Oracle has built a suite of AI agents in its cloud-based enterprise applications that it claims can autonomously make and execute decisions within business processes. But analysts are urging caution, given unanswered questions about data integration and accountability.
Fusion Agentic Applications, announced this week in London, will be integrated with the Oracle Fusion Cloud Applications suite covering finance, ERP, human resources, payroll and supply chain management. Oracle claims there is a structural advantage here. The data needed to train and run these agents already exists within the enterprise application.
Steve Miranda, application development executive at Big Red, framed the transition from process-focused software to results-driven automation as “applications that can reason, make decisions, and act in pursuit of defined business goals.”
Oracle, for example, is promising a Design-to-Source Workspace Agentic Application that it says can work across engineering, supplier, and procurement decisions to create one “coordinated, continuous process.”
However, Balaji Abbabatulla, vice president at Gartner and vendor lead for Oracle, was more cautious, pointing out unanswered questions about how this technology will be implemented in enterprise environments.
“Our position is that this sounds good, but we need to be careful. It doesn’t necessarily look as bright as it sounds. There are challenges internally, which have not been overcome at the moment, but may be resolved over time,” he said.
Gartner said in January that global corporate boards are under pressure to deploy AI agents to their technology teams. Application, database, service layer, and cloud vendors are all vying for the expected bonanza and are looking to build influence over enterprise AI strategies.
Oracle’s pitch is to sell AI Agent Studio for Fusion Applications, which houses an AI agent within a suite of enterprise applications and enables organizations to build, connect, and run AI automation and agent applications. Oracle also launched an AI data platform that integrates data from various sources to build AI agents.
Gartner’s Abbabatulla said that through this platform, Oracle wants to connect to and extract information from non-Oracle repositories and legacy applications such as SharePoint repositories. Big Red provides tools for data and technology professionals to do that, but it’s not automated.
“There is no autonomous way to synchronize these disparate data repositories in the background,” he said.
He added that building agents to run application-based processes requires a lot of effort and will likely cost Oracle money to get the right engineering expertise.
This is a hurdle for some large enterprises that have already invested in data platforms from Databricks, Snowflake, Cloudera, or other vendors, with some efforts reminiscent of the “big data” investment era. Abbatula sees Oracle’s offer as partly defensive, using data in context as an incentive to keep customers within its ecosystem.
“The overhead of the transition is huge because these are investments that people have made over the years,” Abbatula said. “It’s unlikely that this will actually encourage companies to let go of this investment, but I’m sure some organizations will want to try this in addition to other investments they’ve made.”
Oracle and other vendors still have to answer the question of who is responsible when AI decisions go wrong. register I’ve been growing it for several years.
If AI agents make incorrect decisions at scale and quickly, cascading errors can spread unnoticed. So far, Oracle’s answer has been monitoring and auditing tools, but Abbabatulla is unconvinced: “I haven’t seen a clear answer from any vendor on the liability issue.”
Mickey North Riza, vice president of enterprise software at IDC Group, was more bullish, calling it a “significant change” for agent systems that continuously complete work within enterprise software systems.
“Overall, this is a great move for Oracle to position itself as a market maker for agents-as-apps. It won’t be the app with the best UI that will do well, but rather the agent that reliably completes big results with trust and delivers lasting economic impact,” she said.
Oracle, like other major platform vendors, is vying for a piece of that pie as boards of directors put pressure on technology teams to deploy agents. ®
