At the Oracle AI World event held in London, Oracle Corporation announced the launch of 22 new agentic applications integrated within the Oracle Fusion Cloud Applications Suite. This strategic expansion represents a significant pivot in the enterprise software landscape, moving beyond traditional generative artificial intelligence to a framework defined by autonomous, goal-oriented agents. These "agentic" applications are designed to perform complex tasks, reason through business problems, and execute workflows with minimal human intervention, effectively transitioning the role of enterprise software from a passive system of record to an active system of outcomes.
The announcement comes at a critical juncture for the global Enterprise Resource Planning (ERP) and Human Capital Management (HCM) markets. As organizations move beyond the initial excitement of Large Language Models (LLMs), the focus has shifted toward practical utility and the mitigation of technical debt. By embedding these agents directly into the Fusion architecture, Oracle aims to provide a turnkey solution for enterprises seeking to operationalize AI without the risks associated with siloed data or unmanaged external AI tools.
The Evolution of Agentic AI in the Enterprise Framework
The introduction of 22 agentic applications is characterized by Oracle leadership not as a radical departure, but as an evolutionary leap made possible by a decade of cloud migration. Steve Miranda, Executive Vice President of Application Development at Oracle, emphasizes that these agents are built to inhabit the existing Fusion ecosystem. Unlike third-party AI "wrappers," these agents inherit the security protocols, legislative compliance settings, and organizational hierarchies already established within a customer’s cloud environment.
Architecturally, the agents function by interrogating the system to provide actionable proposals. For instance, in a financial context, an agent can be tasked with optimizing a company’s cash position. The AI analyzes current liquidity, identifies risks, and suggests specific actions—such as accelerating a particular order or adjusting payment terms. Once a human operator approves the proposal, the agent executes the transaction within the system. This "cockpit" approach transforms the user experience from manual data entry to high-level strategic oversight.
Chronology of Development and Market Context
The path to the London announcement follows a multi-year trajectory of AI integration within Oracle’s stack.
- Phase One (2018–2021): Oracle focused on embedding machine learning (ML) for predictive analytics within ERP and HCM modules, primarily for anomaly detection and forecasting.
- Phase Two (2022–2023): The rise of generative AI led to the introduction of "assistive" features, such as automated job description generation and summarization tools.
- Phase Three (2024–Present): The current shift toward "agentic" AI, where the system moves from assisting the user to executing tasks autonomously based on high-level business goals.
Market data suggests that this shift is timely. According to industry research, approximately 60% to 80% of corporate IT budgets are currently consumed by "keeping the lights on"—maintaining legacy systems, patching on-premise software, and managing technical debt. Oracle’s strategy posits that by migrating to a cloud-native agentic environment, organizations can redirect a portion of this spend toward innovation. The promise is that agentic AI can handle the "drudge work" of business processes, thereby reducing the operational overhead that has historically plagued large-scale IT departments.
Data Sovereignty and the Role of Organizational Discipline
A recurring theme in the deployment of agentic AI is the necessity of high-quality data. Oracle’s approach relies on the fact that Fusion Cloud customers have already undergone the rigorous process of data cleaning and process standardization required for cloud migration. Because the agents sit on top of this standardized data, they have immediate access to historical SEC reporting, legal entity definitions, and approval rules.
Miranda notes that for AI to be effective, it must understand the specific context of the business it serves. "We know where your legal entities are. We know what your balance is saying," Miranda stated during the London briefing. This level of context is often missing in generalized AI models. By leveraging the "system of record" as the foundation, Oracle ensures that the agents operate within the legal and operational boundaries of the enterprise.
For customers still operating in hybrid environments—combining cloud services with on-premise legacy systems—the transition presents more challenges. While AI can be layered over legacy systems for "quick wins," Oracle warns that this does not retire technical debt. The full efficiency of agentic AI is only realized when the underlying business processes, such as payables automation or supply chain logistics, are fully modernized and documented.
Official Responses and Strategic Guidance for Implementation
In response to inquiries regarding the practical implementation of these tools, Oracle has advised a pragmatic, pain-point-driven approach. The company suggests that enterprises should not necessarily start with the most complex AI projects but rather with the most "documented" ones.
Internal case studies from Oracle’s own operations provide a benchmark for success. The company utilizes its own agentic AI for internal technical support. Historically, the "time to resolution" for internal support tickets was measured in days, largely due to the back-and-forth communication required between humans. By implementing agents trained on a robust library of knowledge articles and FAQs, Oracle has reported reducing resolution times to hours.
For external customers, the advice is similar: start where the process is well-defined. If an organization has a clear "desk manual" or set of standard operating procedures for accounts payable, an agent can be deployed almost instantly. The agent learns the workflow, the approval hierarchies, and the relevant regulations, making it an efficient digital extension of the workforce.
Pricing Models and the Consumption of "Action Units"
The economic viability of AI remains a significant concern for Chief Information Officers (CIOs). Oracle has addressed this by integrating agentic AI into existing Fusion subscriptions. However, to account for the high computational costs of AI inference, the company has introduced a consumption-based model.
Usage beyond a base allotment is measured in "Action Units." This model is intended to align costs with value; the enterprise pays for the outcomes the agents achieve rather than just the availability of the software. While pricing for AI remains a moving target across the industry, Oracle’s full-stack control—from the OCI (Oracle Cloud Infrastructure) layer up to the application layer—allows for efficiencies that the company claims will keep costs predictable for the end user.
Broader Implications: The Shift from Transaction to Exploration
The introduction of 22 agentic applications signals a broader shift in User Experience (UX) design for enterprise software. For decades, ERP systems have been designed as a series of screens that guide a user through a step-by-step transaction. Oracle is now rethinking this paradigm, suggesting that the primary purpose of the user interface will shift from "transacting" to "exploring."
In a compliance-heavy environment, the human need to verify data remains paramount. The new UI philosophy focuses on "hiding complexity behind intuitive navigation." Instead of clicking through ten screens to complete a task, a user might interact with a "cockpit" that summarizes the agent’s findings. If the user needs to verify a specific transaction, they can "explore" the source documents linked by the agent. This allows for brute-force simplicity in the daily workflow while maintaining the ability to perform deep-dive audits when necessary.
Market Reaction and Competitive Analysis
The move by Oracle is seen as a direct challenge to other enterprise giants like Salesforce, SAP, and Microsoft, all of whom are racing to define the "agentic" era. Analysts observe that Oracle’s advantage lies in its integrated database and infrastructure layer, which can theoretically provide lower latency and better data security for agentic workflows.
However, the success of these 22 applications will ultimately depend on customer field stories and the ability of organizations to manage change. The "autonomous enterprise" is a concept that requires not just new software, but a shift in organizational culture. As agents take over more decision-making tasks, the role of the human employee will evolve from a data processor to an AI orchestrator.
Conclusion
Oracle’s announcement of 22 Fusion Agentic Applications marks a definitive move to capture the next wave of enterprise AI demand. By focusing on outcomes, inheriting existing security frameworks, and addressing the realities of technical debt, Oracle is positioning itself as a pragmatic partner for the AI-driven future. As these applications move into general availability, the focus will turn to how effectively they can transform the "messy landscapes" of global enterprise data into streamlined, autonomous engines of business growth. For now, the message from London is clear: the era of the passive system of record is ending, and the era of the agentic system of outcomes has begun.