AI agents are rapidly changing how enterprises interact with applications, data, and business processes. But the real promise of enterprise AI lies beyond conversational responses — it lies in enabling intelligent agents that can understand intent, execute actions, and orchestrate workflows across business functions.
Oracle Fusion AI Agent Studio brings this vision to life by providing a framework to design, extend, and customize AI-driven experiences within Oracle Fusion Applications. Among its capabilities, Workflow Agents introduce a powerful pattern for automating business processes by combining conversational intelligence with task execution and decision logic.
If you are getting started with Oracle Fusion AI Agent Studio, building a simple “Hello World” Workflow Agent is the perfect way to understand the fundamentals. While the example itself may be straightforward, the concepts behind it lay the foundation for creating more sophisticated enterprise agents capable of handling approvals, notifications, data retrieval, and process orchestration.
The business value of Workflow Agents is significant. Organizations can streamline repetitive tasks, improve operational efficiency, reduce manual intervention, and deliver faster, more contextual user experiences. Instead of navigating multiple screens or manually triggering actions, users can interact naturally with AI agents that understand requests and coordinate the underlying workflow execution.
This hands-on example is designed to demystify the process and help you understand how Workflow Agents are configured, how actions are defined, and how Oracle Fusion AI Agent Studio enables a low-friction path from idea to implementation.
In this blog, we will build our first Workflow Agent (“Hello World”) in Oracle Fusion AI Agent Studio, walk through the configuration steps, and explore how this simple use case serves as a stepping stone toward building intelligent, business-ready AI workflows.
Create Workflow Agent
Login to your Oracle Fusion instance and navigate to Tools > AI Agent Studio. Within AI Agent Studio, click on Agent Teams from the panel.
Click on Add.
Enter an Agent Team Name. Agent Team Code will auto-populate once you tab out of the Agent Team Name field.
Under Family, I enter Common and enter Other for Product. Make sure to choose the correct Family and Product based on the requirement.
Under Type, I choose Workflow. This is where we define that we are creating a Workflow Agent.
I enter a brief description.
If you want to expose this workflow agent to Agentic Applications, you can do so by enabling the option below. For the purpose of this demo, we will keep it disabled.
Navigate to the LLM tab. You can use the default large language model.
You can also chose from the list of other available LLMs. Note that there might be additional charges to use these models. I would highly recommend checking with your Oracle CSM on the usage costs, in case you want to use any of these models.
If you are using one of the paid models, you have the flexibility to restrict the usage.
Under the Security tab, you can specify who should have access to use this Agent Team. This is for testing purposes, so I add users having the Application Implementation Consultant role to have access to the agent. In your real-world scenario, add the roles appropriately.
Navigate to the Error Handling tab to configure email notifications for Workflow Agent failures. If the agent encounters an error during execution, notifications can be automatically sent using the settings defined here.
Click Create.
The Workflow agent will be creayted.
You can toggle the layout of the editor by clicking on the Change Layout option.
Now let us add a LLM node to our workflow. You can do that in two ways.
Option 1: Select the stop icon denoting the end of the pipeline and then click on the Add button beside the LLM node in the tools panel.
Option 2: You can click on the Add button between the start and end of the pipeline, and select the LLM node, as shown below.
Either way, the New Node page will pop up.
Enter a Name of the node, the Code will auto-populate.
I will now add a prompt. However, I want the prompt to be dynamic rather than hardcoded. Oracle Workflow Agents provide the capability to inject dynamic values into prompts, allowing prompts to adapt based on runtime context.
For this demonstration, I will add two dynamic values to the prompt: the agent name and the user name.
First, I add the agent name by clicking Add, then navigating to Workflow > Name.
This makes the name of the agent dynamic.
Next I add the user name by clicking Add, then navigating to User > Name.
This completes my Prompt.
Click Create. The new LLM node will be added.
We will keep this workflow agent simple. Now let us test our workflow agent. Click on the Debug button to bring up the chat panel.
The Workflow Agent successfully generates a response for the user — in this case, HCM_IMPL_30 — and correctly identifies itself as XXRM_HELLO_WORLD_WF_AGENT_TEAM. This confirms that the agent is functioning as expected.
Congratulations! You have successfully built your first Workflow Agent using Oracle Fusion AI Agent Studio. The next step is to publish your agent.
Publishing the Workflow Agent
To publish the agent team, click on the Publish button on the top-right-hand corner.
Hope this was helpful. Happy Learning.
References
Leave a Reply