Oracle Fusion Appsが企業向けに安全で専門的なAIエージェントを提供

3 Min ReadAs enterprises seek to use agents within safe, secure environments, Oracle on Tuesday updated its AI Agent Studio for Fusion Applications and introduced Fusion Agentic Applications, a new group of enterprise applications powered by specialized AI agents. The update to Oracle AI Agent Studio for Fusion applications lets enterprises build, connect and run AI automation and agentic applications. It includes an Agentic Applications builder that enables enterprises to build agentic applications from Oracle, its partners, and external agents using natural language, without coding or application development. Meanwhile, a new workflow orchestration tool lets enterprises coordinate multi-step, multi-agent execution. The studio also includes an Agent ROI dashboard that enterprises can use to understand the business effect of their AI workflows. The monitoring, observability, and prompt playground help organizations build trust in agent deployments, adjust prompts, and scale agents through real time visibility and testing, Oracle said.Related:OpenAI Updates Agents SDK, Aims at Building Secure AgentsThe updates to Oracle AI Agent Studio support the new Fusion Agentic Applications platform, Oracle said. Fusion Agentic Applications can make and execute decisions within business processes by accessing unified enterprise data, workflows, policies, approval hierarchies, permissions and transactional contexts. Applications are composed of teams of AI agents with specific roles, expertise, and decision authority. They maintain shared context, can reason through each situation and adjust accordingly, and run AI-powered workflows based on each role's access level and approval frameworks.Both the update to Oracle AI Studio and the new Fusion Agentic Applications come at a key time in the AI market, as agentic AI is growing in popularity, and enterprises are trying to incorporate agentic AI into their workflows. With the release of OpenClaw, now part of OpenAI, and major vendors such as Nvidia adapting the popular personal framework for more palatable use in enterprise environments, Oracle has also adapted its AI studio to help businesses run AI agents in secure, safe environments.Trusting AgentsThe updates and new agentic applications are directed at helping enterprises hesitant to grant AI agents full autonomy."The core enterprise pain point Oracle is addressing is the lack of trust in autonomous execution, as organizations are hesitant to allow agents to take actions across critical systems without clear boundaries, traceability, and human involvement," said Arun Chandrasekaran, an analyst at Gartner.Related:As AI Infosec Woes Heighten, IBM Intros Autonomous Security ServiceHe added that Oracle's approach to helping enterprises build that trust is to embed the agents in "controlled environments with strict governance, identity management and auditability.""However, this safety may result in constraining the agent's scope and autonomy," Chandrasekaran continued.DifferentiatorNevertheless, Oracle’s latest updates and new platform are valuable for organizations interested in building and experimenting with agentic AI, said David Nicholson, an analyst at Futurum Group."Being able to play around with and create your own agentic workflows, your own agents within the walled garden that Oracle represents and then the ability to use off-the-shelf agents that Oracle has developed to work within that environment that you already have is very, very powerful," Nicholson said. He added that what differentiates Oracle's agent strategy from those of other vendors, such as Nvidia (which is also trying to help enterprises run agents within secure environments with its NemoClaw framework), is that the Oracle is known for managing and protecting data. Therefore, for enterprises choosing between Oracle and others, the question is: what core competency does a cloud provider have that will be helpful to the enterprise's core business value.Related:Anthropic Tool Speeds up AI Agent Development for Enterprises"This is one of those decision points where enterprises have to ask themselves, 'who is better suited to help us with agents that are going to be interacting with our data?'" Nicholson said.About the AuthorNews Writer, AI BusinessEsther Shittu brings four years of expertise covering artificial intelligence technologies and industry trends. As co-host of the "Targeting AI" podcast, she talks to thought leaders and practitioners exploring critical AI developments. Previous to AI Business, she wrote for several publications including the New York Daily News, Bklyner and the Brooklyn Daily Eagle. When she's not diving deep into the world of AI, she spends her time on passion projects and raising her three daughters.