top of page

AIP-C01 practice exam and study guide

AWS Certified Generative AI Developer, Professional

Timed practice exams for AIP-C01 on the real exam's 180 minute limit. Each practice exam serves the same number of questions as the real exam. Every question is original and every option is explained. Your score is broken down by the domains AWS publishes, so you see what to study next. The question bank holds hundreds of practice questions.

Exam facts

  • Questions: 75 [1]
  • Passing score: 750 on a scaled score of 100 to 1,000 [1]
  • Time limit: 180 minutes [1]
  • Exam languages (real exam): English, Japanese, Korean, Simplified Chinese [1]
  • Delivery: Pearson VUE testing center or online proctored exam [1]

Tokens: 1 for $4.99, 5 for $14.99, 10 for $24.99 or 25 for $49.99. One token runs a full timed AIP-C01 practice exam, and every shorter session on that exam is then free. The Unlimited pass is $149.99 for a year of every practice session on every exam.

Get tokens

Exam domains and weights

AWS publishes these domains and objectives for AIP-C01, quoted here as AWS words them. [2]

Domain 1: Foundation Model Integration, Data Management, and Compliance, 31% of the exam

  • "Task 1.1: Analyze requirements and design GenAI solutions."
  • "Task 1.2: Select and configure FMs."
  • "Task 1.3: Implement data validation and processing pipelines for FM consumption."
  • "Task 1.4: Design and implement vector store solutions."
  • "Task 1.5: Design retrieval mechanisms for FM augmentation."
  • "Task 1.6: Implement prompt engineering strategies and governance for FM interactions."

Domain 2: Implementation and Integration, 26% of the exam

  • "Task 2.1: Implement agentic AI solutions and tool integrations."
  • "Task 2.2: Implement model deployment strategies."
  • "Task 2.3: Design and implement enterprise integration architectures."
  • "Task 2.4: Implement FM API integrations."
  • "Task 2.5: Implement application integration patterns and development tools."

Domain 3: AI Safety, Security, and Governance, 20% of the exam

  • "Task 3.1: Implement input and output safety controls."
  • "Task 3.2: Implement data security and privacy controls."
  • "Task 3.3: Implement AI governance and compliance mechanisms."
  • "Task 3.4: Implement responsible AI principles."

Domain 4: Operational Efficiency and Optimization for GenAI Applications, 12% of the exam

  • "Task 4.1: Implement cost optimization and resource efficiency strategies."
  • "Task 4.2: Optimize application performance."
  • "Task 4.3: Implement monitoring systems for GenAI applications."

Domain 5: Testing, Validation, and Troubleshooting, 11% of the exam

  • "Task 5.1: Implement evaluation systems for GenAI."
  • "Task 5.2: Troubleshoot GenAI applications."

Sample questions

Three questions from the AIP-C01 practice bank, with the answer and why each option is right or wrong.

Question 1

A company's platform team maintains a custom lens for generative AI workloads in the AWS Well-Architected Tool. Forty workloads already have the published lens applied. The team has fixed spelling errors in three questions and replaced two broken resource links. The team wants every workload to pick up the corrections without asking any workload owner to act. Which solution will meet these requirements?

  • A. Publish the updated lens as a minor version, so that the change is applied to every workload that already uses the lens.
  • B. Publish the updated lens as a major version, so that the change is applied to every workload that already uses the lens.
  • C. Publish the updated lens as a major version, and reuse the earlier version name so that workloads treat it as the same version.
  • D. Upload the corrected file as a new custom lens with a new name, and apply the new lens to each of the 40 workloads.

Answer: A. Pick the version type by what changed. Small text and link fixes go out as a minor version and reach workloads on their own. Changes in meaning go out as a major version, and each owner upgrades when ready. [3]

  • A. Right. Minor versions are meant for small changes such as text edits and link updates, and they are applied to workloads automatically without any notification.
  • B. Wrong. Major version changes are not applied automatically. Each workload that uses the lens gets a notification and its owner decides whether to upgrade.
  • C. Wrong. The version name must be a unique identifier for the version change, so an earlier name can't be reused, and a major version still waits for owners.
  • D. Wrong. A lens with a new name is a separate lens. Someone would have to apply it to all 40 workloads and answer its questions, the work the team wants to avoid.

Question 2

A company's IT support agent on Amazon Bedrock AgentCore keeps raw conversation events in AgentCore Memory. The team wants the agent to learn from past troubleshooting sessions: which tools it chose, where it hit errors and which approach finally worked. It also wants insights drawn across many such sessions, without building its own summarization pipeline. Which solution will meet these requirements?

  • A. Raise the retention of the raw events, and have the agent replay every past session into the prompt of each new session.
  • B. Store each session's final answer in an Amazon DynamoDB table, and have the agent look up the latest answer for each user.
  • C. Write a nightly AWS Lambda function that summarizes each session and stores the summaries as records in a new memory.
  • D. Add an episodic memory strategy, so that AgentCore records completed episodes and generates reflections across episodes.

Answer: D. Episodic memory turns interaction histories into compact episode records and reflections. Use it when an agent should learn from how past tasks went. [4]

  • A. Wrong. Replaying raw events puts every past session into the prompt and gives no structured view of what worked across sessions.
  • B. Wrong. A final answer for each user loses the sequence of tool choices and errors, and it draws no insights across sessions.
  • C. Wrong. A custom summarization pipeline is the work the team wants to avoid, and the episodic strategy already extracts and consolidates episodes.
  • D. Right. The episodic strategy detects episode completion, consolidates extractions into episode records with intent, assessment and justification, and generates reflections across episodes.

Question 3

A company's research agent calls a web search tool through an Amazon Bedrock AgentCore Gateway in us-east-1 with Policy in AgentCore attached. Some search results contain hateful text, and the trust team wants those tool responses withheld from the agent while the tool keeps running. A developer wrote a forbid policy whose guardrail condition reads the tool response, and creating the policy fails. Which change will meet these requirements?

  • A. Change the effect to permit and keep the guardrail condition on the tool response, so that the policy grants only clean responses.
  • B. Keep the forbid effect, and add a temporal condition so that the guardrail reads the response after the tool call completes.
  • C. Change the effect to suppressOutput and keep the guardrail condition on the tool response at context.output.
  • D. Move the guardrail check into the agent's system prompt, so that the model discards hateful search results itself.

Answer: C. Authorization effects decide whether a call may run. suppressOutput screens what a call returns, and its condition holds guardrail checks only. [5]

  • A. Wrong. permit is an authorization effect like forbid, and an authorization effect that references context.output is rejected too.
  • B. Wrong. A forbid policy that references context.output is rejected. Outputs are screened with the suppressOutput effect.
  • C. Right. suppressOutput evaluates what an authorized action returns and withholds the response when the guardrail is violated. The tool still runs, and the effect reads only context.output paths.
  • D. Wrong. Instructions in a prompt aren't enforced. Policy in AgentCore evaluates traffic at the gateway, outside the agent's code.

Practice for AIP-C01 on ExamSens.ai

ExamSens.ai is not authorized, endorsed or sponsored by Amazon Web Services.

References

[1] https://aws.amazon.com/certification/certified-generative-ai-developer-professional/

[2] https://docs.aws.amazon.com/aws-certification/latest/ai-professional-01/ai-professional-01-domain1.html

[3] https://docs.aws.amazon.com/wellarchitected/latest/userguide/lenses-publish-update.html

[4] https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/episodic-memory-strategy.html

[5] https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/policy-guardrails-in-policies.html

bottom of page