AIF-C01 Practice Question 77: Deterministic Multistep Workflow
AWS Certified AI Practitioner (AIF-C01) · Test 8 · question 7 of 10
A team needs a repeatable pipeline: take a support ticket, classify it with a prompt, look up the matching article in a knowledge base, then call an AWS Lambda function to raise a work order. They want the steps and the data passed between them declared explicitly rather than left to a model's judgement, and they want to snapshot the pipeline and promote it from test to production. Which Amazon Bedrock capability should they use?
Show answer and explanation
Answer:
Amazon Bedrock Flows lets you *build workflows by linking prompts, foundational models, and other AWS services to create end-to-end solutions*. You *integrate with Amazon Bedrock offerings such as FMs, knowledge bases, and other AWS services such as AWS Lambda by transferring data between them*, which covers all three steps. Because the flow is a set of nodes and explicit connections, the ordering is declared rather than inferred. For promotion, AWS provides *immutable workflows to move from testing to production*: publishing a version preserves the flow definition as a snapshot, and an alias points at the version your application invokes (option B).
- A is wrong because prompt caching reduces the cost of reprocessing a repeated prompt prefix. It is a cost optimisation applied to individual calls and declares no steps, no data passing and no promotion path. - C is wrong because a single prompt hands the ordering, the decision to consult the knowledge base and the shape of each intermediate result back to the model's judgement, the exact opposite of what the team asked for. A prompt also cannot invoke a Lambda function by itself. - D is wrong because intelligent prompt routing chooses *which model* answers one request, to balance quality against cost. It orchestrates nothing and cannot chain a classification, a retrieval and an action together.
Reference: Build an end-to-end generative AI workflow with Amazon Bedrock Flows