AI is transforming enterprise applications by enabling automation, enhanced decision-making, and improved efficiency. The integration of AI into enterprise systems is moving towards AI-native applications, where AI is a core architectural element, rather than an add-on feature.
Oracle has been making significant progress in this area. Oracle Fusion AI Agents Studio offers enterprises a powerful framework to build, manage, and deploy AI-powered agents within the Oracle Fusion Cloud Applications Suite. These AI agents empower business users by automating tasks, accelerating insights, and enabling intelligent decision-making across Finance, Supply Chain, HR, and Customer Experience (CX).
The future of enterprise applications is intelligent, adaptive, and proactive. Oracle’s AI Agents Studio, part of Oracle Fusion Cloud Applications, brings AI directly to business users with no-code and low-code tools for building intelligent agents that interact seamlessly with business data, processes, and users.
Fusion AI Agents Studio – What is it all about?
Oracle Fusion AI Agents Studio is a no-code/low-code development platform embedded within Oracle Fusion Applications that is designed for creating, customizing, and deploying AI Agents to automate business processes. This platform aims to integrate AI-powered automation into the Enterprise, empowering users to build an “AI workforce” alongside their human teams.
Real Strength lies in AI Agents Studio Tools
Studio offers several Tools for Various Purposes as listed below, which help in building Custom AI agents from Scratch. Alternatively, users can leverage prebuilt templates or existing standard AI agents and extend their functionality with custom features. These tools have the capability to interact with Fusion Applications and other systems.
Fusion Applications Agent Architecture
***Original Source of this image is from Oracle University Learning Material***
The Studio framework offers the provision to define custom Business Objects, Tools, Agents, Prompts, and Agent Teams.
Agent Team: A structured sequence of steps or actions that the AI agent follows to accomplish a specific business task or answer a user query. Think of it like a playbook or blueprint for the agent, guiding it through
- User Interactions (What questions to ask, What information to collect)
- Backend Operations (What data to fetch, What systems to interact with)
- Decision Points (How to handle different paths based on user inputs/system results)
- Responses (How to reply clearly and helpfully)
Prompt: The System Prompt defines agent’s persona and capabilities. It establishes the scope within which it can operate and the tools it can access. It also describes how it should think about achieving any goals or tasks for which it was designed.
Topics: They are Reusable Instructions and can refine the Scope of the Agent.
Pre-built Templates
AI Agents Studio Offers number of Pre-built Templates across functional Streams HCM, CX, SCM, FINANCE. For e.g.: Under HCM Family we have a prebuilt template for Absences called Leave and Absence Analyst. These Templates can be leveraged to Build New Custom Agents as required and tailored to the business requirement.
Putting AI Agents Studio into Action
Let’s try to build a sample AI Agent use case, “Sales Order Discount Advisor” using the Studio Capabilities and Tools.
The scope of this agent is to assist and support the Order Management and Pricing teams to advise on the eligibility of discounts and discount percentage for a given sales order based on sales order amount.
To Build this kind of agent, we would need Business Object Tools like:
- GetOrderDetails: This Tool will retrieve order details about an order using order number. This Tool Leverages the Business Object “Get Sales Order Details”. This Business Object further leverages the Business Function “GetOrderDetailsByOrderNumber”
- Get_Sales_Order_Amount: This Tool will get the sales order amount based on HeaderID. This Tool Leverages the Business Object “Get Sales Order Amount”. This Business Object further leverages the Business Function “getOrderTotalByHeaderId”
- We also use another Tool GetUserSession to fetch the PersonNumber of the Logged In User.
Defining Tools
Defining “GetOrderDetails” Tool
Tool Type: Business Object, Tool Name: GetOrderDetails , Tool Code: GETORDERDETAILS
Family: SCM, Product: Order Management
Description: This tool will retrieve order details about an order using order number, source order number, product descriptions, customer name, customer PO Number, Order Date.
Add business object “Get Sales Order Details” to the Tool
Defining “Get_sales_order_amount” Tool
Tool Type: Business Object, Tool Name: Get_Sales_Order_Amount, Tool Code: GETSALESORDERAMOUNT
Family: SCM , Product: Order Management
Description: Tool To Get Sales Order Amount.
Add business object “Get Sales Order Amount” to the Tool
Defining Agents
Define Agent “Sales Order Discount Advisor”
Agent Name: Sales Order Discount Advisor
Agent Code: SALES_ORDER_DISCOUNT_ADVISOR
Family: SCM, Product: Order Management, Maximum Interactions: 10
Description: Agent to advice discount for a given sales order based on Sales Order amount.
Prompt: Provide detail Instructions to the agent as shown in screenshot below.
LLM: Default
Add Tools to the Agent
Defining Agent Teams
Define Agent Team “Sales Order Discount Advisor Agent Team”
Agent Team Name: Sales Order Discount Advisor Agent Team
Agent Team Code: SALES_ORDER_DISCOUNT_ADVISOR_AGENT_TEAM
Family: SCM, Product: Order Management, Maximum Interactions: 10
Description: Agent Team to advise the eligibility of discount and discount percentage for a given sales order based on sales order amount.
LLM: Default
Define a logical starter question:
Add Agent to the Agent Team.
Save the Agent Team and Publish.
“Sales Order Discount Advisor” Agent Team is now ready for debug or test execution. To test or debug the agent, you need to click the highlighted icon
Below Screenshots show test execution of agent:
Fusion AI Agent Enablement
The AI agents are deployed on guided journeys to assist with certain tasks in a business process. The agents use contextual documents that are pre-loaded during the configuration and the enablement of the agents. The agents engage with the users in a chat interface which is embedded in the Fusion Applications transactional pages.
Enablement process for AI agent consists of the following tasks.
Enable profile values at the site level to Y (yes)
- ORA_HCM_VBCS_PWA_ENABLED
- ORA_PER_GUIDED_JOURNEYS_SETUP_REDWOOD_ENABLED
- ORA_PER_AGENT_TASK_TYPE_GUIDED_JOURNEYS_ENABLED
Create a new role with following privileges and grant the role to users of AI agents
- HRC_MANAGE_AI_AGENT_PRIV
- HRC_ACCESS_AI_AGENT_CHAT_PRIV
Configure Agents and Publish the configuration
Create a Guided Journey
- Create the Guided Journey in the relevant functional area
- Attach the agent created earlier to the journey
- Activate the Guided Journey
Modify the UI to embed the journey
- Edit the relevant UI page in Visual Builder Studio
- Embed the journey created earlier
- Test the agent functionality in the VB UI by switching to Live mode
- Publish the changes made in the page for final deployment
Navigate to the UI Page Where the Journey is Embedded to access the Agent Functionality.
You will be able to see the Agent Task with “Ask Oracle” Option.
Invoke “Ask Oracle” to use the Agent Functionality in Live Production Environment.
Benefits
Benefits of inbuilt AI in Enterprise Applications include:
Roadmap & Future Enhancements
- Robust Multimodal Agent Support
- Cross-Functional Workflows
- Deeper OCI Gen AI Integration
- Enhanced Adaptive Learning Capabilities
Conclusion
Oracle Fusion AI Agents Studio is a transformational step forward in embedding AI within enterprise applications. By putting AI creation tools in the hands of business users securely and scalably.
Oracle is enabling organizations to become truly intelligent enterprises.
#EnterpriseAI_IsNowBuiltIn_NoMoreaBoltOn
Credits: References are taken from Oracle University Learning Content for the Course Oracle AI Agent Studio for Fusion Applications
