Oracle introduces new agent artificial intelligence capabilities to its AI database and launches Fusion Agentic Applications, expanding its efforts to embed AI deeper into business data and enterprise software.
Database updates are targeted at customers building AI agents that work directly with operational and analytical data. The Fusion release targets business users in finance, human resources, supply chain, and customer management. The new software is designed to enable organizations to run AI-driven processes within their existing systems, rather than using separate tools.
Database update
Oracle has added tools that allow customers to build and run agent AI applications without moving data to external systems. This approach combines AI and data in the same environment across operational databases and analytics lakehouses, giving AI agents real-time access to existing enterprise data.
Among the added features is the Oracle Autonomous AI Vector Database, which enables developers and data scientists to build vector-based applications through application programming and web interfaces. This product is currently available on a limited basis and can be accessed through Oracle’s cloud free tier or low-cost developer tier, providing an upgrade path to the broader Autonomous AI Database.
Oracle also introduced AI Database Private Agent Factory, a no-code tool for building and deploying data-driven AI agents and workflows in public cloud or on-premises environments. It is intended to allow customers to create and manage agents without sharing their data with third parties. It includes database knowledge, structured data analysis, and pre-built agents for deeper data exploration.
Another new component, Oracle Unified Memory Core, is designed to maintain AI agent context across vector, JSON, graph, relational, text, spatial, and columnar data within a single system. Oracle says this reduces latency and eliminates the need for external synchronization between different data stores.
Focus on security
Oracle has built much of its database launch around security and data control. New features include Oracle Deep Data Security, which enforces end-user-specific access rules within the database, allowing users and the AI agents acting on their behalf to see only the data they are authorized to see.
This feature is designed to combat risks such as prompt injection and unintentional data leaks by controlling access at the database level rather than in application code. Oracle also launched Oracle Private AI Services Container, which enables customers to run private instances of AI models in public clouds, private clouds, or on-premises environments (including air-gapped configurations).
Oracle Trusted Answer Search was provided as another safeguard. Rather than relying on large-scale language models to generate answers directly, the tool uses vector search to match a user’s question to previously created reports, striving to reduce hallucinations and inaccurate answers.
Oracle has also added features aimed at open standards and interoperability. While Oracle Vectors on Ice supports vector data stored in Apache Iceberg tables, Oracle Autonomous AI Database MCP Server is intended to provide external AI agents and clients with secure access to the database without any bespoke integration work.
“The next wave of enterprise AI will be determined by our customers’ ability to use AI in business-critical production systems to securely deliver breakthrough innovation, insight, and productivity,” said Juan Loaiza, executive vice president of Oracle Database Technologies at Oracle.
“With Oracle AI Database, customers don’t just store data, they activate it for AI. By architecting AI and data together, we help customers quickly build and manage agent AI applications that can securely query and interact with real enterprise data with stock exchange-grade robustness on all major clouds and on-premises.”
HyperFRAME Research analyst Stephen Dickens said Oracle has addressed a central problem in agent-based AI deployments by using a single system for multiple data types.
“In the age of agent AI, a unified memory core is essential for agents to maintain context across a variety of data types, including vector, JSON, graph, columnar, spatial, textual, and relational, without the latency or staleness of external synchronization,” said Dickens, CEO and principal analyst at HyperFRAME Research.
“Only Oracle AI Database accomplishes this in a single mission-critical engine with concurrent transactional and analytical processing, high availability, and robust security, enabling real-time inference on live business data. While organizations without this foundation will struggle with fragmented and unreliable agents, organizations leveraging Oracle will gain a decisive advantage in scalable AI deployments.”
application layer
In parallel with the database changes, Oracle launched Fusion Agentic Applications. It is described as a new class of enterprise applications built into Oracle Fusion Cloud Applications. The software uses a group of specialized AI agents to make and execute decisions within business processes, leveraging corporate data, workflows, approval structures, and permissions.
Oracle said the applications are different from AI assistants and add-on tools because they reside within the transaction system itself, allowing it to operate in real-time with governance controls already in place. They are designed to navigate routine tasks within guardrails and direct humans to exceptions and tradeoffs when human judgment is required.
22 Fusion Agentic applications are available at launch. Oracle highlighted examples such as the Workforce Operations Agentic Application for scheduling and payroll issues, the Design-to-Source Workspace Agentic Application for procurement and engineering decisions, the Cross-Sell Program Workspace Agentic Application for sales teams, and the Collectors Workspace Agentic Application for cash collection.
“Too much time is spent managing processes instead of driving results, and the way we work is no longer aligned with the speed, complexity, and expectations of modern business,” said Steve Miranda, Oracle’s executive vice president of application development.
“With Fusion Agentic Applications, we are moving enterprise software beyond passive systems of record, giving our customers applications that can reason, decide, and act in pursuit of defined business goals. This is a huge step forward for the industry, helping our customers achieve results faster, focus valuable time on strategic activities, and redefine how work works.”
Industry analysts said the announcement signals a broader shift from AI assistants to software that can perform multi-step tasks within core business systems.
“The introduction of Oracle Fusion Agentic Applications represents a meaningful shift in enterprise software by moving beyond task automation to results-driven execution in the transition to an autonomous enterprise,” said Mark Smith, Chief AI and Software Analyst at ISG.
“As organizations look to expand automation across their businesses, having a platform that can orchestrate agents across functions while maintaining security and authorization within the application suite will be a key differentiator.”
