renegotiating terms or consolidating orders; and a salesperson will get help prioritizing opportunities and responding to customers.
According to Oracle, these capabilities are designed to respect security and privacy policies of each organization, operating on data that the user can already consult and recording the actions for subsequent auditing.
How they work and what they can do
Agents combine natural language understanding ’s relationship with the operational knowledge of the Fusion environmentThey can initiate their work based on a user question or proactively when they detect a relevant event, such as an anomaly in an approval flow or a significant change in an indicator.
In addition to explaining the reason for a recommendation, agents are trained to execute actions with prior confirmation, respecting business rules, spending limits and segregation of dutiesIf they require more information, they continue the conversation by asking questions to close the task with confidence.
Typical capabilities include: case summaries, draft generation (responses, communications, or descriptions), variance analysis, and creation of prioritized lists. When the task requires human intervention (e.g., validating an exception), the agent prepares the context to expedite the decision.
- Proactivity: alerts and suggestions when relevant data changes.
- Conversation: Natural language questions and answers, with context history.
- Execution: Create and update records within Fusion, with traceability.
- controls: Enforcement of policies, permissions, and audit logs.
The experience is integrated into the usual Oracle Fusion Cloud screens, so the user doesn't have to jump between applications or copy data. The agent provides the right context at the right time and suggests the next step with one or two clicks.
Impact by business areas and use cases
In finance (ERP), the agent can detect deviations in expenses or income and suggest corrective actions, from a budget adjustment to pushing a pending approval. You can also prepare reconciliations with supporting documentation and summaries for review.
In human resources (HCM), it helps prioritize selection processes, identify bottlenecks in ups and downs, summarize performance feedback or draft internal communications, always within HR's own access limits.
In supply chain and purchasing (SCM), you can recommend consolidating orders, look for alternative suppliers in the event of an incident, and point out risks in delivery times, preparing the necessary orders for validation.
In customer experience (CX), assists sales and customer service teams with account summaries, prioritization of opportunities and draft responses consistent with the history, allowing the CRM to be updated upon completion of each interaction.
For IT and management, agents offer options for configuration and governance that define what each profile can do, what data is visible, and how each interaction is logged for compliance and auditing purposes.
Government, security and availability
Oracle proposes a governance model with access limits, traceability, and controls so that agents operate under the same rules as the rest of the system. This includes Role permissions, data policies, and logging of actions executed by attendees with clear attribution.
Regarding adoption, the rollout of these features will be gradual across the various Oracle Fusion Cloud Applications modules, and may vary by region and product. Details are not provided in the shared materials. specific commercial conditions, so each organization is expected to validate availability and scope in its environment.
The approach seeks to have AI work where critical processes reside, respecting security and simplifying the teams' daily work. With agents tailored to each function, the expectation is to reduce times In repetitive tasks, improve the quality of decisions and standardize good practices without complicating the user experience.
