In March 23 story, corrects headline and paragraphs 1, 4 and 6 to reflect that Oracle introduced agentic apps that work across its software suite, rather than individual AI agents

By Stephen Nellis

SAN FRANCISCO, March 23 (Reuters) - Oracle ORCL.N is revamping its cloud-based software suite used by large companies to work with artificial intelligence agents as "agentic apps," with a goal of having humans ask the system for business outcomes.

The changes, which Oracle planned to announce at an event in London on Tuesday local time, are part of a broader trend in which providers of highly specialized corporate software are revamping it to be used by AI agents that can carry out tasks on behalf of human users.

Oracle's shares are down about 40% this year as the company has been swept by investor concerns that AI tools will largely supplant complicated business software. Oracle's executives have argued that the company is embracing AI tools to keep its software ahead of those changes.

In the latest case, Oracle is updating its Fusion suite of software, which includes core business tasks such as planning production in factories and collecting money from customers, among other functions such as human resources.

Steve Miranda, executive vice president of applications development at Oracle, said the company's goal is to make it easier to focus on business questions, such as how to make a new product design cheaper and faster, while minimizing the risks of supply chain disruptions.

The data needed for those decisions, Miranda said, is scattered among the various applications in Oracle's suite and third-party software connected to it. Teams of AI agents will take on tasks such as entering and gathering data and making recommendations needed to reach business outcomes, while for human employees there will be more emphasis on skills like knowing how to negotiate with suppliers and what kind of risk tolerance for supply disruption a company has, Miranda said.

"Typing in an invoice isn't a particularly high-value skill to your enterprise or to the person you know who does that part of their job," Miranda said.

"Decision making is still kind of up to that human and weighing the different pros and cons of that case. But certainly the execution, the typing of the invoices, the typing of the purchase order, that is what is going to be replaced in whole in AI."

(Reporting by Stephen Nellis in San Francisco; Editing by Bill Berkrot and Matthew Lewis)

((stephen.nellis@thomsonreuters.com))