Oracle Fusion AI Agent Studio Foundations Associate (1Z0-1145-1) Practice Test – 35 Free Exam Questions with Answers

Correct answer: D. Reasoning agents can plan, optimize, and execute multi-step workflows, while Answer agents primarily provide contextual guidance or responses.

Reasoning agents are designed to plan, optimize, and execute complex, multi-step workflows using available tools. Standard Answer agents are primarily built to provide contextual guidance or fetch information, lacking the multi-step execution capabilities of their Reasoning counterparts.

Correct answer: B. System Prompts define the agent's role and instructions; Topics refine the scope or context for specific tasks

System prompts establish the overarching persona, role, and global instructions for the agent, while topics narrow the operational scope for specific tasks. Avoid distractors that confuse these configuration elements with visual generation or automated backups.

Correct answer: C. Correctness score · D. Median or P99 latency

Correctness scores and latency metrics like Median or P99 are key operational indicators tracked in the analytics dashboard. Metrics such as user interface color schemes or meeting counts are irrelevant to technical agent performance and runtime behavior.

Correct answer: B. Confirm that the agent response wait time is acceptable and indicates the financial cost of running the agent

Latency metrics confirm the agent responds within an acceptable timeframe, ensuring a smooth user experience. Token counts directly indicate the financial cost and resource consumption of running the agent, which are crucial operational quality checks.

Correct answer: C. Business Object tool for accessing invoice data and Email tool for sending customer communications

The Business Object tool is required to access and interact with the native invoice data within Fusion applications. The Email tool is then necessary to automate the delivery of those generated communications to the appropriate customers.

Correct answer: B. Comparison of evaluation runs for changes · D. Evaluation test data management for storing test cases

Managing evaluation test data and comparing runs across changes are core features of a structured testing framework. These features allow developers to measure agent regression accurately, unlike irrelevant infrastructure tweaks like server overclocking.

Correct answer: C. It enables agents to exchange information from systems outside of Fusion Applications through defined APIs

The External REST tool allows agents to connect and exchange information with third-party systems outside of Fusion Applications via defined APIs. Remember that deep links are strictly used for navigating within Fusion Applications.

Correct answer: C. External REST tool

The External REST tool is the correct choice to retrieve data from external integrations like a separate customer relationship management system. The Business Object tool is only used for standard data structures inside Fusion Applications.

Correct answer: C. Document tool, to enable semantic search to base responses in unstructured documents

The Document tool performs semantic search on unstructured files, enabling the agent to summarize manuals and provide accurate answers. Deep links simply direct users to application pages and cannot interpret document contents.

Correct answer: A. Email tool, to write and send summary emails to recipients

The Email tool is explicitly designed to write and send outbound summary emails to users after transactions occur. Business Object tools interact with application records rather than managing outbound communication notifications.

Correct answer: C. External REST tool, to connect to a system outside of Fusion

The External REST tool is utilized to connect an agent to third-party systems outside of Fusion Applications using public APIs. The Business Object tool is limited to querying internal Oracle Fusion data structures.

Correct answer: C. Using role-based security and Fusion user tokens to authorize sensitive operations · D. Adding human review (human-in-the-loop) before completing sensitive transactions

Role-based security and human-in-the-loop approvals are the primary safeguards for preventing unauthorized sensitive operations. Options suggesting unrestricted access or automatic deletions directly violate core security and governance principles.

Correct answer: A. Reviewing and adjusting the agent's system prompt and topics as needed · D. Performing design time/offline testing to ensure the agent works correctly

Publishing an agent from a template requires reviewing its system prompt and topics, then completing offline design-time testing. Business object groups and semantic search configurations are not prerequisites for publishing. Always validate the agent's offline behavior first.

Correct answer: B. Customize the system prompt of the Benefits Agent template with company policy details · D. Adjust the agent's topics to cover organization-specific benefits scenarios

Administrators tailor prebuilt templates by customizing the system prompt with policy details and adjusting the topics for specific scenarios. Source code modification is not supported in AI Agent Studio, and direct document uploads are handled through assisted data, not templates.

Correct answer: B. Customize the system prompt to align the agent's tone and content to company standards · C. Use a prebuilt agent template designed for narrative reporting

Report generation is streamlined by using a prebuilt narrative reporting agent template and customizing the system prompt for company tone. Modifying visualization libraries or workflow settings for multi-language support are not standard features available within the agent template configuration.

Correct answer: B. Add topics detailing specific compliance requirements by promotion type. · C. Update the system prompt with compliance language for industry-specific promotions.

Updating the system prompt and adding specific topics are the correct ways to enforce compliance language and context. Changing the AI model does not inject specific business rules, and adding data assets alone will not reliably enforce strict language requirements.

Correct answer: B. Modify the system prompt to specify tone and language preferences. · C. Create additional topics depicting various customer interaction scenarios.

Modifying the system prompt directly enforces the desired tone and language rules for the agent's outputs. Adding relevant topics helps guide the agent through various interaction scenarios, whereas creating new skills or adding unprompted knowledge will not dictate the writing style.

Correct answer: C. Select and customize a pre-built agent template.

Selecting and customizing a pre-built agent template is the most efficient way to handle standard workflows involving both internal and external data. Building from scratch is unnecessarily complex, and generative AI alone cannot execute structured tool-based workflows.

Correct answer: B. To visualize execution steps and identify process errors

Agent tracing is used to visualize the execution steps of an agent, making it easier to identify where process errors occur. For the exam, associate tracing directly with debugging and troubleshooting an agent's logic flow rather than editing fields.

Correct answer: A. Correctness and Latency

Correctness and latency are the primary metrics for evaluating both the accuracy of the agent's responses and the customer wait time. Token usage is helpful for cost tracking but does not directly measure workflow accuracy.

Correct answer: A. Prune the response fields returned from business object queries to only what's needed. · B. Implement role-based controls to allow only authorized roles to execute certain queries.

Pruning response fields minimizes unnecessary data exposure, while role-based controls enforce standard security perimeters. Removing tools entirely would break core functionality, and monitoring metrics is reactive rather than a preventative security measure.

Correct answer: A. Access business objects within Fusion where the secured data resides

Accessing business objects directly within Fusion ensures the agent respects native role-based security while retrieving real-time data. Exporting data broadly or masking access undermines both security and the accuracy needed for live operations.

Correct answer: B. To review specific prompts for analyzing inputs, outputs, and performance metrics

Agent tracing allows administrators to inspect individual prompts to evaluate inputs, outputs, and performance metrics. The dashboard is strictly for monitoring and analytics, so it cannot modify security roles or permissions.

Correct answer: C. Run an evaluation of each agent using expected inputs and outputs

Running a formal evaluation uses expected inputs and outputs to measure accuracy and compare agent versions objectively. Token counts only indicate resource usage, while live prompt changes do not provide reliable quality comparisons.

Correct answer: A. To customize the agent's system prompts and topics for organization-specific requirements · C. To extend the agent's functionality by adding new tools or integrations unique to a use case

Cloning an out-of-the-box template allows developers to safely customize prompts and add unique integrations for specific business needs. You cannot bypass security controls, as cloned agents still inherit standard Fusion role permissions.

Correct answer: C. Benefits Agent

The Benefits Agent is the correct out-of-the-box template for answering employee questions about medical, dental, and vision plans. Help Desk agents focus on service requests rather than specific HCM benefit enrollments.

Correct answer: B. Google Search Agent · C. Benefits Agent

The Google Search Agent and Benefits Agent are provided as prebuilt templates. Inventory Audit and Payslip agents are not standard prebuilt templates in the studio. Memorize the core Oracle-provided templates, as custom agents must be built entirely from scratch.

Correct answer: C. Performance Monitoring Tool

The Performance Monitoring Tool provides detailed trace information, allowing developers to dissect the agent's step-by-step execution. Basic configuration screens or standard analytics dashboards lack the deep diagnostic capabilities needed to review this granular trace data.

Correct answer: A. Set agent permissions to restrict certain update actions to specific business users.

Configuring role-based security and permissions restricts specific update actions to authorized business users, directly protecting sensitive data. However, adding a human-in-the-loop step is also a primary security feature, creating slight ambiguity between defensive controls.

Correct answer: C. Implement human-in-the-loop validation to review the agent's responses

Implementing human-in-the-loop validation allows a person to review critical decisions before execution. While natural language processing analyzes intent, it cannot guarantee accuracy for critical decisions. Rely on human validation to catch potential errors.

Correct answer: D. Error Rate and Median/P99 Latency Reporting

Error rate tracks how often agents fail, while latency reporting measures query resolution speed. LLM requests and prompt injection metrics do not provide direct insights into resolution speed or general performance quality. Map executive goals directly to these operational metrics.

Correct answer: D. Error Rate and Median/P99 Latency Reporting

Error rate and latency reporting provide continuous, automated telemetry on agent reliability over time. Correctness is usually measured through targeted test evaluations rather than live dashboard monitoring of production traffic.

Correct answer: A. When an agent needs to access data from a third-party logistics system · C. When an agent needs to retrieve data from an internal compliance database

The External REST tool is required to interact with systems outside Oracle Fusion, such as third-party logistics or isolated internal databases. Native Fusion data is accessed using standard business object tools instead.

Correct answer: C. Compensation Agent

The Compensation Agent specifically handles salary planning, benchmarking, and compensation decisions. The Payslip and Benefits agents serve distinct purposes like viewing pay statements or managing insurance enrollments.

Correct answer: D. Personal and Employment Details Agent, which retrieves personal and employment details and can generate transactions.

The Personal and Employment Details Agent provides the necessary access to employee data and can initiate transactions. The other options describe incorrect capabilities for their respective agents, like managing benefits or sending notifications.