Oracle has unveiled significant updates to the Oracle AI Agent Studio, part of its Fusion Applications suite. This all-encompassing platform is designed for the creation, testing, and deployment of AI agents and their teams across various business sectors. The recent enhancements broaden the Oracle Fusion Cloud Applications AI ecosystem, introducing a novel AI Agent Marketplace, expanded support for large language models (LLMs), and an array of resources for building agents, alongside a comprehensive network of Oracle-certified AI agent specialists. Chris Leone, Oracle’s executive vice president of Applications Development, emphasized that companies are facing increasing operational complexity and a pressing requirement to expedite AI integration. He stated, “By establishing a robust AI ecosystem centered on Oracle Fusion Applications, we empower customers to tackle intricate challenges efficiently, securely, and with assurance. The AI Agent Marketplace and other advancements in the AI Agent Studio allow our clients to enhance the embedded AI agents in Fusion Applications with industry-specific capabilities developed by our expanding network of systems integrators and independent software vendors.”
### New AI Agent Marketplace Harnesses Partner Expertise for Enhanced AI Integration
The newly launched Oracle Fusion Applications AI Agent Marketplace provides customers with a streamlined method to implement Oracle-validated AI agents created by partners directly within Oracle Fusion Applications. The marketplace includes pre-configured partner templates integrated into the Oracle AI Agent Studio, which aids clients in accelerating automation efforts, enhancing productivity, and addressing complex, sector-specific business issues. Unlike other AI agent marketplaces, the Oracle AI Agent Marketplace is seamlessly integrated into Oracle Fusion Applications, allowing users to access, test, and deploy third-party AI agents directly within their current workflows. Clients can install and oversee validated templates crafted by certified members of the Oracle PartnerNetwork alongside Oracle’s own pre-built agents, all within a cohesive environment.
### Enhanced LLMs and Features Facilitate Multi-Step Workflows in AI Agent Studio
Oracle AI Agent Studio allows users and partners to choose the most effective large language model (LLM) tailored to their business requirements, with compatibility for several LLMs, including those from OpenAI, Anthropic, Cohere, Google, Meta, and xAI. Recent updates to Oracle AI Agent Studio introduce new features such as: Integration and Extensibility Capabilities, which include Model Context Protocol (MCP) support to allow users to enhance agent functionalities with external data and tools through smooth integration with outside systems. The addition of A2A agent cards promotes collaboration among different agents by enabling communication and context sharing via standardized connectors. A credential store has been implemented to help guarantee secure access for AI agents to external services, protecting sensitive data through secure management of API keys and authentication tokens within the Oracle AI Agent Studio.
### Monitoring and Evaluation Features for Enhanced Agent Performance
The updated monitoring dashboard provides users with real-time insights into agent performance, enabling quick detection and resolution of issues related to sessions, latency, error rates, and token usage. The agent performance evaluation feature enhances the effectiveness of AI agents by allowing systematic testing and assessment of performance against predefined quality, accuracy, and safety metrics. Additionally, agent tracing captures detailed execution data related to agent workflows, assisting users in debugging and optimizing agent performance. Performance metrics enable users to track key indicators, such as accuracy, latency, API errors, and token usage, facilitating continuous improvement of agent effectiveness over time. Furthermore, token usage tracking aids in making costs more predictable for clients by measuring consumption for premium LLMs.
### Comprehensive Management Tools for Prompt and Agent Lifecycle
The introduction of prompt management capabilities, including prompt libraries and lifecycle management, allows users to oversee agents throughout different lifecycle stages—from creation and testing to version control—by storing prompts and use cases in a centralized repository. Topics management enhances agent consistency by consolidating all topics used across various AI agents, providing visibility into the capabilities available to specific agents and ensuring uniform prompt parameters for those operating within similar fields.
### Expanded AI Agent Capabilities
The availability of agent templates accelerates the configuration and deployment of AI agents, offering predefined blueprints for common applications within the Oracle AI Agent Studio. The agent builder assistant feature enables users to create agents from scratch, specifying topics, prompts, and tools based on high-level user guidance. Additionally, the AI Agent Studio FAQ agent supports builders with a Q&A assistant that responds to natural language inquiries regarding tasks or projects in Oracle AI Agent Studio, including template configuration, agent creation, publishing, and performance evaluation.
### Advanced Multimodal and Retrieval-Augmented Generation Features
The introduction of Multimodal Retrieval-Augmented Generation (RAG) enhances agent Q&A capabilities by enabling the analysis of diverse content types, including documents, images, and tables. Furthermore, RAG can be utilized over external sources, improving agent performance by incorporating data from documents stored in external repositories like SharePoint.
### Workflow Agents and Node Features
New features for workflow agents include deterministic execution, which ensures consistent and reliable outcomes for critical business processes by allowing predefined workflows with specific results for certain AI agents. The chaining of workflows enhances outcomes for complex multi-step tasks by linking multiple workflows together. The agent node feature enables users to tackle more intricate processes by integrating additional agents into a workflow when dynamic actions are necessary, such as making decisions or interpreting contextual information. The human-in-the-loop capability allows for a balance between automation and oversight by integrating human review and approval within workflows.
### A Trusted Network of Certified Experts
Oracle boasts a network of over 32,000 certified experts who have undergone extensive training in creating effective agents within the Oracle AI Agent Studio. This pool of specialists allows customers to leverage high-performing agents and optimize AI utilization across their workflows. Additionally, this expanding network will enhance the Oracle AI Agent Marketplace by introducing new expert-built agents and templates, ensuring clients have access to reliable and secure AI agents poised to transform their business operations.
### Analyst Insights
“In the ongoing enterprise AI competition, Oracle has established itself as a strong contender. With the introduction of the AI Agent Marketplace, Oracle is setting a new standard,” remarked Mickey North Rizza, Group Vice-President of Enterprise Software at IDC. He noted that the new marketplace, featuring a continually growing selection of partner-built AI agents supported in Fusion Applications, uniquely positions Oracle customers to accelerate their AI adoption. This ongoing innovation reflects Oracle’s dedication to delivering real value and productivity enhancements to its clients in a highly competitive AI environment. Holger Mueller, vice president and principal analyst at Constellation Research, added, “Enterprise application suites that integrate AI capabilities and provide flexible agent development environments are clearly leading the market. Expanding these ecosystems with accessible marketplaces for partner-built AI agents represents a natural evolution, simplifying the adoption and scaling of AI-driven automation for enterprises.”
