Oracle Fusion AI Agent Studio External REST API Tool Integration: Complete Step-by-Step Guide

Oracle AI Agent Studio External REST API Integration Guide

Learn how to integrate External REST APIs with Oracle Fusion AI Agent Studio. Step-by-step tutorial with real business examples, screenshots, best practices, and a configuration guide.

Oracle Fusion AI Agent Studio: How to Integrate External REST APIs with AI Agents (Complete Guide)

The External REST Tool in Oracle AI Agent Studio is a game-changer.

Using External REST APIs, your AI agent can securely communicate with virtually any third-party application that exposes REST services. Whether you need to retrieve live weather information, validate tax details, obtain exchange rates, check shipment status, or fetch customer data from another CRM, Oracle AI Agent Studio enables your AI agents to invoke these APIs during a conversation and present the results in natural language.

In this guide, you'll learn what External REST APIs are, why they're important, how they work, and how to configure them step by step using the Weather Forecast example shown in your screenshots.

What is an External REST API Tool in Oracle AI Agent Studio?

An External REST Tool is a reusable integration component within Oracle AI Agent Studio that enables an AI agent to invoke REST-based web services hosted outside Oracle Fusion.

Instead of relying only on data stored in Oracle Fusion Applications, the AI agent can retrieve live information from external applications and incorporate it into its response.

Think of it as giving your AI agent the ability to communicate with the outside world.

For example:

• Weather forecasting services
• Currency exchange APIs
• Shipping providers
• Banking systems
• Tax validation services
• CRM applications
• ERP systems
• Custom enterprise applications
• Government APIs

Rather than responding only with Oracle Fusion data, your AI agent becomes an intelligent orchestrator capable of combining enterprise data with external information.

Real Business Example

Imagine you're using Oracle Procurement Cloud.

A buyer asks:

"Can I safely schedule delivery for tomorrow in Jaipur?"

Oracle Fusion knows:

• Purchase Order
• Supplier
• Delivery Location

But it doesn't know tomorrow's weather.

The AI Agent calls a Weather REST API, retrieves the forecast, and responds:

"Tomorrow's forecast for Jaipur predicts heavy rainfall. You may want to reschedule delivery to avoid transportation delays."

This is exactly the type of integration demonstrated in the Weather Forecast example in your document.

How External REST APIs Work

The AI agent doesn't "know" the answer beforehand. Instead, it identifies when an external tool is needed, invokes the REST API, processes the returned data, and generates a conversational response.

Before configuring an External REST Tool, ensure you have:

• Oracle AI Agent Studio enabled
• Appropriate permissions to create Tools and Agents
• A working REST API endpoint
• Authentication details (if required)
• A clear understanding of the API request and response structure

For this walkthrough, the Weather API is used as the sample external service.

Step 1 – Create an External REST Tool

Navigate to:

AI Agent Studio → Tools

Click Create Tool and select External REST as the Tool Type.

Provide:

• Tool Name
• Tool Code
• Product
• Family
• Description

In the sample, a tool named Weather Forecasting Tool is created to retrieve weather information for a specified city.

Why this matters: The tool acts as a reusable connector. Once created, multiple AI agents can invoke it without redefining the REST integration.

Step 2 – Configure Authorization

Open the Authorization tab.

Click Add to create a new connection.

Enter:

• Instance URL
• Authentication Type

In the sample, the Weather API endpoint is configured with no authentication because it uses a public endpoint. In production, you may use API Keys, Basic Authentication, OAuth 2.0, or Bearer Tokens depending on the service you're integrating with.

Best Practice: Never hardcode credentials in prompts. Use the supported authentication mechanism provided by the tool configuration.

Step 3 – Define the REST Function

Switch to the Functions tab and add a new endpoint.

Configure:

• Function Name
• HTTP Method (GET, POST, PUT, DELETE, etc.)
• Resource Path
• Description

The Weather example uses an HTTP GET request to retrieve forecast data for a specified city.

Step 4 – Configure Request Parameters

Once the REST function is created, the next step is to define the parameters that the API expects.

Parameters allow your AI Agent to pass user inputs dynamically when invoking the REST API.

For the Weather Forecast example, the API requires information such as:

• City Name

Instead of hardcoding these values, Oracle AI Agent Studio allows you to define them as input parameters.

When a user asks:

"What's the weather in London tomorrow?"

The AI Agent automatically extracts London from the user's question and passes it as a parameter to the REST API.

Step 5 – Configure HTTP Headers

Many enterprise REST APIs require HTTP Headers.

Headers commonly contain:

• API Keys
• Authentication Tokens
• Content-Type
• Accept
• Version Information

In real Oracle Fusion implementations, you'll often configure:

Authorization
Bearer Token

or

x-api-key

This configuration is shown in the Headers section of the External REST Tool.

Step 6 – Click "Create" the External REST API Tool and your tool is ready 

Step 7 – Create Your AI Agent

Now it's time to create the AI Agent.

Navigate to

AI Agent Studio
→ Agents
→ Create Agent

Provide:

• Agent Name
• Description
• Topic
• Instructions

Step 8 – Add the External REST Tool

This is where the magic happens.

Inside the AI Agent configuration, you'll find the Tools section.

Click

Add Tool

Select

Weather Forecasting Tool

Now the AI Agent knows it has an external capability.

Instead of answering solely from Oracle Fusion data, it can now invoke the Weather API whenever required.

Think of Tools as the AI Agent's "skills."

Without tools:

AI Agent = Knowledge Only

With tools:

AI Agent = Knowledge + Actions

The prompt is one of the most important parts of the configuration.

Instead of writing:

Answer weather questions.

Write something like:

When the user asks about weather conditions, use the Weather Forecast Tool to retrieve live weather information. Summarize the response in a conversational format. If the location is missing, ask the user to provide the city name before calling the API.

Good instructions dramatically improve agent performance.

Step 10 – Create an Agent Team

Oracle AI Agent Studio allows multiple AI Agents to work together.

Instead of building one giant agent, Oracle recommends creating specialized agents.

Example:

Travel Agent
        │
        ├── Weather Agent
        ├── Hotel Agent
        ├── Flight Agent
        └── Currency Agent

When a user asks

"I'm travelling to Tokyo next week."

Oracle automatically delegates work to the appropriate agent.

The Weather Forecasting Agent shown in your document can become one member of a larger Agent Team.

Step 11 – Test Your AI Agent

Open the built-in chat window.

Ask:

What's the weather in Jaipur?

Final Thoughts

External REST integration is one of the most powerful capabilities in Oracle AI Agent Studio because it transforms AI agents from simple conversational assistants into intelligent enterprise orchestrators.

Rather than limiting conversations to data stored in Oracle Fusion, AI agents can securely access real-time information from external systems such as logistics providers, banking platforms, government services, CRM applications, and custom enterprise APIs. This enables organisations to automate richer business processes and provide more context-aware responses.

As Oracle continues expanding its Agentic AI platform, mastering External REST Tools will become an essential skill for Oracle developers, integration specialists, and solution architects. By combining Oracle Fusion data with trusted external services, you can build AI agents that are not only conversational but also actionable—helping users make better decisions with information drawn from across the enterprise ecosystem.

Oracle AI Agent Studio, Oracle Fusion AI Agent, External REST API, Oracle AI Agent Integration, Oracle Fusion REST API, Oracle AI Agent Studio Tutorial, Oracle AI Agent External Tool, Oracle Fusion AI, Oracle AI Agent Configuration, Oracle Agentic AI, Oracle AI Agent Studio Guide, Oracle Fusion AI Integration.

 

Oracle Fusion 26B Introduces a New Data Extraction Framework: Goodbye BICC, Hello Read-Only Data Store

                                    Oracle Fusion 26B Data Extraction | BICC Replacement Guide

Oracle Fusion 26B Introduces a Smarter Way to Extract Data – Here's Why It Matters

If your organisation relies on Power BI, Oracle Analytics Cloud (OAC), Snowflake, Microsoft Fabric, or any external reporting platform connected to Oracle Fusion Cloud, Oracle 26B brings one of the most significant platform enhancements in recent years.

Oracle has introduced a new Data Extraction Framework that fundamentally changes how application data is extracted from Oracle Fusion.

The Challenge with Traditional Data Extraction

For years, Oracle Fusion customers have used tools like:

• BI Cloud Connector (BICC)
• BI Publisher
• OTBI
• Custom SQL-based integrations

to extract business data for reporting, analytics, and downstream applications.

The problem?

All these extraction jobs accessed the same production database that users relied on for day-to-day business operations.

Imagine your finance team is posting journals during month-end close while, at the same time:

• Power BI is refreshing dashboards
• A data warehouse is extracting millions of records
• Integration jobs are pulling supplier and invoice data
• Business analysts are running large reports

All of these workloads compete for the same database resources. As data volumes grow, this can impact application performance and increase extraction times.

How Oracle Fusion 26B Changes Everything

Oracle's new Data Extraction Framework solves this challenge by introducing a Read-Only Replicated Data Store powered by Oracle Autonomous Data Warehouse (ADW).

Instead of extracting data directly from the transactional Oracle Fusion database, Oracle continuously replicates business data into a dedicated read-only environment.

Your reporting and integration tools now connect to this replicated data store rather than the live production system.

Production users and reporting workloads are now separated, resulting in a more scalable architecture.

Why This Is a Big Deal

This isn't simply a new user interface or another Redwood enhancement.

It changes the architecture of Oracle Fusion reporting.

Instead of running analytical workloads on your production environment, Oracle now follows the same best practice used by modern cloud platforms:

Operational Workloads → Production Database

Reporting & Analytics → Read-Only Analytics Database

This separation significantly improves system performance and data availability.

New Features Introduced in Oracle 26B

Oracle Is Replacing BICC

Perhaps the biggest announcement is Oracle's recommendation that customers adopt this framework for all new extraction use cases.

Over time, this framework is expected to replace:

• BI Cloud Connector (BICC)
• BI Publisher-based extraction

For organizations planning new reporting or integration projects, this is the strategic direction Oracle is investing in.

Real Business Example

Imagine a global retail company with operations in 20 countries.

Every hour, it refreshes dashboards showing:

• Sales performance
• Supplier payments
• Inventory
• Purchase orders
• Financial KPIs

Before Oracle 26B

These refreshes query the production Oracle Fusion environment, competing with finance users processing invoices and journals.

After Oracle 26B

The dashboards connect to the replicated Autonomous Data Warehouse.

Finance users continue working without reporting workloads affecting application performance.

The result is faster reporting and a smoother user experience.

Final Thoughts

Oracle Fusion 26B's new Data Extraction Framework is much more than a replacement for BICC—it's a foundational change in how Oracle envisions enterprise reporting and data integration.

By moving extraction workloads to a dedicated, read-only Autonomous Data Warehouse, Oracle delivers a modern architecture that improves performance, simplifies development, and prepares organizations for the next generation of analytics and AI.

For customers starting new reporting, data warehouse, or integration initiatives, adopting this framework from the outset will help align with Oracle's long-term roadmap while reducing complexity and improving scalability.

As Oracle continues investing in Redwood, AI, and cloud-native services, this new Data Extraction Framework is likely to become the standard approach for accessing Oracle Fusion data.

Oracle Fusion 26B, Oracle Data Extraction, Oracle BICC Replacement, Oracle Autonomous Data Warehouse, Oracle Redwood, Oracle Fusion Reporting, Oracle Fusion Analytics, Oracle Business Objects, Oracle Cloud ERP 26B, Oracle Data Integration

Oracle fusion AI Agent Pricing Explained (2026): When Is It Free, When Does It Become Chargeable?

Oracle Fusion AI Agent Pricing Explained 2026 | Free vs Chargeable AI Agents

Learn when Oracle Fusion AI Agents are free, when they become chargeable, how AI Units work, and how Oracle calculates AI costs in Oracle Fusion Cloud.

Oracle Fusion AI Agent Pricing Explained: Free vs Pricing

Oracle Fusion is making AI a core part of Oracle Fusion Cloud Applications. Whether you're using ERP, HCM, SCM, CX, or Procurement, AI Agents can automate repetitive work, answer business questions, and improve productivity.

One of the most common questions customers ask is:

"Is Oracle fusion AI Agent Studio free, or do I have to pay for every AI Agent I build?"

The answer depends on what you're building and which AI model you're using.

Let's break it down.

What's Included with Oracle Fusion?

If you're an Oracle Fusion Cloud customer, several AI capabilities are already included with your subscription.

You can typically access:

• Oracle AI Agent Studio

• Oracle-delivered AI Agents

• AI Agent Marketplace

• Built-in monitoring and analytics

• Monthly AI Unit allocation (subject to your subscription)

This allows organizations to explore AI without making an immediate investment in additional AI services.

When Does Oracle Start Charging?

Oracle's latest AI pricing model is based on AI Units, not simply on the number of AI Agents you create.

In general, AI consumption starts becoming relevant when:

• Premium AI models are used

• AI Unit consumption exceeds the allocation included with your subscription

• Agentic Applications are deployed to production (where applicable)

• Additional AI Units need to be purchased

The key takeaway is that the AI model and workload drive costs more than the number of agents.

What Are AI Units?

Think of AI Units (AUs) as Oracle's way of measuring AI usage. 

Every AI interaction consumes resources, such as:

• Running an AI model

• Processing documents

• Calling enterprise APIs

• Generating responses

Instead of charging separately for each action, Oracle measures overall AI consumption using AI Units.

User Request
      │
      ▼
Oracle AI Agent
      │
      ▼
LLM Processes Request
      │
      ▼
AI Units Consumed

When Does an Oracle Fusion AI Agent Become Chargeable?

This is the question almost every Oracle Fusion customer asks.

The simple answer is:

Your AI Agent doesn't become chargeable just because it's custom. It becomes chargeable when it starts consuming billable AI Units—primarily through the use of Premium LLMs or when your included AI Unit allocation is exceeded.

Scenario 1: Using Oracle's Basic LLM (GPT-OSS)(No AI Unit Charges)

If your AI Agent uses Oracle's Basic LLM (for example, GPT-OSS), Oracle does not deduct AI Units for standard AI interactions. Whether you're using an Oracle-delivered agent, a modified agent, or even your own custom-built agent, the AI model itself does not generate AI Unit charges in this scenario.

Examples:

• HR Policy Assistant

• Leave Balance Agent

• Procurement Help Agent

• A custom agent built with GPT-OSS

These remain within the included capabilities as long as you stay on the supported Basic LLM and don't consume chargeable premium AI services.

Scenario 2: Using a Premium LLM (AI Units Start Consuming)

The moment you switch your AI Agent to a Premium LLM, Oracle starts measuring AI usage through AI Units (AUs).

Examples of Premium LLMs include:

• GPT-5 mini

• GPT-4.1 mini

Each AI interaction consumes AI Units based on:

• The selected LLM

• Prompt size

• Response size

• Document processing

• Workflow complexity

In Oracle's published examples, a typical premium AI interaction may consume around 5 AI Units, although actual consumption varies by workload.

How Many AI Units Do You Get for Free?

Every eligible Oracle Fusion subscription includes a monthly allocation of 20,000 AI Units.

Think of it as a monthly AI allowance.

Oracle Fusion Subscription
          │
          ▼
 20,000 AI Units / Month
          │
          ▼
Used by Premium AI Requests
          │
          ▼
When the allocation is exhausted,
additional AI Units must be purchased.

Unused AI Units do not roll over to the next month—they reset with each billing cycle.

Once you consume the monthly allocated AI units additonal 100,000=$1,000 AI Units are pooled as additional capacity.

How Are AI Units Consumed?

AI Unit consumption depends on the type of work your agent performs.

The key takeaway is that AI Units are consumed based on AI processing, not simply because an agent exists.

So, Is It the Agent or the LLM That Determines Cost?

The easiest way to think about Oracle's pricing model is:

• Standard Agent + Basic LLM = Included with your subscription (subject to Oracle's licensing terms).

• Custom Agent + Basic LLM = AI Unit charges generally don't apply for the Basic LLM itself, though other licensing considerations may still apply.

• Any Agent (Standard or Custom) + Premium LLM = AI Units are consumed.

• Once your included monthly AI Unit allocation is exhausted, you'll need to purchase additional AI Units to continue using premium AI capabilities.

Key Takeaway

The biggest misconception is that "custom agents cost money."

Under Oracle's current AI Units model, the primary driver of cost is the AI model (LLM) and the amount of AI processing performed—not whether the agent is Oracle-delivered, modified, or custom. Choosing a Premium LLM and increasing AI usage will consume your monthly AI Unit allocation. Once those included AI Units are exhausted, additional usage becomes chargeable.

Disclaimer: Oracle updates AI capabilities, licensing policies, and pricing with each release. Always refer to the latest Oracle Fusion Cloud documentation, Oracle AI pricing guides, and Oracle Cloud Service Price List before making licensing or purchasing decisions. How oracle charges for ai agents ?