Oracle launches agentic AI tools & Deep Trust architecture

Oracle has introduced new Fusion Agentic Applications and a Deep Trust AI Architecture, extending its push to embed AI across its business software portfolio.

The updates span Oracle Fusion applications in finance, human resources, supply chain, sales and service. Oracle says 25 Claw-powered agentic applications are now available, designed to analyse options, prepare actions and carry out approved work within company policies and controls.

Alongside the applications, Oracle launched Fusion Companion and Fusion Agentic Home, two new AI-based interfaces for employees. The tools are intended to let users define business goals, assign work to AI systems and monitor progress, including through supported external tools such as ChatGPT.

Oracle also outlined a security and governance framework called Deep Trust AI Architecture. It is designed to apply security, privacy and business rules across four layers of the technology stack: software platform, data, data access and application development.

Chris Leone, Executive Vice President of Applications Development at Oracle, said the changes reflect a shift in how companies use enterprise software. "The next era of enterprise AI is all about delivering software that increasingly does the work, while people define the objectives and set the policies," Leone said.

"Fusion Agentic Applications, powered by Fusion Claw, can take on highly complex business processes economically and at scale. And with new AI experiences like Fusion Companion and Fusion Agentic Home, people can describe what they want to accomplish and delegate that work to AI agents and agentic applications. These innovations change how people work with enterprise software, giving them more capacity to focus on the decisions that matter while retaining control and accountability," he said.

Application rollout

The first set of agentic applications covers a wide range of operational tasks. In finance, Oracle highlighted tools such as Ledger and Payroll Activity Workspace, which review accounting entries, reconcile information, compare pay and cost data with prior periods, and propose follow-up steps for approval.

In HR, Oracle pointed to Workforce Orchestration, which assesses hiring, redeployment and external talent options to help shape workforce plans. In supply chain, Supply Planning Allocation is aimed at scarce-material allocation decisions, while Procurement Spend is designed to identify off-contract buying and savings opportunities.

Customer-facing functions are also included. Oracle says CPQ Test-to-Production Setup & Configuration Migration helps prepare and move approved configuration changes into production, while Service Health Check and Strategic Field Demand Planning focus on service performance issues and field workforce coverage.

These applications rely on Fusion Claw, an execution runtime intended for longer-running and more complex work. Oracle says the runtime supports replanning, deterministic execution and tighter governance around autonomous actions.

Intent to action

Fusion Companion acts as an intelligent layer across Fusion Applications, using business context, enterprise data and user permissions to determine which workflows, transactions, approvals, agents and applications should be involved in a task. Oracle says it can also work with information from other enterprise systems while respecting existing permissions and controls.

Fusion Agentic Home is positioned as a personalised interface built on top of that layer. It lets staff state an objective, hand work to AI systems and track the status of that work through to completion.

The announcements suggest Oracle is trying to move AI in enterprise software beyond assistance and into execution, while keeping a human approval step for higher-stakes processes. The company describes agentic applications as a new class of outcome-based software in which users set objectives and authority, and the software coordinates work to reach the result.

Trust framework

The Deep Trust AI Architecture addresses a related issue: how companies manage risk as AI systems gain wider access to business data and processes. Oracle says the framework brings together functions from Oracle AI Database, Oracle Autonomous AI Database and Oracle APEX AI Application Generator.

At the platform level, Oracle says Autonomous AI Database can automatically apply security updates across the database engine, operating system, storage and firmware before the updates are made public. It adds that rolling patching allows those changes to be introduced while applications continue to run.

At the data layer, Oracle is promoting Deep Data Security and SQL Assertions as ways to enforce user access rules and business limits in the database itself rather than only in application code. That approach is intended to stop AI-generated applications or direct database access tools from bypassing policy controls.

For data access, Oracle says Deep Trust Data APIs limit AI systems to specific business objects rather than unrestricted SQL access. It also introduced Deep NL2SQL, which adds business and data context to natural-language database queries, alongside feedback and verification tools designed to improve the accuracy of AI-generated responses.

On application development, Oracle says APEX AI Application Generator lets AI coding systems translate natural-language requests into application definitions for Oracle APEX instead of producing large volumes of executable code. Oracle argues that this model reduces maintenance and security risks because the runtime engine, rather than generated code, enforces core protections.

Juan Loaiza, Executive Vice President of AI Database Technologies at Oracle, said trust remains a central barrier to broader AI use in business systems. "AI models can now enable amazing productivity and innovation. However, enterprises can't realize these benefits if they can't trust what AI generates. AI trust is the biggest challenge we face today," Loaiza said.

"With Oracle Deep Trust AI Architecture, we are building trust into the underlying software platform, data, data access, and application development layers. This enables organizations to put AI to work while maintaining control over the information and actions that AI agents and AI-generated applications can take," he said.

Steven Dickens, Chief Executive Officer of HyperFRAME Research, said Oracle is focusing on a foundational issue in enterprise AI adoption. "The enterprise AI race is entering a new phase, and Oracle is shifting the competitive battleground from model intelligence to architectural trust," Dickens said.

"While much of the industry is still grappling with securing AI agents and governing access to enterprise data, Oracle is addressing these challenges at the source. With Deep Data Security enforcing policies directly in the database and Deep NL2SQL bringing business context, continuous learning, and verification to AI-generated answers, Oracle is fundamentally changing the equation for enterprise AI. What sets the Deep Trust AI Architecture apart is its unified approach to embedding trust into how AI accesses, understands, and acts on enterprise data. Oracle is raising the industry bar by unifying intelligence, security, and governance at the foundation of the agentic enterprise, empowering organizations to innovate faster, scale AI with confidence, and deliver more trusted business outcomes," he said.