Amazon AWS Certified Generative AI Developer - Professional AIP-C01 Practice Test 28

Question 270

A document-processing state machine needs human escalation. Which integration fix is appropriate for this failure?

Mark the document validated before a required human reviewer examines the exceptions.
Delete the parsed source/decision evidence after sending a human notification.
Let the model impersonate an approver by filling the manager username in output JSON.
Model explicit exception state and preserve parsed evidence for review.

Explanation

Correct Answer

Model a real human-review exception state, retaining parsed evidence and workflow identity for authenticated escalation and resumption.

Incorrect Answers

A pending exception is not an approval.

The reviewer needs the evidence to adjudicate the case.

A generated name does not supply independent human authorization.

Question 271

An integration API's model parameter changes break consumers. What must the application pipeline enforce before promotion?

Replace response field names in the production gateway without versioning.
Tell old clients to parse arbitrary free-text model replies instead of their expected response fields.
Version gateway adapter and test old/new client contracts.
Ship unverified vendor-specific model parameters directly from external callers into Bedrock.

Explanation

Correct Answer

Use a tested versioned adapter that preserves old client contracts while explicitly mapping changed provider parameters to a new schema.

Incorrect Answers

Existing consumers may fail deserialization.

Unstructured output breaks the machine-readable contract.

Provider fields require backend validation and compatibility checks.

Question 272

A workflow needs multiple FM calls with branches. Which development or release-control change meets this requirement?

Define prompt flow or explicit state-machine stages with typed outputs.
Concatenate all stage prompts into one untyped message and assume intermediate validation still runs.
Treat a flow's first successful model call as proof later branches and writes completed.
Retry the entire flow with new transaction IDs after a downstream timeout.

Explanation

Correct Answer

Use supported Bedrock Flows or explicit Step Functions-style stages with typed outputs, conditions and error paths across separate model calls.

Incorrect Answers

A single model response does not reproduce deterministic branch/error handling.

Multi-stage workflows require per-stage execution receipts.

This can duplicate committed side effects.

Question 273

An accessible web interface shows long AI answer but no source link. Which integration fix is appropriate for this failure?

Display a plain URL-looking string without a working citation target or link semantics.
Include verifiable citations and keyboard/screen-reader navigable controls.
Use only a tiny color-coded citation icon with no accessible label or keyboard focus.
Link directly to private S3 objects for everyone without caller authorization.

Explanation

Correct Answer

Expose verifiable source links through authorized accessible UI controls, including keyboard focus and screen-reader descriptive text.

Incorrect Answers

Visual resemblance is not a usable, traceable source link.

Assistive users cannot navigate or identify the supporting source.

A citation must respect access rights as well as navigability.

Question 274

An Amplify frontend needs responsive AI state management. What must the application pipeline enforce before promotion?

Distinguish loading/partial/error/verified states and test mobile/assistive access.
Show a permanent success badge as soon as the request starts.
Hide error text and expose failures only through changing spinner colors.
Store entire customer prompts in client logs by default to debug responsive behavior.

Explanation

Correct Answer

Model explicit loading, partial, failed, canceled and verified states, then test responsive screen-reader and keyboard interactions.

Incorrect Answers

A pending request has not produced a verified answer.

Color alone is inaccessible and does not explain recovery.

UI debugging must still protect sensitive content.

Question 275

A CRM assistant must write a ticket only after validation. Which development or release-control change meets this requirement?

Invoke backend transaction with user permissions and return real ticket ID.
Generate a plausible ticket number before the CRM returns its create response.
Let the model bypass backend validation because its requested JSON matches the tool schema.
Treat a queued ticket write as confirmed without checking persisted status.

Explanation

Correct Answer

Validate the authenticated user's ticket permissions and input values at backend, commit the ticket, then return the actual ticket ID.

Incorrect Answers

A model-produced ID is not a persisted CRM record.

Schema shape does not establish record-level authorization.

Accepted asynchronous processing is not a completed ticket transaction.

Question 276

A billing app wants AI summary after a refund. Which integration fix is appropriate for this failure?

Have the model compose a 'refund completed' message from the user's intent before the backend runs.
Trigger summary from confirmed refund event rather than predicted action.
Trigger the summary from an initial refund request event and label it successful before settlement.
Use an unrelated cached payment record to infer the refund amount and status.

Explanation

Correct Answer

Emit the AI summary only after the refund platform has committed and published a confirmed refund event, preserving transaction references.

Incorrect Answers

Intent can differ from the actual financial outcome.

A request is not the refund's confirmed committed event.

Cached inference cannot replace the actual refund receipt.

Question 277

A developer uses Amazon Q to generate IAM policy code. What must the application pipeline enforce before promotion?

Review generated code for least privilege before merging.
Deploy Amazon Q-generated IAM JSON automatically because it was produced by an AWS tool.
Accept wildcard resources on all actions to avoid permission denials during testing.
Review only JSON syntax while ignoring privilege escalation implications.

Explanation

Correct Answer

AI-generated IAM policies are draft code; review effect/actions/resources/conditions and run least-privilege validation before merge.

Incorrect Answers

Code generation does not constitute security review.

Excessive policy scope violates least privilege and enlarges blast radius.

A syntactically valid IAM policy can grant dangerous unintended rights.

Question 278

A new AI-generated integration passes normal tool calls but its error paths have not been tested. Select TWO negative test categories: rejected/unauthorized input and uncertain results after a downstream timeout.

Test invalid arguments and unauthorized caller behavior.
Check only that a well-formed example returns HTTP 200.
Assume authentication failure cannot occur because the UI hides the tool button.
Test downstream timeout and uncertain operation states.
Treat timed-out tool calls as succeeded because the model's text looks normal.

Explanation

Correct Answer

Test malformed and unauthorized requests to confirm contract/auth enforcement.

Test backend timeouts and unknown-commit states to confirm truthful retry and error handling.

Incorrect Answers

Happy paths do not cover input, identity or dependency failures.

Direct API calls can bypass UI affordances.

A timeout leaves backend effect uncertain until verified.

Question 279

A developer wants automatic code refactoring in a protected repo. Which integration fix is appropriate for this failure?

Allow the agent to merge changes to the protected branch immediately after generation.
Run only a linter and skip behavioral tests for generated refactors.
Attach privileged tokens to the agent prompt to let it bypass repository checks.
Require code review, test suite and secret scan before merging suggestions.

Explanation

Correct Answer

Automated refactoring suggestions require protected-branch review, tests and secret/policy scans before merge, even if they compile.

Incorrect Answers

Unreviewed code can introduce regressions or leak secrets.

Formatting validity does not establish functional equivalence.

Embedding secrets and bypassing controls violates the protected merge policy.

Question 280

A multi-agent application uses Strands to invoke tools on behalf of individual employees. Select THREE mandatory elements: typed tools, caller-specific executor authorization and a verifiable result of each side effect.

Register a broad untyped wildcard tool that executes arbitrary commands from model text.
Define typed Strands tool contracts.
Invoke every action with a global administrator credential regardless of caller.
Record only the final natural-language answer and omit backend tool execution results.
Enforce actor-specific executor permissions.
Retain authoritative tool execution receipts.

Explanation

Correct Answer

Typed signatures constrain Strands tool inputs at the interface boundary.

Executor must independently authorize the authenticated actor for each requested capability.

Store tool request/correlation and authoritative result receipts, not just model narration.

Incorrect Answers

Free-form tools cannot reliably enforce schema and operation allowlists.

Shared admin execution loses user-specific permissions.

Without receipts, the workflow cannot prove what actions actually ran.

Pass Your Amazon AWS Certified Generative AI Developer - Professional AIP-C01 Exam Easy!

Question 270

Explanation

Correct Answer

Incorrect Answers

Question 271

Explanation

Correct Answer

Incorrect Answers

Question 272

Explanation

Correct Answer

Incorrect Answers

Question 273

Explanation

Correct Answer

Incorrect Answers

Question 274

Explanation

Correct Answer

Incorrect Answers

Question 275

Explanation

Correct Answer

Incorrect Answers

Question 276

Explanation

Correct Answer

Incorrect Answers

Question 277

Explanation

Correct Answer

Incorrect Answers

Question 278

Explanation

Correct Answer

Incorrect Answers

Question 279

Explanation

Correct Answer

Incorrect Answers

Question 280

Explanation

Correct Answer

Incorrect Answers

AWS Certified Generative AI Developer - Professional AIP-C01 Practice Tests