Most people think about AI through the lens of chatbots and task-based productivity tools: things that assist humans in the moment, rather than tools that take action or work in the background. Oracle, a cloud and AI leader that has served the majority of Fortune 500 companies for decades, is redefining how AI influences business outcomes with teams of AI agents that autonomously handle complex workflows, act behind the scenes, and execute goal-oriented reasoning.
To do that, Oracle just launched Fusion Agentic Applications, a new class of AI-powered enterprise applications. With 22 agentic applications and hundreds of pre-configured AI agents spanning finance, HR, supply chain, and customer experience, Oracle aims to reinvent how work works.
“Where we see things headed and what we're delivering now is a new agent pattern,” says Chris Leone, Oracle’s EVP of Applications Development. “Rather than having a workflow with a predefined path or advisor agent that helps in moments in time, we're assembling teams of agents that all work together collaboratively to solve a particular business objective or business outcome that's defined by what we're calling an agentic application.”
Built Into the Business, Not Bolted On
The conceptual leap at the heart of agentic applications moves software from a passive repository—a place where work gets recorded—to an active driver of work. Leone describes the traditional enterprise application environment as “stagnant systems of record,” and positions Fusion Agentic Applications as a “system of outcomes.”
These agentic applications use multiple AI agents at once that work as a team — first defining the problem, then assembling the right specialists, aligning their efforts, and solving collaboratively. Each agent has a distinct role and works toward a common objective.
In a supplier negotiation, for example, those goals might be to reduce supplier spend by 15 or 20 percent, cut inventory lead times, and consolidate sourcing among strategic partners. Every AI agent in the workspace—the one building out requests for quotes, the one analyzing bid performance, the one recommending award decisions—operates with those targets in view. “Any action or recommendation that they make will take that outcome into context,” says Leone, adding that agents explain their recommendations clearly to employees involved. “They say, ‘Hey, you should pick this supplier because they're meeting the 15% criteria and the lead times are down.” In the new Design-to-Source Workspace, agents translate product specifications into qualified supplier options and simulate cost and lead-time trade-offs in real time.
Employee involvement in this workflow can look a few different ways, says Leone: “human in the loop,” where every agent decision requires human approval; “human in the lead,” where routine actions happen automatically within defined guardrails while higher-stakes decisions surface to humans; and fully autonomous operations. As businesses get more comfortable with this style of enterprise AI adoption, they can shift between levels of human oversight.
From Systems of Record to Systems of Outcomes
Embedding agentic AI inside enterprise software relies heavily on a foundation of good, clean data. Oracle Fusion Applications serve as a system of record for enterprise operations with context that third-party tools lack: the actual policies, security frameworks, business objects, approval hierarchies, role-based access controls, and transactional history that define how a specific organization runs.
As pressure mounts in boardrooms to incorporate AI tools, Leone has seen many organizations quickly deploy products like chatbots. But without a systematic, comprehensive approach, he says, these organizations “haven’t really achieved any kind of lasting value where it truly is driving automation.”
Consider the job of a head nurse managing hundreds of other nurses. They must constantly balance schedules, absences, skills, accreditations, and payroll—all while focusing on patient care. Oracle Fusion Applications can reason across all of that data simultaneously, simulate the impact of competing absence requests on coverage, flag compliance risks in real time, and surface recommended decisions for the nurse to approve. “All that reasoning, all of that cognitive load can now be part of the system,” says Leone. The result is that a single employee becomes a more productive contributor, with a team of AI agents handling the cognitive heavy lifting behind the scenes so they can redirect their focus elsewhere.
A finance team can similarly deploy a Collectors Workspace to consolidate account data across systems, assess customer risk, and help enable faster cash collection—compressing days of work into a continuous, real-time process.
“When I come into the system, it’s completely different every single time,” he says of the agentic experience. “Work is always happening. It’s bringing me the latest updates, calls my attention to more complex cases where my expertise is needed, and it’s moving work forward continuously.”
Building the Ecosystem: 22 Apps, Hundreds of Agents, and a Low-Code Builder
Alongside the 22 prebuilt agentic applications and hundreds of pre-configured agents that organizations can immediately put into production, Oracle is expanding the AI Agent Studio for Fusion Applications—its platform for building, composing, and deploying AI agents and agentic applications within Fusion Applications, which is backed by 63,000-plus certified experts available to accelerate deployments. Customers can use the studio to customize Oracle’s out-of-the-box agents for their industry, add specialist agents to existing agentic application teams, or build entirely new agentic applications from scratch.
Oracle is also introducing new natural language, low-code capabilities that make app development accessible to non-engineers. An ROI dashboard quantifies time saved, cost reductions, and productivity gains per agent, giving enterprise leaders the data to justify continued AI investment.
Leone puts it simply: every enterprise system will have to move this direction—the only variable is how fast. “They might approach the problem a little bit differently,” he says. “But everybody will have to move to this system of outcomes where teams of AI specialists are really driving and solving problems.” And for the head nurse, the finance team, or the HR manager, AI is making the day-to-day experience of work more seamless and efficient so they can redirect their attention toward the objectives that demand their unique expertise.
