Oracle 1Z0-1155-2 Real Exam Dumps [September 2026 Update]

The 1Z0-1155-2 is the Developer Professional exam, not the Foundations Associate exam, and the implementation depth that distinguishes them is where most candidates misallocate their preparation. The Foundations Associate (1Z0-1145-1) tests whether you understand what Oracle Fusion AI Agent Studio is and how agents are conceptually structured. The Developer Professional tests whether you can make the correct implementation decisions: which tool type an agent needs for a specific data access requirement, how RAG knowledge grounding behaves differently depending on the knowledge base configuration, when multi-agent orchestration is appropriate versus a single-agent workflow, and how the testing framework’s semantic accuracy evaluation differs from tracing. Candidates who prepare using Foundations Associate materials or general AI/LLM literacy content arrive well-prepared for the conceptual domains and consistently underperform on the tool selection, knowledge grounding configuration, and advanced orchestration questions that make up the majority of domains three through six. The 1Z0-1155-2 requires knowing Oracle AI Agent Studio implementation at the level of someone who has built, tested, and deployed agents inside Oracle Fusion Applications – not someone who knows what an agent is.

The Oracle 1Z0-1155-2 (Oracle Fusion AI Agent Studio Developer Professional – Rel 26-2) certifies advanced expertise in designing, building, configuring, deploying, testing, and managing AI agents and Fusion Agentic Applications within Oracle Fusion Cloud Applications. Validated against Oracle Fusion Applications Release 26A / 26-2. Format: 50 multiple-choice and scenario-based questions in 90 minutes. Passing score: 68%. Delivered through Pearson VUE.

Cert Empire’s 1Z0-1155-2 exam questions are built at the Developer Professional implementation-depth the real exam requires – tool selection scenarios, knowledge base configuration decisions, testing framework usage, and pro-code development with the Fusion Agentic Applications builder.

Exam Snapshot

Oracle Fusion AI Agent Studio: What It Is and Why the Certification Matters

Oracle AI Agent Studio for Fusion Applications is the complete development platform for building, connecting, and running AI agents and Fusion Agentic Applications natively inside Oracle Fusion Cloud Applications – ERP, HCM, SCM, CX. It is included with Oracle Fusion Applications subscriptions at no additional cost (as of the 26A/26-2 release).

Unlike generic LLM platforms where agents are built in isolation and connected to enterprise systems through custom integrations, Oracle AI Agent Studio runs inside Oracle Fusion’s security model, data model, and workflow engine. Agents access Fusion business objects through configured tools that respect existing role-based access controls. Every agent action is logged to Fusion’s immutable audit trail. This inside-Fusion-boundaries architecture is a fundamental architectural characteristic that the 1Z0-1155-2 exam tests repeatedly – it determines how tools are configured, how permissions are assigned, and how orchestration integrates with existing Fusion workflows.

Six Exam Domains

Domain 1: Identifying and Configuring Components of AI Agent Studio

AI Agent Studio architecture: The exam tests the structural components of an agent in Oracle Fusion AI Agent Studio: the agent definition (persona, instructions, and behavioral constraints), the LLM configuration (which LLM backs the agent and what parameters control its behavior), tools (the capabilities the agent can invoke), knowledge bases (the data sources grounding the agent’s responses), and the agent context window (what information is assembled and presented to the LLM for each interaction).

GenAI and RAG mechanics in the Oracle context: The exam tests Retrieval-Augmented Generation at the implementation decision level. RAG grounds an agent’s responses in specific enterprise data sources rather than relying solely on the LLM’s training data. In Oracle AI Agent Studio, RAG is implemented through knowledge bases: the agent receives a user query, the knowledge base retrieves semantically relevant content from configured data sources, and the LLM uses that retrieved content plus the query to generate a grounded response. The exam tests what types of content can be indexed in a knowledge base, how retrieval relevance is determined, and when RAG-grounded responses are more appropriate than tool-based responses.

LLM orchestration within enterprise boundaries: Oracle AI Agent Studio orchestrates LLM interactions while enforcing Oracle Fusion’s security, compliance, and data access boundaries. The exam tests that agents cannot access data outside what their configured tools and role-based permissions allow – even if the LLM could theoretically infer or reference such data from training. This enforcement happens at the tool and data security layer, not at the LLM layer.

The four tool types – the most implementation-specific domain 1 topic:

• Document Tool: Enables the agent to search and retrieve content from unstructured documents (Oracle Content Management, uploaded files, knowledge articles). Used when the agent needs to ground responses in document-format content – policy documents, procedures, FAQs, product manuals. The exam tests that Document Tool works through semantic search against indexed document content, not structured database queries.
• Business Object Tool: Enables the agent to read from and write to Oracle Fusion business objects (invoices, purchase orders, employee records, customer accounts) through Fusion APIs. The exam tests that Business Object Tool respects the Fusion security model – the agent can only access records that the configured security context permits, and write operations trigger standard Fusion business process validations.
• Deep Link Tool: Enables the agent to navigate users to specific Oracle Fusion UI pages, records, or transactions. The exam tests Deep Link Tool as a navigation mechanism, not a data retrieval mechanism – it facilitates user action within Fusion rather than retrieving data for the agent’s reasoning.
• External REST Tool: Enables the agent to call external REST APIs outside of Oracle Fusion. The exam tests when External REST Tool is appropriate (integration with third-party systems, external data sources) versus when a Business Object Tool is the correct choice (access to data in Oracle Fusion). The exam also tests what is required to configure an External REST Tool: the endpoint URL, authentication method, and the request/response schema that the agent uses to formulate correct API calls.

Domain 2: Creating and Assigning Roles and Access Permissions

The Oracle Fusion security model applied to AI agents: Oracle AI Agent Studio operates within Oracle Fusion’s role-based access control (RBAC) framework. Agents do not have their own independent identity – they operate with the permissions of the role context in which they are deployed. The exam tests how agent security is configured: which Oracle Fusion roles determine what data the agent can access through its tools, how data security policies (row-level security) apply to agent data access, and what happens when a user without appropriate permissions interacts with an agent that has access to restricted data.

Agent-specific roles and permissions: The exam tests the two key permission dimensions for AI Agent Studio: who can build and configure agents (developer/administrator roles with access to the AI Agent Studio authoring environment) versus who can use deployed agents (end-user roles with access to the deployed agent interface). Separating these permission layers is an exam-tested best practice – production agents should not be accessible for modification by end users, and agent developers should not need production data access to build and test agents.

Data security in Business Object Tool configurations: When a Business Object Tool is configured to access Fusion business objects, the data security policies defined in Oracle Fusion’s security model determine which records are accessible. The exam tests that the Business Object Tool configuration must specify a security context that aligns with the intended user population’s data access – misconfiguring security context can result in either the agent refusing to return data (overly restrictive) or the agent exposing data to users who should not see it (overly permissive).

Audit and compliance: Every agent action that accesses or modifies a Fusion business object through a configured tool is logged to Oracle Fusion’s audit trail. The exam tests that this audit capability is automatic and cannot be bypassed – it is a consequence of the inside-Fusion-boundaries architecture, not a separate configuration.

Domain 3: Configuring and Deploying an Agent Using Guided Journey

The Guided Journey approach – no-code agent configuration: Oracle AI Agent Studio’s Guided Journey provides a structured, wizard-style interface for configuring agents without writing code. The exam tests what Guided Journey configures: the agent’s name and description (displayed to end users), the agent’s persona (the behavioral identity that shapes how the agent communicates), the agent’s system instructions (the behavioral constraints and objectives that guide the LLM’s responses), the tools assigned to the agent, and the knowledge bases connected to the agent.

Persona configuration: A persona defines the agent’s communication style, professional tone, domain expertise framing, and contextual identity. The exam tests how persona instructions shape LLM behavior: a Collections Specialist persona constrains the agent to communicate about accounts receivable topics in a professional, specific tone, while preventing the agent from providing advice outside its configured domain. The exam also tests that persona instructions are part of the system prompt sent to the LLM – they are not a filtering layer that operates independently of the LLM.

System instructions and behavioral constraints: System instructions define what the agent should and should not do: the scope of topics it addresses, how it escalates to human agents when it cannot resolve a request, how it handles ambiguous inputs, and what disclaimers or limitations it communicates. The exam tests the difference between system instructions (behavioral guidance for the LLM) and guardrails (technical controls that enforce constraints on agent outputs regardless of LLM behavior).

Knowledge base connection in Guided Journey: The exam tests how knowledge bases are selected and connected to an agent during Guided Journey configuration: which knowledge bases are available depends on what has been provisioned in the organization’s AI Agent Studio environment, the agent can reference multiple knowledge bases for different knowledge domains, and the agent’s response quality depends on the relevance and freshness of the connected knowledge base content.

Deploying an agent: After configuration, an agent is deployed to make it accessible to end users through configured deployment surfaces: embedded in Oracle Fusion UI pages, accessible through Digital Assistant channels (web, mobile), or surfaced through the Fusion Agentic Applications interface. The exam tests the deployment configuration steps and what testing is required before production deployment.

Domain 4: Agent Knowledge Grounding, Tools, Actions, and Workflow Integration

Knowledge base configuration and content intelligence: Oracle AI Agent Studio knowledge bases index unstructured and structured content for agent retrieval. The exam tests knowledge base configuration decisions:

Semantic search retrieves content based on meaning similarity between the query and the indexed content – effective for natural language queries about concepts, policies, and procedures where the exact keywords may not match. Keyword search retrieves content based on term occurrence – more effective for queries about specific identifiers, codes, or proper nouns. The exam tests when each retrieval strategy is appropriate and how Oracle AI Agent Studio combines both approaches for hybrid retrieval.

Content grounding quality: The exam tests what determines retrieval quality: document chunking strategy (how content is divided for indexing – chunk size affects whether retrieved content has sufficient context), document metadata (enables filtering by category, date, or type), and content freshness (knowledge bases must be reindexed when source content changes, otherwise agents generate outdated responses). The exam presents scenarios where agents provide inaccurate or outdated responses and tests which knowledge base configuration change addresses the root cause.

Actions vs. Tools distinction: The exam tests a precision distinction between tools and actions in Oracle AI Agent Studio. Tools are capabilities the agent can invoke during a conversation to retrieve data or execute operations (Document Tool, Business Object Tool, etc.). Actions are defined sequences of steps the agent can execute as part of completing a specific workflow – they may combine multiple tool invocations with decision logic to accomplish a multi-step task. An action for “process a supplier invoice exception” might query the invoice details, check approval policy, submit for approval, and notify the vendor – coordinating multiple tool calls in a defined sequence.

Workflow orchestration – multi-step and multi-agent execution: The March 2026 Oracle AI World London updates introduced workflow orchestration capabilities that the 1Z0-1155-2 exam validates. Orchestration allows coordinating multi-step, multi-agent execution with rules that control how work moves between steps, built-in logic, and human oversight points. The exam tests orchestration design decisions: when to use a single agent with multiple tools versus a multi-agent workflow where specialized agents handle different stages, how human oversight checkpoints are configured (what triggers a pause for human review versus what proceeds autonomously), and how orchestration maintains state across steps.

Contextual memory: Agents can remember context across interactions, workflows, and agent collaborations. The exam tests contextual memory configuration: what types of context are persisted (user preferences, prior conversation summaries, workflow state), how long context is retained, and how memory boundaries align with security requirements (ensuring one user’s context is not accessible to another).

Domain 5: Testing, Evaluation, and Monitoring

The Oracle AI Agent Studio testing framework: The exam tests the testing framework at the implementation detail level – not just that testing exists, but how each testing capability is used and what it reveals.

Dataset management for evaluation: Test datasets contain representative input-output pairs – sample user queries and the expected agent responses or actions. The exam tests how test datasets are created (manually curated, synthetically generated, or captured from production interactions), how they are maintained as the agent’s instructions, tools, and knowledge bases change, and how dataset coverage affects evaluation reliability.

Semantic accuracy evaluation: Semantic accuracy measures whether the agent’s outputs correctly represent the underlying facts and context – not whether the response is grammatically correct or matches a template exactly, but whether the meaning and factual content of the response align with the ground truth. The exam tests how semantic accuracy is measured (comparison against expected responses using embedding similarity, not exact string matching) and what a low semantic accuracy score indicates about the agent’s configuration.

Tracing: Tracing records the sequence of steps an agent takes to produce a response: which tools were called, in what order, with what parameters, what results they returned, and how those results influenced the final response. The exam tests tracing as a diagnostic tool – when an agent produces an unexpected response, tracing reveals whether the issue originated in knowledge retrieval (the wrong content was retrieved), tool execution (the tool returned incorrect data), or instruction interpretation (the LLM reasoned incorrectly from correct inputs).

A/B comparison testing: The exam tests A/B comparison as a method for evaluating changes to agent configuration – comparing an agent’s performance before and after changing an instruction, swapping a knowledge base, or modifying a tool configuration. The exam tests what metrics are compared (semantic accuracy, task completion rate, response latency, token consumption) and how to interpret comparison results to make configuration decisions.

Guardrails: Guardrails are technical controls that enforce boundaries on agent behavior independently of the LLM’s instructions. The exam tests what guardrails can and cannot do: guardrails can filter harmful content, prevent disclosure of specific data patterns (PII, confidential identifiers), restrict the agent to a specific topic domain, and limit response length. Guardrails operate at the output layer – they evaluate and potentially modify what the agent returns before it reaches the user, not how the LLM reasons. The exam tests why guardrails are necessary even when persona and system instructions already address the same constraints (LLM instructions can be overridden through adversarial prompting; guardrails cannot).

Performance monitoring and token usage: The exam tests monitoring of production agent performance: tracking response latency, token consumption per interaction (affects cost and context window utilization), error rates, tool call failure rates, and escalation rates (how often users are transferred to human agents). The exam tests how token usage monitoring informs optimization – high token consumption may indicate overly verbose system instructions, excessive retrieved content, or inefficient multi-turn conversation design.

Domain 6: Advanced Agent Development and Agentic Applications

Pro-code development with VS Code, CLI, and APIs: The July 2026 Oracle announcement of the AI-native builder experience introduced pro-code development capabilities the 1Z0-1155-2 exam validates. Developers can use Visual Studio Code, standard CLIs, and AI coding assistants (including Codex and Claude Code) alongside Git-based lifecycle management, local validation, debugging, and CI/CD workflows to build complex agentic applications.

The exam tests when pro-code development is appropriate versus Guided Journey no-code configuration: Guided Journey handles most standard agent configurations; pro-code is necessary for complex custom logic, integration with non-standard external systems, programmatic generation of agent instructions, advanced multi-agent orchestration patterns, and deployment automation through CI/CD pipelines.

Oracle AI Agent SDK and APIs: The exam tests the Oracle AI Agent SDK as the programmatic interface for interacting with Oracle AI Agent Studio from code: creating and updating agent definitions, connecting knowledge bases, configuring tools, triggering agent invocations, and reading evaluation results. The exam tests the SDK usage patterns for common development scenarios and when the SDK is necessary versus when the UI-based Guided Journey configuration is sufficient.

Git-based lifecycle management: The exam tests how Git-based lifecycle management applies to agent development: storing agent definitions, tool configurations, knowledge base metadata, and test datasets in version control, enabling rollback to prior agent configurations, supporting collaborative development with multiple agent developers working in parallel branches, and enabling CI/CD automation for agent deployment to different environments (development, testing, production).

Fusion Agentic Applications: Fusion Agentic Applications are a new class of enterprise application introduced with Oracle Fusion Applications Release 26A. They are outcome-driven systems backed by teams of specialized AI agents that reason, coordinate, and decide, then execute work through Fusion business objects, workflows, tools, policies, approvals, and logged actions. The exam tests what distinguishes Fusion Agentic Applications from standard agents: they orchestrate multiple specialized agents, operate against specific business objectives over extended time horizons (not just single interactions), and coordinate execution across the full Oracle Fusion Applications suite.

Multi-agent orchestration patterns: The exam tests three orchestration patterns:

Sequential orchestration: Agent A completes its task and passes its output to Agent B, which completes its task and passes to Agent C. Used when tasks have strict dependencies – each agent’s work depends on the prior agent’s output.

Parallel orchestration: Multiple agents execute simultaneously on different aspects of the same problem. Used when subtasks are independent – each agent works on a different data domain or business area and results are aggregated.

Hierarchical orchestration (orchestrator-worker): An orchestrator agent decomposes a high-level objective into subtasks, assigns them to specialized worker agents, monitors their progress, and synthesizes their outputs into a final result. Used for complex business objectives where no single agent has sufficient scope or tool access.

Responsible AI and enterprise governance: The exam tests Oracle’s responsible AI framework as applied within AI Agent Studio: transparency (agent disclosures to users about AI involvement), fairness (preventing bias in agent recommendations), accountability (audit trail for all agent actions), and privacy (data handling within Oracle Fusion’s privacy controls). The exam tests how guardrails, role-based permissions, and audit logging together implement responsible AI governance – not as abstract principles but as specific configured features.

5 Study Tips for Oracle 1Z0-1155-2

• Tip 1: Study the four tool types (Document, Business Object, Deep Link, External REST) as implementation decision questions rather than definitions. Know which is appropriate for each described data access or navigation requirement and why the alternative tools are incorrect for the same scenario.
• Tip 2: Study the testing framework components at the operational level: what semantic accuracy measures versus what tracing reveals, when A/B comparison is used, and how guardrails differ from LLM instructions as enforcement mechanisms.
• Tip 3: Study knowledge base configuration decisions: semantic versus keyword retrieval, chunking strategy effects on retrieval quality, and how content freshness affects agent response accuracy.
• Tip 4: Study multi-agent orchestration patterns (sequential, parallel, hierarchical) with use case examples for each. The exam presents a business automation requirement and tests which orchestration pattern is most appropriate.
• Tip 5: Practice with Cert Empire’s 1Z0-1155-2 exam questions built at Developer Professional implementation depth – tool selection scenarios, knowledge grounding decisions, testing framework usage, and Fusion Agentic Applications orchestration questions.

Best Study Resources

• Cert Empire 1Z0-1155-2 exam questions PDF and practice simulator (Rel 26-2 aligned, August 2026 edition).
• Oracle AI Agent Studio documentation (docs.oracle.com/en/cloud/saas/fusion/agentai).
• Oracle Fusion AI Agent Studio public GitHub repository (github.com/oracle-fusion-agent-studio).
• Oracle AI Agent Studio learning path on MyLearn (mylearn.oracle.com).
• Oracle AI World London March 2026 session recordings (oracle.com/ai-world).
• Udemy: “1Z0-1155-2: Oracle Fusion AI Agent Studio Developer Exam” practice test course.
• CertificationPractice.com Oracle 1Z0-1155-2 free practice exams.

Career Opportunities After 1Z0-1155-2

• Oracle Fusion AI Engineer
• Enterprise AI Agent Developer (Oracle Cloud)
• Oracle Fusion Applications Architect (AI Specialization)
• AI Automation Consultant (Oracle ERP/HCM/SCM)
• Fusion Agentic Applications Developer

The 1Z0-1155-2 is among the most current and market-relevant Oracle certifications available in 2026 – validated against the 26A/26-2 release, it reflects the platform capabilities that Oracle positioned at Oracle AI World London in March 2026 and expanded with the AI-native builder announcement in July 2026. Organizations deploying Oracle Fusion Applications are actively seeking developers who can build agents and Fusion Agentic Applications to automate enterprise processes. Oracle Fusion AI developers with this credential earn between USD 110,000 and USD 165,000 in enterprise and consulting roles.

Why Candidates Choose Cert Empire for Oracle 1Z0-1155-2 Preparation

✔ Tool type selection scenario questions. Our 1Z0-1155-2 questions present data access or navigation requirements and test which of the four tool types (Document, Business Object, Deep Link, External REST) is correct and why the alternatives are incorrect.

✔ Testing framework implementation questions. We test semantic accuracy measurement versus tracing diagnostics, A/B comparison usage, and guardrail configuration – each at the operational detail the Developer Professional exam requires.

✔ Knowledge base configuration decision questions. Our questions test semantic versus keyword retrieval selection, chunking strategy effects, and content freshness implications for agent response accuracy.

✔ Multi-agent orchestration pattern questions. We present business automation scenarios and test whether sequential, parallel, or hierarchical orchestration is most appropriate, with the reasoning that distinguishes each.

✔ Practice under real exam conditions with the Cert Empire Exam Simulator. Our 1Z0-1155-2 simulator runs 50 scenario-based questions in 90 minutes across all six developer professional domains.

✔ Instant access, 90-day free updates, and 24/7 support. As Oracle updates AI Agent Studio and releases new exam guides, your materials update automatically. Our support team is available around the clock.

✔ Backed by a full money-back guarantee. If our exam questions do not help you pass, we refund your purchase with no conditions.

FAQ’s

What is Oracle 1Z0-1155-2?

1Z0-1155-2 is the Oracle Fusion AI Agent Studio Developer Professional certification exam, validated against Oracle Fusion Applications Release 26A / 26-2. It certifies advanced skills in designing, building, deploying, testing, and managing AI agents and Fusion Agentic Applications within Oracle Fusion Cloud Applications.

What is the passing score for 1Z0-1155-2?

68% – 34 of 50 questions must be answered correctly. The exam runs 90 minutes.

What is the difference between 1Z0-1145-1 and 1Z0-1155-2?

1Z0-1145-1 is the Oracle Fusion AI Agent Studio Foundations Associate exam – it tests foundational conceptual knowledge of what AI Agent Studio is and how agents work. 1Z0-1155-2 is the Developer Professional exam – it tests implementation decisions: which tools to configure for specific requirements, how to design knowledge bases, how to build and test agents, and how to develop Fusion Agentic Applications using pro-code tools.

What are the four agent tool types tested on 1Z0-1155-2?

Document Tool (search and retrieve from unstructured documents), Business Object Tool (read from and write to Oracle Fusion business objects), Deep Link Tool (navigate users to Fusion UI pages), and External REST Tool (call external APIs outside Oracle Fusion).

What is a Fusion Agentic Application?

A Fusion Agentic Application is a new class of enterprise application introduced with Oracle Fusion Applications Release 26A. It consists of multiple specialized AI agents that reason, coordinate, and execute work autonomously through Oracle Fusion business objects, workflows, tools, and approvals – operating against specific business objectives rather than just responding to individual queries.

Does 1Z0-1155-2 require knowledge of pro-code development?

Yes. Domain 6 of the exam covers advanced agent development using VS Code, CLI tools, the Oracle AI Agent SDK, Git-based lifecycle management, and CI/CD workflows. Candidates without development experience should focus on ensuring Domain 6 content is covered in their preparation.

Related Certifications Worth Exploring

Oracle 1Z0-1155-2 certified developers building a complete Oracle AI credential portfolio will find our Oracle Fusion AI Agent Studio Foundations Associate (1Z0-1145-1) exam questions page covers the Associate-level credential that provides the conceptual foundation the Developer Professional builds upon – valuable for candidates who want to validate foundational knowledge before attempting the Developer Professional exam. For those expanding their Oracle Cloud technical expertise alongside AI Agent Studio development, our Oracle Cloud Infrastructure 2025 Application Integration Professional (1Z0-1042-25) exam questions page covers the integration platform credential relevant to developers connecting AI agents to external systems through Oracle Integration Cloud.