Getty Images/iStockphoto
Oracle calls Fusion Agentic Applications next-level AI for ERP
Analysts concur that the 22 new applications represent a significant leap from narrowly focused AI to teams of agents that coordinate tasks to achieve business objectives.
Oracle announced 22 new agentic AI applications consisting of teams of agents that work together to achieve business objectives rather than being limited to narrow tasks. The applications represent a significant leap forward in how AI is used in ERP, Oracle said.
Holger Mueller
The vendor's claim found support among industry analysts. Holger Mueller, vice president and principal analyst at Constellation Research, said Oracle has moved ahead of its ERP rivals, including market-share leader SAP. "Now we have to see when people go live and humans are supervising multiple agents," Mueller said. He expects competitors like SAP and Workday to have more spring in their steps as they race to match what Oracle has done.
In an interview, Chris Leone, Oracle's executive vice president of applications development, said the applications represent an architectural break from Fusion applications' traditional role as systems of record.
"It was always on the human to move work forward. If we had to apply judgment or understand a policy, we did that offline. We came back into the system and updated the system of record," Leone said. In contrast, Agentic Applications "understand" the business context, including policies and histories, and can advance work on their own or recommend actions to the user.
Leone likened the process of using the agents to a manager assembling a team of experts to address a problem or objective -- say, reducing supplier expenditures by 20%. "We can start to have these specialized agents understand what the process is, understand context, have all the expertise that they need, and then recommend specific actions you need to take to drive work forward," he said.
Agents still handle small tasks, but now they can do more by working together to achieve outcomes that serve larger business goals. "Big outcomes are a collection of smaller outcomes," Leone explained. "To hit an earnings-per-share or gross-margin number, you need to do a bunch of smaller things," such as reducing supplier spend by 10% and lead times by a certain percentage or only working with strategic suppliers to reduce risk.
A user can run the agents in human-in-the-loop mode that requires them to approve every action recommended by the AI. Another mode, human in the lead, lets AI agents make some decisions but leaves the more important ones to the human user's judgment. "In the future, as customers feel more confident, they can start to run these agentic applications more autonomously to execute work on behalf of the user," Leone said.
The applications are expected to be available by April 2.
Teams of experts
Tagged with the new brand name Fusion Agentic Applications, the 22 applications, or "workspaces," are teams of AI agents "with specific roles, expertise, and decision authority to determine why, when, and how work should happen to achieve a given objective," Oracle said in a press release.
They cover four broad business functions that each have a dedicated module in the Fusion ERP suite: HR, supply chain, finance and customer experience. For example, a workforce operations workspace is designed to make workforce management more proactive and intelligent by reducing manual data gathering, speeding up approval of schedule requests and reducing payroll issues. A design-to-source workspace connects engineering, supplier and sourcing functions for improved decision-making and lower costs.
A collections application helps finance teams collect cash more quickly, creating a more intelligent, continuous cash flow to improve working capital, Oracle said. A cross-sell workspace purportedly makes campaigns more proactive and revenue more predictable by helping sales teams identify growth opportunities.
Collectively, the applications have capabilities in common that help them reach objectives. They work from a shared, persistent context, which enables them to "remember" the original intent and history of previous decisions and actions, which also means users are less likely to have to reconstruct the context at later steps.
The applications are constantly reasoning and evaluating situations so they can adjust their actions toward the objective. Oracle also said it built enterprise-grade governance and auditability into the applications to maintain role-based access and approvals throughout each workflow.
Brian Sommer
Brian Sommer, president of TechVentive, attributed Oracle's success with agentic AI to its broad, integrated suite of "solid" ERP applications and the methodical way it has built a portfolio of special-purpose agents fine-tuned for working within the suite. "They're still offering virtually all of this stuff as part of a standard subscription price, not as an add-on deal, but they're being smart and efficient about how this agentic AI is being deployed, not just grabbing headlines and costs be damned for the customer."
But Sommer cautioned that ERP vendors, including Oracle, have yet to achieve the promise of AI that can be trusted to respond appropriately to market triggers that have an impact across a company, such as a rise in interest rates. The new Agentic Applications move Oracle closer, however. "They've got their hands full with so much to do with the more discrete, short-term things," he said. "But they have an architecture where they can make this go live really quick, and I guarantee you their smartest integrator partners are going to be working on those things."
New Agent Studio features
The vendor also announced upgrades to its Oracle AI Agent Studio for Fusion Applications expressly to support developing and deploying the new agentic applications.
A new application builder works with agents from Oracle, its partners and other vendors. Users interact with an AI-driven natural-language UI to choose agents, compose workflows and connect enterprise data without having to use traditional development tools, according to Oracle. Other additions support orchestrating enterprise-wide workflows that cross business processes, blending unstructured content with transactional data, maintaining context awareness and monitoring agent performance.
The Agentic Applications and Studio upgrades were unveiled at the Oracle AI World Tour conference in London.
David Essex is an industry editor who creates in-depth content on enterprise applications, emerging technology and market trends for several Informa TechTarget websites.
