AIP-C01 Online Practice Questions and Verified Answers

Question 1 Single choice

A company has a customer service application that uses Amazon Bedrock to generate personalized responses to customer inquiries. The company needs to establish a quality assurance process to evaluate prompt effectiveness and model configurations across updates. The process must automatically compare outputs from multiple prompt templates, detect response quality issues, provide quantitative metrics, and allow human reviewers to give feedback on responses. The process must prevent configurations that do not meet a predefined quality threshold from being deployed.

Which solution will meet these requirements?

A
Create an AWS Lambda function that sends sample customer inquiries to multiple Amazon Bedrock model configurations and stores responses in Amazon S3. Use Amazon QuickSight to visualize response patterns. Manually review outputs daily. Use AWS CodePipeline to deploy configurations that meet the quality threshold.
B
Use Amazon Bedrock evaluation jobs to compare model outputs by using custom prompt datasets.
Configure AWS CodePipeline to run the evaluation jobs when prompt templates change. Configure CodePipeline to deploy only configurations that exceed the predefined quality threshold.
C
Set up Amazon CloudWatch alarms to monitor response latency and error rates from Amazon Bedrock.
Use Amazon EventBridge rules to notify teams when thresholds are exceeded. Configure a manual approval workflow in AWS Systems Manager.
D
Use AWS Lambda functions to create an automated testing framework that samples production traffic and routes duplicate requests to the updated model version. Use Amazon Comprehend sentiment analysis to compare results. Block deployment if sentiment scores decrease.

Show answer and explanation

Correct answer: B

Explanation

Bedrock evaluation jobs run the competing prompt templates and model configurations against a consistent custom prompt dataset, producing comparable quality measurements and outputs for review. Triggering those jobs from CodePipeline whenever templates change makes evaluation repeatable across updates. The pipeline gate then prevents deployment unless the measured result exceeds the predefined quality threshold.

Question 2 Single choice

A medical company is building a generative AI (GenAI) application that uses Retrieval Augmented Generation (RAG) to provide evidence-based medical information. The application uses Amazon OpenSearch Service to retrieve vector embeddings. Users report that searches frequently miss results that contain exact medical terms and acronyms and return too many semantically similar but irrelevant documents. The company needs to improve retrieval quality and maintain low end-user latency, even as the document collection grows to millions of documents.

Which solution will meet these requirements with the LEAST operational overhead?

A
Configure hybrid search by combining vector similarity with keyword matching to improve semantic understanding and exact term and acronym matching.
B
Increase the dimensions of the vector embeddings from 384 to 1536. Use a post-processing AWS
Lambda function to filter out irrelevant results after retrieval.
C
Replace OpenSearch Service with Amazon Kendra. Use query expansion to handle medical acronyms
and terminology variants during pre-processing.
D
Implement a two-stage retrieval architecture in which initial vector search results are re-ranked by an ML model hosted on Amazon SageMaker.

Show answer and explanation

Correct answer: A

Explanation

The current vector search captures conceptual similarity but can miss exact medical terms and acronyms. Hybrid search combines that vector similarity with keyword matching, so exact terminology contributes directly to retrieval while semantic matches remain available. OpenSearch performs both retrieval signals without adding a separately hosted reranking model, preserving low latency and limiting operational overhead as the corpus grows.

Question 3 Single choice

Inference costs are rising, and requests vary widely in complexity while answer quality must remain stable. What should the team do?

A
Route every request to the lowest-priced model and use longer prompts to compensate for any quality difference.
B
Measure token use, trim context, tune retrieval, route by complexity, and recheck quality.
C
Truncate every input to the same token count because shorter context cannot affect answers to simple requests.
D
Reduce only maximum output length because input context and repeated retrieval do not materially affect inference cost.

Show answer and explanation

Correct answer: B

Explanation

Cost optimization must identify where tokens are consumed before changing behavior. Measuring usage exposes excess input, output, or retrieval context; trimming context and tuning retrieval reduce unnecessary material. Routing simple and complex requests to suitable models avoids paying the same rate for every case. Rechecking quality verifies that these savings preserve the required answer performance.

Question 4 Single choice

Retrieved documents can contain instructions that attempt to make a support agent disclose data or call a privileged tool. Which control set best reduces this risk?

A
Strengthen the system prompt and allow every tool call that follows the system prompt's response format.
B
Filter direct user input, then trust retrieved documents because they came from an internal vector store.
C
Use private network connectivity for retrieval and let the model determine whether tool arguments are authorized.
D
Delimit untrusted content, limit tools and data, and authorize consequential actions outside the model.

Show answer and explanation

Correct answer: D

Explanation

Retrieved documents remain untrusted even when stored internally because their text can contain instructions aimed at the agent. Delimiting that content separates evidence from privileged directions, while limiting available tools and accessible data reduces what a malicious passage can influence. Consequential actions must pass authorization outside the model, preventing generated tool requests from becoming permission decisions.

Question 5 Single choice

A company runs a Retrieval Augmented Generation (RAG) application that uses Amazon Bedrock Knowledge Bases to perform regulatory compliance queries. The application uses the RetrieveAndGenerateStream API. The application retrieves relevant documents from a knowledge base that contains more than 50,000 regulatory documents, legal precedents, and policy updates. The RAG application is producing suboptimal responses because the initial retrieval often returns semantically similar but contextually irrelevant documents. The poor responses are causing model hallucinations and incorrect regulatory guidance. The company needs to improve the performance of the RAG application so
it returns more relevant documents.

Which solution will meet this requirement with the LEAST operational overhead?

A
Deploy an Amazon SageMaker endpoint to run a fine-tuned ranking model. Use an Amazon API Gateway REST API to route requests. Configure the application to make requests through the REST API to rerank the results.
B
Use Amazon Comprehend to classify documents and apply relevance scores. Integrate the RAG application's reranking process with Amazon Textract to run document analysis. Use Amazon Neptune to perform graph-based relevance calculations.
C
Implement a retrieval pipeline that uses the Amazon Bedrock Knowledge Bases Retrieve API to perform initial document retrieval. Call the Amazon Bedrock Rerank API to rerank the results. Invoke the InvokeModelWithResponseStream operation to generate responses.
D
Use the latest Amazon reranker model through the reranking configuration within Amazon Bedrock Knowledge Bases. Use the model to improve document relevance scoring and to reorder results based on contextual assessments.

Show answer and explanation

Correct answer: D

Explanation

The initial vector retrieval already finds semantically similar documents, but their contextual relevance is weak. Configuring an Amazon reranker within Amazon Bedrock Knowledge Bases adds a second relevance assessment that rescoring and reorders those candidates according to the query context. Because reranking is enabled in the existing knowledge base configuration, it improves the documents supplied for generation without a custom retrieval pipeline.

Question 6 Single choice

A financial services company is building a customer support application that retrieves relevant financial regulation documents from a database based on semantic similarity to user queries. The application must integrate with Amazon Bedrock to generate responses. The application must search documents in English, Spanish, and Portuguese. The application must filter documents by metadata such as publication date, regulatory agency, and document type.
The database stores approximately 10 million document embeddings. To minimize operational overhead, the company wants a solution that minimizes management and maintenance effort while providing low-latency responses for real-time customer interactions.

Which solution will meet these requirements?

A
Use Amazon OpenSearch Serverless to provide vector search capabilities and metadata filtering.
Integrate with Amazon Bedrock Knowledge Bases to enable Retrieval Augmented Generation (RAG) using an Anthropic Claude foundation model.
B
Deploy an Amazon Aurora PostgreSQL database with the pgvector extension. Store embeddings and metadata in tables. Use SQL queries for similarity search and send results to Amazon Bedrock for response generation.
C
Use Amazon S3 Vectors to configure a vector index and non-filterable metadata fields. Integrate S3
Vectors with Amazon Bedrock for RAG.
D
Set up an Amazon Neptune Analytics database with a vector index. Use graph-based retrieval and
Amazon Bedrock for response generation.

Show answer and explanation

Correct answer: A

Explanation

Amazon OpenSearch Serverless supplies managed vector search for the large embedding collection and supports metadata filters for publication date, agency, and document type. Amazon Bedrock Knowledge Bases can use that retrieval layer for RAG across the multilingual documents. The serverless design reduces infrastructure management while supporting low-latency customer interactions.

Question 7 Single choice

A hotel company wants to enhance a legacy Java-based property management system (PMS) by adding AI capabilities. The company wants to use Amazon Bedrock Knowledge Bases to provide staff with room availability information and hotel-specific details. The solution must maintain separate access controls for each hotel that the company manages. The solution must provide room availability information in near real time and must maintain consistent performance during peak usage periods.

Which solution will meet these requirements?

A
Deploy a single Amazon Bedrock knowledge base that contains combined data for all hotels. Configure
AWS Lambda functions to synchronize data from each hotel's PMS database through direct API connections. Implement AWS CloudTrail logging with hotel-specific filters to audit access logs for each
hotel's data.
B
Create an Amazon EventBridge rule for each hotel that is invoked by changes to the PMS database.
Configure the rule to send updates to a centralized Amazon Bedrock knowledge base in a management AWS account. Configure resource-based policies to enforce hotel-specific access controls.
C
Implement one Amazon Bedrock knowledge base for each hotel in a multi-account structure. Use direct data ingestion to provide near real-time room availability information. Schedule regular synchronization for less critical information.
D
Build a centralized Amazon Bedrock Agents solution that uses multiple knowledge bases. Implement
AWS IAM Identity Center with hotel-specific permission sets to control staff access.

Show answer and explanation

Correct answer: C

Explanation

A separate Amazon Bedrock knowledge base for each hotel creates a clear data and access boundary, and the multi-account structure reinforces that isolation. Direct data ingestion carries room availability changes into the relevant knowledge base for near real-time queries. Regular synchronization is reserved for less critical hotel details, avoiding unnecessary real-time processing while keeping each hotel's corpus current.

Question 8 Single choice

A healthcare application uses Amazon Bedrock to summarize clinical notes. The team must reduce unsupported claims in generated summaries and block unsafe output with minimal custom code. The team also needs an audit trail of model requests and responses.

Which solution should the team implement?

A
Use AWS CloudTrail only to record InvokeModel API calls, and rely on manual review to detect unsupported claims.
B
Create a custom classifier in AWS Lambda that checks each generated sentence after the response is returned to the user.
C
Configure Amazon Bedrock Guardrails with contextual grounding and content filters, and enable model invocation logging to an Amazon S3 destination.
D
Move the workload to Amazon SageMaker AI training jobs so that the model learns to avoid hallucinations from future examples.

Show answer and explanation

Correct answer: C

Explanation

Contextual grounding checks compare generated claims with the supplied clinical context, helping reduce unsupported statements. Bedrock Guardrails content filters block output that violates the safety policy with little custom code. Model invocation logging records requests and responses in the configured S3 destination, providing the audit trail separately from the controls that evaluate generation quality.

Question 9 Single choice

A financial services company is developing a real-time generative AI (GenAI) assistant to support human call center agents. The GenAI assistant must transcribe live customer speech, analyze context, and provide incremental suggestions to call center agents while a customer is still speaking. To preserve responsiveness, the GenAI assistant must maintain end-to-end latency under 1 second from speech to initial response display. The architecture must use only managed AWS services and must support bidirectional streaming to ensure that call center agents receive updates in real time.

Which solution will meet these requirements?

A
Use Amazon Transcribe streaming to transcribe calls. Pass the text to Amazon Comprehend for sentiment analysis. Feed the results to Anthropic Claude on Amazon Bedrock by using the InvokeModel API. Store results in Amazon DynamoDB. Use a WebSocket API to display the results.
B
Use Amazon Transcribe streaming with partial results enabled to deliver fragments of transcribed text before customers finish speaking. Forward text fragments to Amazon Bedrock by using the InvokeModelWithResponseStream API. Stream responses to call center agents through an Amazon
API Gateway WebSocket API.
C
Use Amazon Transcribe batch processing to convert calls to text. Pass complete transcripts to
Anthropic Claude on Amazon Bedrock by using the ConverseStream API. Return responses through an Amazon Lex chatbot interface.
D
Use the Amazon Transcribe streaming API with an AWS Lambda function to transcribe each audio segment. Call the Amazon Titan Embeddings model on Amazon Bedrock by using the InvokeModel API. Publish results to Amazon SNS.

Show answer and explanation

Correct answer: B

Explanation

Amazon Transcribe streaming with partial results emits text fragments before the customer finishes speaking, shortening the speech-to-model delay. InvokeModelWithResponseStream returns model output incrementally instead of waiting for a complete response. An API Gateway WebSocket API carries those updates to the call center agent, creating a managed streaming path in both directions for real-time suggestions.

Question 10 Multiple choice

A platform team wants to improve governance for prompts that are shared by several production GenAI applications.

Which two actions should the team take? (Choose two.)

A
Manage reusable prompt templates with Amazon Bedrock Prompt Management and require reviewed versions for production use.
B
Allow each application to modify prompt text directly in production without recording the change.
C
Store the only copy of each prompt in a chat transcript so developers can copy it when needed.
D
Use AWS CloudTrail and operational logs to track prompt asset access, deployment activity, and
runtime behavior.
E
Remove prompt tests because prompt changes do not affect application correctness.

Show answer and explanation

Correct answers: A, D

Explanation

Bedrock Prompt Management supplies a shared home for reusable templates and reviewed production versions, preventing applications from drifting through unrecorded edits. CloudTrail tracks access and deployment activity associated with those prompt assets, while operational logs capture runtime use and behavior. Together, version governance and traceability support both controlled releases and later investigation.