Oracle AI agent builder brings no-code, low-code and pro-code together

To bring AI agent development under one roof, Oracle today updated AI Agent Studio for Fusion Applications.

The AI agent builder combines no-code, low-code and pro-code tools, which can accommodate line-of-business users building simple task-based agents. It also supports developers, who can build more complex agents with Git-based workflows and coding tools such as Claude Code, and they have access to sandboxes and other familiar testing tools.

Oracle also released Fusion Agentic Applications, native runtime apps that can manage Oracle and third-party AI agents and AI automations within Fusion cloud apps, spanning CX, HR, ERP and supply chain.

The new tools and apps build on Oracle's previous "agent swarm" orchestration, which deploys multiple lightweight agents together to execute more complex work, said Holger Mueller, an analyst at Constellation Research. Fusion AI Apps update traditional apps for the AI era.

"You don't use your Fusion apps and add the agent," Mueller said. "You build a new app, which uses an agent for what an agent does well, and then there's other stuff that the human still has to do."

Many cloud companies in the CX realm -- including Pegasystems, Adobe and Genesys -- are working to limit LLM calls by combining deterministic workflows with probabilistic ones and by setting limits such as monthly caps on AI agent costs.

Mueller added that, compared to this time last year, enterprises are no longer looking to LLMs to accomplish everything in sight. Rather, they want to apply AI only where it makes sense -- logically and economically -- in the workflow.

"Either my agent returns me my sales numbers, or it doesn't," Mueller said. "There's nothing probabilistic in that."

Fusion Apps aren't strictly an AI harness; they are business applications with an outcome in mind, such as improving collections, workforce optimization or service call deflections. But they utilize some of the same capabilities, such as security and governance rules within the Fusion environment, said Kaushal Kurapati, head of the Agent platform at Oracle.

"[It's a] combination of AI-assisted reasoning with LLMs, but yet a deterministic runtime execution," Kurapati said. "So you get the flexibility and power of the LLM and the predictability, reliability, consistency and speed you need for executing an enterprise workflow."