Oracle shifts focus of Fusion Cloud Applications to AI agents

New agentic AI for Fusion apps

The new agents work across the Fusion suite's finance, HR, supply chain and customer experience (CX) applications.

"We are taking a more holistic approach than ever before," said Rondy Ng, Oracle's executive vice president of applications development, in an interview. "In finance operations, all the static workflows that people have been counting on to actually run the business, or even some of the manual activities, are going to be superseded by autonomous agentic capabilities that are a lot more context-aware. The system is going to inform you about the things that you care about."

Four new finance agents are available -- at no cost, Oracle said -- in the Fusion ERP and Enterprise Performance Management applications.

• A payables agent takes over much of the invoice processing that burdens accounts payable departments. Oracle claims it can extract and normalize data from invoices in various sources, such as emails and PDFs, match it to purchase orders and receipts, run compliance checks and route invoices for approval and payment.
• A ledger agent alerts accountants to discrepancies between the general ledger and the records that hold its data and automatically makes adjustments.
• Geared to the needs of financial planning and analysis professionals, a planning agent provides trend and variance analysis, makes predictions based on real-time data and guides what-if simulations.
• A payments agent is designed to optimize cash flows by evaluating and managing payment and financing options while supporting the onboarding process between banks and suppliers. It also monitors the process for exceptions and payment acknowledgements.

For HCM, Oracle released three agents for performance management and career development. One agent provides information to make one-on-one meetings more actionable for managers and employees. Another agent answers questions about team members' performance and goals and recommends next steps for career advancement. A concierge agent helps managers answer employees' questions about compensation, leave and other issues.

Bonnie Tinder, CEO of Raven Intelligence, said she was impressed by the sophistication of the HCM agents. "A nurse can ask, 'How can I increase my take-home pay without working more hours?'" Tinder explained, adding that the agent gathers the context from the employee's records to provide a useful answer. It knows the certification level of the nurse and suggests additional certifications needed to move into new roles. The employee can then ask the agent for a plan. In that scenario, the agent responds like a human career coach, she said.

In supply chain management, three new agents partly automate requisition, sales-order and fulfillment processes. For example, the fulfillment agent retrieves orders, optimizes picking and packing and expedites high-priority orders.

The three CX agents focus on supporting sales with customer analytics and service. One agent identifies customers who are most likely to buy, while another watches customer sentiment to spot service requests at risk of being escalated. A deal advisor suggests price data, product guides and other information that could help close specific deals.

AI-infused databases form the foundation

The new AI agents, like the SaaS Fusion applications, run exclusively on Oracle Cloud Infrastructure (OCI), the vendor's suite for computation, database, networking, storage and related services, including integration and development. OCI is Oracle's entry in the hyperscale cloud market led by AWS, Google Cloud and Microsoft Azure.

"All of the applications run on that singular database infrastructure," Tinder said. "It makes it so much easier for the applications and AI to be unified on a single database as opposed to integration between the different suites."

OCI is also the mechanism from which Oracle delivers most of its AI technology and its biggest differentiator in the AI arms race against SAP and other ERP vendors, according to analysts. These vendors piggyback on other hyperscalers or have less feature-rich cloud services of their own.

Indeed, many SAP customers still run their SAP ERP systems on Oracle databases, despite SAP's 15-year effort to lure them to its own HANA in-memory database.

"It's not just the breadth and depth of the application portfolio -- which is significant and impressive -- it's the realization of how much of a flywheel, if you will, they're sitting on," said Brian Sommer, president of the research firm TechVentive. He likened Oracle's investments in its cloud, data and AI infrastructure to a mass that is accelerating at speeds that other vendors are challenged to match.

Nevertheless, Oracle customers also generally lag behind its pace of innovation, which Sommer said is not a knock against them but a reflection of the technical debt most face in consolidating their data -- typically a prerequisite of realizing business value from AI. Many also have not figured out where to put the "human in the loop" of their increasingly AI-driven business processes. "I'm just not sure they're all there yet," Sommer said.

The other major news on the infrastructure front was the AI Data Platform also unveiled at the conference.

Stephanie Walter, AI stack practice leader at HyperFRAME Research, acknowledged that the data platform is a collection of previously released products but said what's significant is that they are being consolidated.

"You can't just have a great, standalone database if you want to have enterprise AI in your organization," Walter said. "You have to use a lot of different functionality end to end, and having the data platform and making sure that the integrations are frictionless is a big step forward."

Walter called the databases the anchor of Oracle's AI stack and said the vendor is investing so heavily in them for two reasons. One is data consolidation, which relieves customers of having to move around data so AI can use it. The other reason is low latency. "All of these AI processes are working on extremely large amounts of data," she said, and executing AI processes as close to the data as possible or having the database handle the work has speed advantages.

Oracle's moves mirror what SAP showcased at its new SAP Connect conference last week in Las Vegas. SAP also announced new agents that work across its SaaS enterprise applications and shored up its data offerings to make them AI-friendly. It also added role-specific capabilities to its Joule copilot and repositioned it as an assistant that can call AI agents to carry out common tasks.

David Essex is an industry editor who covers enterprise applications, emerging technology and market trends, and creates in-depth content for several TechTarget websites.