A bank is building a Bedrock-powered chatbot that will answer account-opening questions using public bank documents. The team will use prompt engineering to shape model responses. Which prompt engineering technique fits this scenario?

Directional stimulus prompting is the most suitable technique because it involves providing explicit instructions or constraints within the prompt to guide the model's response toward specific information or a desired format. In this scenario, the bank wants the chatbot to answer questions based only on public bank documents, which requires directing the model to use that specific knowledge base and avoid hallucinating or using general knowledge. Complexity-based prompting is not a standard or widely recognized prompt engineering technique. Zero-shot prompting involves asking the model to perform a task without any examples, which might lead to less precise answers if not combined with strong directional stimuli. Few-shot prompting provides a few examples of desired input/output pairs, which could be useful but is less direct than explicitly instructing the model to use specific documents.